http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479
MATLAB Central Newsreader  MATLAB Central Spring Contest
Feed for thread: MATLAB Central Spring Contest
enus
©19942014 by MathWorks, Inc.
webmaster@mathworks.com
MATLAB Central Newsreader
http://blogs.law.harvard.edu/tech/rss
60
MathWorks
http://www.mathworks.com/images/membrane_icon.gif

Tue, 29 Apr 2008 21:05:04 +0000
MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429465
Helen Chen
Just a reminder to let everyone know that the Spring Contest<br>
launches tomorrow, Wednesday May 30th at high noon. Be<br>
there or be square!<br>
<br>
See you then!<br>
Helen

Wed, 30 Apr 2008 00:34:22 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429485
D. Ismay
Helen Chen wrote on 29Apr08 14:05 :<br>
> Just a reminder to let everyone know that the Spring Contest<br>
> launches tomorrow, Wednesday May 30th at high noon.<br>
<br>
perhaps it does, on your planet. but on /this/ one, May 30th doesn't <br>
arrive for another 30 days.

Wed, 30 Apr 2008 16:23:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429622
Helen Chen
"D. Ismay" <!@a.com> wrote in message <br>
<cuSdnYaj8fQSI4rVnZ2dnUVZ_qqgnZ2d@earthlink.com>...<br>
> Helen Chen wrote on 29Apr08 14:05 :<br>
> > Just a reminder to let everyone know that the Spring <br>
Contest<br>
> > launches tomorrow, Wednesday May 30th at high noon.<br>
> <br>
> perhaps it does, on your planet. but on /this/ one, <br>
May 30th doesn't <br>
> arrive for another 30 days.<br>
<br>
Sorry about that, yes, I got over excited looking at the <br>
calendar!<br>
<br>
But still, the time has arrived. The contest begins.. <br>
<br>
The contest home page is at <br>
<a href="http://www.mathworks.com/contest/wiring/home.html">http://www.mathworks.com/contest/wiring/home.html</a> .<br>
This page includes important links including the blog, <br>
rules, and the contest queue. <br>
<br>
If you've participated in the contest before, you can get <br>
started right now by going right to the rules page at <br>
<a href="http://www.mathworks.com/contest/wiring/rules.html">http://www.mathworks.com/contest/wiring/rules.html</a> . <br>
<br>
Good luck to everyone!<br>
<br>
Helen

Wed, 30 Apr 2008 16:32:04 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429625
Alan Chalker
Good luck everyone. Looks to be yet another interesting<br>
contest. I'll try to do a mid contest commented code this<br>
weekend as usual, however I might be a little tied up in<br>
trying to teach MATLAB to the newest contest participant, my<br>
new daughter Katie who was born on Monday;)

Wed, 30 Apr 2008 16:44:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429627
srach
"Alan Chalker" <alancNOSPAM@osc.edu> wrote in message<br>
<fva6u4$r8i$1@fred.mathworks.com>...<br>
> Good luck everyone. Looks to be yet another interesting<br>
> contest. I'll try to do a mid contest commented code this<br>
> weekend as usual, however I might be a little tied up in<br>
> trying to teach MATLAB to the newest contest participant, my<br>
> new daughter Katie who was born on Monday;)<br>
<br>
Congratulations, Alan!<br>
<br>
And good luck to everyone, of course. :)<br>
<br>
srach

Wed, 30 Apr 2008 16:45:05 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429629
Duane Hanselman
"Helen Chen" <helen.chen@mathworks.com> wrote in message<br>
<fv82i0$nql$1@fred.mathworks.com>...<br>
> Just a reminder to let everyone know that the Spring Contest<br>
> launches tomorrow, Wednesday May 30th at high noon. Be<br>
> there or be square!<br>
> <br>
> See you then!<br>
> Helen<br>
<br>
The rules state that this is based on<br>
"the problem of wiring up printed circuit boards"<br>
<br>
How does one "wire up"? Is it possible to "wire down" as<br>
well? Or perhaps "wire in" or "wire out"?<br>
<br>
I'm just being picky. Terms such as "wire up", "connect up",<br>
etc. are not grammatically correct. The word "up" is not<br>
needed. Just drop it:<br>
<br>
"the problem of wiring printed circuit boards"<br>
<br>
This is similar to the phrase "In order to...", just drop<br>
the first two words "To..."<br>
<br>
I've written books with both of these things in them. The<br>
copy editors strike them out immediately. I am now working<br>
on my 11th book, and I finally learned to avoid these before<br>
the copy editor sees them. :)<br>
<br>
p.s., this looks like a great contest. I like the penalties<br>
for poor coding!<br>
<br>
Duane Hanselman

Wed, 30 Apr 2008 16:46:05 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429630
Sergey
"Alan Chalker" <alancNOSPAM@osc.edu> wrote in message <br>
<fva6u4$r8i$1@fred.mathworks.com>...<br>
> Good luck everyone. Looks to be yet another interesting<br>
> contest. I'll try to do a mid contest commented code this<br>
> weekend as usual, however I might be a little tied up in<br>
> trying to teach MATLAB to the newest contest participant, <br>
my<br>
> new daughter Katie who was born on Monday;)<br>
<br>
Congratulations!<br>
<br>
SY

Wed, 30 Apr 2008 17:40:04 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429634
Andy Johnson
Oops, I just posted this comment to the wrong place, so now <br>
I'll post it here (sorry)...<br>
<br>
Shouldn’t the last line of this part of the instructions <br>
say [ 4 5 4 6] ?<br>
<br>
The segments for the connector between the 8 pins could be <br>
written like so.<br>
w = [ 2 3 2 4 ]<br>
[ 2 4 3 4 ]<br>
[ 3 4 3 5 ]<br>
[ 3 5 4 5 ]<br>
[ 4 5 5 6 ]

Wed, 30 Apr 2008 18:16:05 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429644
Lucio AndradeCetto
Andy:<br>
<br>
That is correct, thanks for the clarification.<br>
<br>
Lucio<br>
The MathWorks Constest Team<br>
<br>
"Andy Johnson" <matman@summitsolutions.net> wrote in <br>
message <fvaatj$bdc$1@fred.mathworks.com>...<br>
> Oops, I just posted this comment to the wrong place, so <br>
now <br>
> I'll post it here (sorry)...<br>
> <br>
> Shouldn’t the last line of this part of the instructions <br>
> say [ 4 5 4 6] ?<br>
> <br>
> The segments for the connector between the 8 pins could <br>
be <br>
> written like so.<br>
> w = [ 2 3 2 4 ]<br>
> [ 2 4 3 4 ]<br>
> [ 3 4 3 5 ]<br>
> [ 3 5 4 5 ]<br>
> [ 4 5 5 6 ]<br>
> <br>
>

Wed, 30 Apr 2008 19:18:01 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429658
D. Ismay
Duane Hanselman wrote:<br>
[...]<br>
> The rules state that this is based on<br>
> "the problem of wiring up printed circuit boards"<br>
> <br>
> How does one "wire up"? Is it possible to "wire down" as<br>
> well? Or perhaps "wire in" or "wire out"?<br>
> <br>
> I'm just being picky. Terms such as "wire up", "connect up",<br>
> etc. are not grammatically correct. The word "up" is not<br>
> needed. Just drop it:<br>
> <br>
> "the problem of wiring printed circuit boards"<br>
> <br>
> This is similar to the phrase "In order to...", just drop<br>
> the first two words "To..."<br>
<br>
You think that's bad? News anchor's favorite lines/words:<br>
<br>
1. "...each and every..." (how are they different?)<br>
2. "...first and foremost..." (can you have a 'last and foremost'?)<br>
3. "...ramification..." (but not 'implication'?)<br>
4. Everything has "impact", but nothing has "effect".<br>
5. "...incredible!" (literal meaning, "not credible")<br>
6. "...massive..." (while describing objects that <br>
do not posses mass)<br>
7. "There's..." (contraction of "there is", for <br>
plural case)<br>
8. 'behaviors', 'moneys', 'winds', 'rains', 'foods', 'fruits', <br>
'medicines', when all<br>
of these are group nouns and don't change spelling for plural case.<br>
<br>
And it gets better  those people are in a profession that, you would<br>
expect, demands aboveaverage understanding and proper use of<br>
English and commonlyknown rules of grammar.<br>
<br>
Go figure.<br>
<br>
D. Ismay

Wed, 30 Apr 2008 20:03:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429668
Hope
One question:<br>
Bridges are expensive: they cost 25 points each. In this<br>
diagram we've saved 22 points by connecting the two 11 pins,<br>
but we had to buy three connectors and a bridge, for a total<br>
cost of 28. The net score for this move is 17 ...<br>
<br>
Should it be 25+3(11*2) = 6 ?<br>
<br>
Using bridge will also lose credit?<br>
<br>
Thanks

Wed, 30 Apr 2008 20:14:09 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429670
roberson@ibd.nrccnrc.gc.ca (Walter Roberson)
In article <a8SdnUuVdO93WIXVnZ2dnUVZ_hvnZ2d@earthlink.com>,<br>
D. Ismay <noSpam@Woohoo.Woohoohoo.Woohoo.Woohoohoo.HOOHOO.hoohoo.Woohoo.Woohoohoo> wrote:<br>
<br>
>8. 'behaviors', 'moneys', 'winds', 'rains', 'foods', 'fruits', <br>
>'medicines', when all<br>
>of these are group nouns and don't change spelling for plural case.<br>
<br>
You are wrong about each of the words you list. <br>
<br>
<br>
<a href="http://www.oed.com">http://www.oed.com</a><br>
<br>
Note: in the below quotations, some nonASCII letters disappeared,<br>
especially eth and thorn. Sorry.<br>
<br>
<br>
behaviour, n<br>
<br>
1b. Also in pl.<br>
<br>
1538 BALE Comedy in Harl. Misc. (Malh.) I. 211 Your fastynges,<br>
longe prayers, with other holy behauers. 1601 SHAKES. Jul. C. I.<br>
ii. 42 Which giue some soyle (perhaps) to my Behauiours. 1678<br>
CUDWORTH Intell. Syst. I. iv. S19. 366 To observe the actions,<br>
manners and Behaviours of men. a1763 `GEO. PSALMANAZAR' Mem. (1764)<br>
186, I could see..thro' all his artifices and different behaviours.<br>
1959 Camb. Rev. 7 Mar. 405/1 We must surely accept that the pattern<br>
of associated behaviours first noticed by Weber was one of the most<br>
brilliantly successful suggestions in the whole history of<br>
intellectual endeavour.<br>
<br>
money, n<br>
<br>
2e. Chiefly Horse Racing and Gambling (orig. U.S.). With<br>
preceding ordinal number: the prize or prize money associated<br>
with finishing in the placing denoted by the number in a<br>
competitive event; this placing itself. [...] 1894 Vermont Agric.<br>
Rep. 14 96 He trotted in seventeen races..; won nine first<br>
moneys.<br>
<br>
3. In pl. (now chiefly in legal and quasilegal parlance). Sums<br>
or quantities of money.<br>
<br>
c1384 Bible (Wycliffite, E.V.) (Douce 369(2)) 2 Macc. iii. 6<br>
Tolde to hym the tresorie in Jerusalem for to be ful with moneys<br>
[L. pecuniis] vnnoumbreable. 1593 Acct. Bk. W. Morton f. 69,<br>
Reseuet..iii scor ane stane of tolone ane pound les, of moneyes<br>
is sum Ic honderis xxxiiii lib. 4s. 1600 SHAKESPEARE Merchant of<br>
Venice I. iii. 115 You come to me, and you say, Shylocke, we<br>
would haue moneyes. 1625 BACON Ess. (new ed.) 246 No Man will<br>
Lend his Moneyes farre off, nor put them into Vnknown Hands. 1632<br>
W. LITHGOW Totall Disc. Trav. IV. 140 [He] furnished him with<br>
great moneys, and other necessaries. 1734 tr. C. Rollin Anc.<br>
Hist. (1827) VIII. XIX. v. 163 To make him a present of the<br>
monies arising from that sale. 1794 R. CUMBERLAND Jew II. ii. 24,<br>
Why truly, monies is a good thing. 1819 SCOTT Ivanhoe I. x*. 208<br>
`O,' said the Jew, `you are come to pay monies... And from whom<br>
dost thou bring it?' 1822 BYRON Werner II. ii, But to steal The<br>
moneys of a slumbering man! 1865 Morning Star 3 Feb. 3/5 A young<br>
woman, was charged..with stealing from the person of Robert<br>
Tharston,..7s. 6d., his moneys. 1871 R. ELLIS tr. Catullus Poems<br>
xxix. 22 Is not all his act To swallow monies, empty purses heap<br>
on heap? 1927 A. H. MCNEILE Introd. New Test. 132 Schmirdel<br>
objects that it would have been quite irrational to convey monies<br>
from South Galatia to Jerusalem by way of Macedonia. 1959 Times<br>
Rev. Industry Mar. 4/3 There is an ambivalence in the claims on<br>
promotional moneys, for the furtherance of distribution on the<br>
one hand and for the extension of advertising on the other. 1990<br>
J. MCGAHERN Amongst Women 55 He..started to tot up all the monies<br>
he presently held against the expenses he had.<br>
<br>
<br>
wind, n. (1)<br>
<br>
1. Air in motion; a state of movement in the air; a current of<br>
air, of any degree of force perceptible to the senses, occurring<br>
naturally in the atmosphere, usually parallel to the surface of<br>
the ground. a. In general or collective sense. [...]<br>
(b) pl. pl. c825 Vesp. Psalter xvii[i]. 11 [10] Volavit super<br>
pinnas ventorum, fle ofer firu winda. 971 Blickl. Hom. 51 as<br>
windas & as renas syndon ealle his. a1300 Cursor M. 22630 Windes<br>
on ilk side sal rise. 1390 GOWER Conf. I. 34 Right now the hyhe<br>
wyndes blowe. c1460 J. METHAM Wks. (1916) 157 [I]ff Crystemes day<br>
falle vpon Moneday, yt schuld be a gret wyntyr, and fulle off<br>
wyindys. a1593 MARLOWE Ovid's Elegies II. xi, Hither the winds<br>
blow, here the springtide roar. a1614 J. MELVILL Autob. & Diary<br>
(Wodrow Soc.) 261 The Lord of Armies, wha ryddes upon the winges<br>
of the woundes. 163856 COWLEY Davideis I. Notes, Wks. 1710 I.<br>
357 The Matter of Winds is an Exhalation arising out of the<br>
Concavities of the Earth. 1748 GRAY Alliance 43 Command the<br>
Winds, and tame th' unwilling Deep. 1830 TENNYSON Ode to Memory<br>
14 The dewimpearled winds of dawn. 1860 TYNDALL Glac. II. viii.<br>
263 The lighter de'bris is scattered by the winds far and wide<br>
over the glacier.<br>
<br>
<br>
rain, n. (1)<br>
<br>
2. pl. a. Showers of rain; rainfalls. a900 O.E. Martyrol. 20 Mar.<br>
40 aere lyfte ecynd is aet heo teh to a renas of aem sealtan sae.<br>
971 Blickl. Hom. 51 as windas & as renas syndon ealle his. 1154<br>
O.E. Chron. (Laud MS.) an. 1098 urh mycele renas e ealles eares<br>
ne ablunnon. c1200 Vices & Virtues 143 Godd..wiheld alle reines<br>
rie hier & six monees. a1340 HAMPOLE Psalter civ. 30 He set aire<br>
raynys haghil. c1400 MANDEVILLE (Roxb.) vii. 23 are es na<br>
trubling of e aer thurgh raynes. 1556 Chron. Gr. Friars (Camden)<br>
2 Thys yere felle gret raynes. 1625 N. CARPENTER Geog. Del. II.<br>
i. (1635) 5 The extraordinary Raines and showers which those<br>
places suffer. 1738 GRAY Tasso 10 Swoll'n with new force and late<br>
descending rains. 1878 HUXLEY Physiogr. 48 The heavy tropical<br>
rains are usually confined to definite periods. Prov. 1846<br>
DENHAM Prov. (Percy Soc.) 54 Many rains, many rowans.<br>
<br>
<br>
food, n.<br>
<br>
1. e. An article of food; a kind of food.<br>
1393 GOWER Conf. III. 26, I you shall reherce, How that my fodes<br>
ben diverse. c1449 PECOCK Repr. III. v. 303 Hauyng foodis..be we<br>
content. 1526 Pilgr. Perf. (W. de W. 1531) 5b, God sent from<br>
heuen a swete fode for theyr brede called manna. 1617 MARKHAM<br>
Caval. I. 56 In England..we have so many choyces of good foodes.<br>
1674 N. COX Gentl. Recreat. IV. (1677) 45 The larger the Pike the<br>
courser the food. 1754 Dict. Arts & Sc. II. 1288 Foods proper for<br>
preserving health. 1887 Cassell's Fam. Physician 911 What are the<br>
proper fuels, or foods, with which to supply it [the human<br>
machine].<br>
<br>
<br>
fruit, n.<br>
<br>
1. Vegetable products in general, that are fit to be used as food<br>
by men and animals. Now usually in pl. Also fruits of the earth<br>
or the ground.<br>
c1375 Lay Folks Mass Bk. (MS. B.) 392 o froytes of o erthe make<br>
plentuus. 1549 Bk. Com. Prayer, Litany, That it may please thee<br>
to give and preserve to our use the kindly fruits of the earth.<br>
1648 GAGE West Ind. xii. 43 The answer of our Queene<br>
Elizabeth..to some that presented unto her of the fruits of<br>
America. 1665 Ord. Mayor Lond. in De Foe Plague (1840) 46 That<br>
no..musty corn, or other corrupt fruits..be suffered to be sold.<br>
1725 WATTS Logic I. vi. S3 If the husk or seeds are eaten, they<br>
are called the fruits of the ground. 1791 T. NEWTE Tour Eng. &<br>
Scot. 196 At Aberdeen, turnips, carrots, and potatoes, pass,<br>
among the common people, by the name of fruit. 1859 JEPHSON<br>
Brittany ii. 20 The Breton peasant can turn all the fruits of the<br>
earth to account. c1374 CHAUCER Former Age 3 They helde hem<br>
paied of the fructes at ey ete. 150020 DUNBAR Poems xiv. 63<br>
Quhilk slayis the corne and fruct that growis grene. fig. c1374<br>
CHAUCER Boeth. I. pr. i. 3 (Camb. MS.) Thise ben tho<br>
that..destroyen the corn plentyuos of fruites of resone. 1559<br>
Mirr. Mag., Hen. VI, xxxix, See here the pleasaunt fruytes that<br>
many princes reape. 1707 WATTS Hymn, `Come, we that love the<br>
Lord' viii, Celestial Fruits on earthly Ground From Faith and<br>
Hope may grow.<br>
<br>
2. The edible product of a plant or tree, consisting of the seed<br>
and its envelope, esp. the latter when it is of a juicy pulpy<br>
nature, as in the apple, orange, plum, etc. tree of fruit =<br>
fruittree. As denoting an article of food, the word is<br>
popularly extended to include certain vegetable products that<br>
resemble `fruits' in their qualities, e.g. the stalks of<br>
rhubarb.<br>
b. with a and pl., as denoting a kind of fruit. <br>
1375 BARBOUR Bruce x. 191 The treis..Chargit vith froytis on<br>
syndri viss. c1400 Lanfranc's Cirurg. 261 ou schalt purge colre<br>
wi a decoccioun of fretis. c1460 J. RUSSELL Bk. Nurture 667<br>
Speke..For frutes afore mete to ete em fastyngely. 1527 R.<br>
THORNE in Hakluyt Voy. (1589) 252 Our fruites and graines be<br>
Apples, Nuts, and Corne. 1650 FULLER Pisgah I. iv. 11 Dates,<br>
Almonds..Nuts..Pomegranates and other severall fruits. 1795<br>
Gentl. Mag. 540/1 The glow of ripe fruits and declining leaves<br>
mark the autumn. 1842 TENNYSON Gardener's Dau. 190 Fruits and<br>
cream served in the weeping elm. 1858 HOMANS Cycl. Commerce 886<br>
This fruit [currants] is of a violet colour, and hangs in long<br>
loose bunches. 1475 Bk. Noblesse 70 Planted withe treis of<br>
verdure of divers fructis. 1585 JAS. I Ess. Poesie (Arb.) 14 To<br>
taste, and smell..Delicious fruictis, whilks in that tyme abound.<br>
1596 DALRYMPLE tr. Leslie's Hist. Scot. I. 6 Excepte spice and<br>
Vine, and sum fructes.<br>
<br>
<br>
medicine, n. (1)<br>
<br>
1. a. A substance or preparation used in the treatment of<br>
illness; a drug; esp. one taken by mouth. Also: such substances<br>
generally. Also in extended use.<br>
a1398 J. TREVISA tr. Bartholomaeus Anglicus De Proprietatibus<br>
Rerum (BL Add.) f. 99, Skabbe is curable wi metisines at..clensi<br>
wiinne & wioute. 464 M. PASTON in Paston Lett. (1971) I. 291 For<br>
Goddys sake be war what medesyns ye take of any fysissyanys of<br>
London. 1513 H. BRADSHAW Lyfe St. Werburge II. 853 All phisike<br>
and medicyns were founde to her in vayne. 1617 J. WOODALL<br>
Surgions Mate 4 Haue ready your medicines to binde vp the wound<br>
againe. 17413 J. WESLEY Extract of Jrnl. (1749) 15 One of the<br>
mistresses lay..near death, having found no help from all the<br>
medicines she had taken.<br>
<br>
{There are two additional meanings listed that take the plural,<br>
but which I have not quoted here as both are marked as Obs.}<br>
 <br>
"When we all think alike no one is thinking very much."<br>
 Walter Lippmann

Wed, 30 Apr 2008 20:52:04 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429676
sebastian gonzalez
Hi, <br>
<br>
are there any limitations to the size of the grid?<br>
<br>
and maybe a silly question, but just to be sure, there are not periodic boundary <br>
conditions?<br>
<br>
thanks<br>
Sebastian

Wed, 30 Apr 2008 21:03:08 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429678
Soren Christensen
"Helen Chen" <helen.chen@mathworks.com> wrote in message <br>
<fv82i0$nql$1@fred.mathworks.com>...<br>
> Just a reminder to let everyone know that the Spring <br>
Contest<br>
> launches tomorrow, Wednesday May 30th at high noon. Be<br>
> there or be square!<br>
> <br>
> See you then!<br>
> Helen<br>
<br>
Hi Helen,<br>
Great contest. May wires branch out? (fx. can 3 pins be <br>
connected in a 'T' formation).<br>
I might be missing it but it is not used in the examples, <br>
but not explicitly prohibited either it seems.<br>
Thanks<br>
Soren

Wed, 30 Apr 2008 21:43:40 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429684
D. Ismay
Walter Roberson wrote:<br>
> In article <a8SdnUuVdO93WIXVnZ2dnUVZ_hvnZ2d@earthlink.com>,<br>
> D. Ismay <noSpam@Woohoo.Woohoohoo.Woohoo.Woohoohoo.HOOHOO.hoohoo.Woohoo.Woohoohoo> wrote:<br>
> <br>
>> 8. 'behaviors', 'moneys', 'winds', 'rains', 'foods', 'fruits', <br>
>> 'medicines', when all<br>
>> of these are group nouns and don't change spelling for plural case.<br>
> <br>
> You are wrong about each of the words you list. <br>
<br>
bad guess, Mr. Roberson.

Wed, 30 Apr 2008 22:03:04 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429685
Laszlo
Can bridges bridge two 90 degree angled connectors in the<br>
same node?<br>
something like this <br>
<br>
//<br>
<br>
<br>
So North and West is connected to each other and South and<br>
East to each other.<br>
<br>
Laszlo<br>
<br>
"Soren Christensen" <sorench@gmail.com> wrote in message<br>
<fvamqc$r2m$1@fred.mathworks.com>...<br>
> "Helen Chen" <helen.chen@mathworks.com> wrote in message <br>
> <fv82i0$nql$1@fred.mathworks.com>...<br>
> > Just a reminder to let everyone know that the Spring <br>
> Contest<br>
> > launches tomorrow, Wednesday May 30th at high noon. Be<br>
> > there or be square!<br>
> > <br>
> > See you then!<br>
> > Helen<br>
> <br>
> Hi Helen,<br>
> Great contest. May wires branch out? (fx. can 3 pins be <br>
> connected in a 'T' formation).<br>
> I might be missing it but it is not used in the examples, <br>
> but not explicitly prohibited either it seems.<br>
> Thanks<br>
> Soren

Wed, 30 Apr 2008 22:15:54 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429688
roberson@ibd.nrccnrc.gc.ca (Walter Roberson)
In article <L_ednRtBG9SRdYXVnZ2dnUVZ_g2dnZ2d@earthlink.com>,<br>
D. Ismay <noSpam@Woohoo.Woohoohoo.Woohoo.Woohoohoo.HOOHOO.hoohoo.Woohoo.Woohoohoo> wrote:<br>
>Walter Roberson wrote:<br>
>> In article <a8SdnUuVdO93WIXVnZ2dnUVZ_hvnZ2d@earthlink.com>,<br>
>> D. Ismay <noSpam@Woohoo.Woohoohoo.Woohoo.Woohoohoo.HOOHOO.hoohoo.Woohoo.Woohoohoo> wrote:<br>
<br>
>>> 8. 'behaviors', 'moneys', 'winds', 'rains', 'foods', 'fruits', <br>
>>> 'medicines', when all<br>
>>> of these are group nouns and don't change spelling for plural case.<br>
<br>
>> You are wrong about each of the words you list. <br>
<br>
>bad guess, Mr. Roberson.<br>
<br>
No guess. I provided specific citations to a source considered by many<br>
people to be authoratative. The quotations provided in some cases extended<br>
back farther than 1000 years. It is now incumbant upon you to prove your<br>
case.<br>
<br>
<a href="http://en.wikipedia.org/wiki/Mass_noun#Multiple_senses_for_one_noun">http://en.wikipedia.org/wiki/Mass_noun#Multiple_senses_for_one_noun</a><br>
<br>
Many English count nouns can be used as mass nouns, and in these<br>
cases, they take on cumulative reference. For example, one may<br>
say that "there's apple in this sauce," and then apple has<br>
cumulative reference, and, hence, is used as a mass noun.<br>
Conversely, "fire" is generally a mass noun, but "a fire" refers<br>
to a discrete entity, and does not satisfy the criterion for<br>
cumulative reference. Two common situations of this process are<br>
when speaking of either servings/measurements of a substance<br>
("Two waters please") or of several types/varieties ("waters of<br>
the world").[2] One may say that mass nouns that are used as<br>
count nouns are "countified" and that count ones that are used as<br>
mass nouns are "massified." Some mass nouns can't easily be<br>
countified, and some count nouns are hard to massify. For example<br>
the count noun "house" is difficult to use as mass, and the mass<br>
noun "cutlery" is hard to countify:<br>
 <br>
"Ignorance has been our king... he sits unchallenged on the throne of<br>
Man. His dynasty is ageold. His right to rule is now considered<br>
legitimate. Past sages have affirmed it. They did nothing to unseat<br>
him."  Walter M Miller, Jr

Wed, 30 Apr 2008 22:27:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429691
Kenneth Eaton
I have a question that actually pertains to the contest, <br>
not dictionary definitions.<br>
<br>
With bridges, are you allowed to jump then over another <br>
pin, or just a wire connection.<br>
<br>
Ken

Wed, 30 Apr 2008 22:54:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429695
sebastian gonzalez
Hi again, <br>
this is my first constest, so sorry for ask but the darkness implies not answering <br>
the questions?<br>
<br>
just to start looking tomorrow the thread.<br>
<br>
best<br>
s

Thu, 01 May 2008 00:31:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429699
Lucio AndradeCetto
Answers to some of your questions:<br>
<br>
1. Bridges can only be used for crossovers, they can't be <br>
used to have two 90 angles a the same coordinate.<br>
2. You can not bridge pins, if I am not wrong you'll pay <br>
for them but they'll be ignored.<br>
3. The example that described the cost of the bridge should <br>
give a net score of 25+3(11*2) = 6<br>
4. The size of the boards (as well as other parameters in <br>
the testsuite) are sampled from a distribution, therefore <br>
theoretically speaking there is no limit, but in practice <br>
you could expect that most likely the distributions of the <br>
board sizes wil resemble those in the sample testsuite.<br>
5. "T" connections and "+" connections are possible and you <br>
do not pay extra but the number of wires used (3 and 4 <br>
respectively)<br>
6. Darkness does not mean "do not answer question", I'll be <br>
around for a while, let me know if you have questions.<br>
<br>
Good luck, <br>
<br>
P.S. As orginizer I get the chance to peek at some of your <br>
entries :), we have started receiving some interesting <br>
pieces of code.<br>
<br>
Lucio<br>
The MathWorks Contest Team

Thu, 01 May 2008 05:03:04 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429717
Markus Buehren
Hi,<br>
<br>
wasn't the ranking (without scores) visible in the darkness<br>
phase of the last contests? <br>
<br>
Markus

Thu, 01 May 2008 08:29:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429733
Nicholas Howe
Enjoying the new contest! Thank you all for organizing it.<br>
<br>
A question: is the routine that checks for forbidden<br>
functions available? More than once in past contests I have<br>
had entries disqualified because I accidentally left in a<br>
forbidden command I had been using for debugging.<br>
<br>
P.S. Congratulations, Alan!

Thu, 01 May 2008 16:24:05 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429792
Michael Bindschadler
"Nicholas Howe" <NikHow@hotmail.com> wrote in message<br>
<fvbv0f$b3l$1@fred.mathworks.com>...<br>
> Enjoying the new contest! Thank you all for organizing it.<br>
> <br>
> A question: is the routine that checks for forbidden<br>
> functions available? More than once in past contests I have<br>
> had entries disqualified because I accidentally left in a<br>
> forbidden command I had been using for debugging.<br>
> <br>
> P.S. Congratulations, Alan!<br>
<br>
Yes, Congrats Alan!<br>
<br>
On the other note, I would have benefited from having the<br>
routine that checks for forbidden functions! All my entries<br>
in darkness failed because I accidentally left an error()<br>
call in helper code that I copied from a function I wrote<br>
for another purpose. I assure you I had no nefarious<br>
purposes, and it would have been nice to find out they were<br>
going to fail before I submitted them. Please consider<br>
releasing this code in the future.<br>
<br>
Thanks,<br>
<br>
Mike Bindschadler

Thu, 01 May 2008 18:07:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429817
ehsan mirrahimi
The first phase of the Wiring contest has passed. At<br>
noontime today, we finished Darkness and have now entered<br>
Twilight. <br>
<br>
Things were off and running as soon as the challenge was<br>
posted. Entries were posted continually through the noon<br>
deadline. David Jones, a MATLAB Contest veteran, is the<br>
winner at the Darkness close. Congratulations David!<br>
<br>
Here were the top 10 in Darkness:<br>
<br>
1 David Jones <br>
2 Nick Howe <br>
3 Claus Still <br>
4 Steve Hoelzer <br>
5 nathan q <br>
6 Alfonso NietoCastanon<br>
7 Fabio Carnevale<br>
8 Schwabenpower<br>
9 Michael <br>
10 DreadNox <br>
<br>
Best, <br>
Helen

Thu, 01 May 2008 18:08:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429818
Matthew Simoneau
Markus, no we haven't changed anything about Darkness. It's<br>
really quite dark.<br>
<br>
Nicholas and Michael, we'll consider making this checker<br>
available in the next contest. I hate to see it surprise<br>
people at the end of Darkness. Sadly, some of our security<br>
is "security through obscurity", so we'll have to trade that<br>
off against how much security we lose.

Thu, 01 May 2008 19:02:05 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429834
Nicholas Howe
I don't know if it is permitted to ask for help on this, but<br>
I am really puzzled about an error message that my last four<br>
submissions have been generating:<br>
<br>
Line: 1 Column: 21 Unexpected MATLAB expression.<br>
<br>
The first line of my program was a comment, and I've tried<br>
changing it/deleting it/etc. but still get the same error<br>
(always Line: 1 Column: 21). The code works perfectly on my<br>
machine on the testsuite.

Thu, 01 May 2008 19:41:04 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429842
Matt Butts
While this is actually just a problem with my current<br>
algorithm, I can make a case that you should accept the<br>
empty set as a valid answer.<br>
<br>
Suppose that all pin values are on the order of 1e9. Based<br>
on the scoring of this puzzle, I can not imagine ever<br>
wanting to connect any pins. The cost of the connector far<br>
out ways the benefits of eliminating the pins.<br>
<br>
That being said, I'll go find out what is wrong with my<br>
algorithm.

Thu, 01 May 2008 20:34:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429852
Michael Bindschadler
"Matt Butts" <mattbutts@gmail.com> wrote in message<br>
<fvd6cg$of3$1@fred.mathworks.com>...<br>
> While this is actually just a problem with my current<br>
> algorithm, I can make a case that you should accept the<br>
> empty set as a valid answer.<br>
> <br>
> Suppose that all pin values are on the order of 1e9. Based<br>
> on the scoring of this puzzle, I can not imagine ever<br>
> wanting to connect any pins. The cost of the connector far<br>
> out ways the benefits of eliminating the pins.<br>
> <br>
> That being said, I'll go find out what is wrong with my<br>
> algorithm.<br>
> <br>
<br>
Actually, this is possible, and is exactly what the<br>
contestsupplied solver does. You just supply a 0 x 4<br>
matrix, for example by W = zeros(0,4).<br>
<br>
Cheers, <br>
Mike Bindschadler

Thu, 01 May 2008 20:41:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429854
Michael Bindschadler
"Matthew Simoneau" <matthew@mathworks.com> wrote in message<br>
<fvd0u3$i9i$1@fred.mathworks.com>...<br>
> Nicholas and Michael, we'll consider making this checker<br>
> available in the next contest. I hate to see it surprise<br>
> people at the end of Darkness. Sadly, some of our security<br>
> is "security through obscurity", so we'll have to trade that<br>
> off against how much security we lose.<br>
<br>
Even if you left the sneakier bits off the distribution<br>
version, it would still be helpful. Something which did a<br>
straightforward check for common functions which happen to<br>
be disallowed in the contest (error, eval,<br>
debugging/benchmarking functions) would help those of us<br>
just making honest and forgetful errors. <br>
<br>
I don't want to whine too much about it, I know it's my<br>
responsibility to make sure my code complies with the posted<br>
rules; this is just a suggestion for a nice feature for<br>
future contests.<br>
<br>
Thanks,<br>
<br>
Mike Bindschadler

Thu, 01 May 2008 21:57:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429858
Matthew Simoneau
Nicholas, it took me a while to figure this out because<br>
there was a bug in the error reporting, but your code<br>
contains TITLE a handle graphics command and on the<br>
restricted list. Sorry for the confusion.

Thu, 01 May 2008 23:31:04 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429862
Matt Butts
"Michael Bindschadler" <mikeREMOVETHISbind@gmail.com> wrote<br>
in message <fvd9fr$435$1@fred.mathworks.com>...<br>
> "Matt Butts" <mattbutts@gmail.com> wrote in message<br>
> <fvd6cg$of3$1@fred.mathworks.com>...<br>
> > While this is actually just a problem with my current<br>
> > algorithm, I can make a case that you should accept the<br>
> > empty set as a valid answer.<br>
> > <br>
> > Suppose that all pin values are on the order of 1e9. Based<br>
> > on the scoring of this puzzle, I can not imagine ever<br>
> > wanting to connect any pins. The cost of the connector far<br>
> > out ways the benefits of eliminating the pins.<br>
> > <br>
> > That being said, I'll go find out what is wrong with my<br>
> > algorithm.<br>
> > <br>
> <br>
> Actually, this is possible, and is exactly what the<br>
> contestsupplied solver does. You just supply a 0 x 4<br>
> matrix, for example by W = zeros(0,4).<br>
> <br>
> Cheers, <br>
> Mike Bindschadler<br>
<br>
Thanks for the clarification. I was mistakenly assuming that<br>
[] would be treated the same as a 0x4 matrix.

Fri, 02 May 2008 07:59:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429892
Alan Chalker
Sorry all, but it appears I have the dubious honor of being<br>
the first to crash the queue this time around. I was trying<br>
to determine the scoring coefficients and submitted a<br>
program with what should be a very high node count. <br>
However, after testing it on my version of R2008a, it<br>
appears that mtree 'wraps around' the node count as a fault<br>
and it's returning a value of 1 for the node count.

Fri, 02 May 2008 08:24:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429894
Alan Chalker
Even though I accidentally crashed the queue, I think I have<br>
been able to figure out the scoring formula and am posting<br>
it here as I traditionally do. I’ve determined it’s very<br>
similar to the recent contests:<br>
<br>
score = k1*result + k2*e(k3*runtime) +<br>
k4*max(complexity10,0) + k5*nodes<br>
<br>
Where:<br>
<br>
k1 = 0.1<br>
k2 = 2<br>
k3 = 2/30 (0.06666666…)<br>
k4 = 1<br>
k5 = 0.001<br>
<br>
The current leading entry has a time of 56s, result of<br>
141891, cyc of 25, and nodes of 4196. Here’s a breakdown of<br>
the current tradoffs:<br>
<br>
cyc and score are a 1:1 ratio (i.e. each point shaved off<br>
cyc is a point shaved off the score)<br>
time and score are a 1:5.7 ratio<br>
result and score are a 1:0.1 ratio<br>
node and score are a 1:0.001 ratio<br>
<br>
David Jones entries have already settled in just at the<br>
‘knee’ of the time exponential curve, which is rather flat<br>
until about ~50s. However, because of the new time exponent<br>
constant, we are going to see much more payoff in this<br>
content in focusing on reducing the execution time versus<br>
other scoring elements, probably down to the ~10 second range.<br>
<br>
Unfortunately that probably also means that people are going<br>
to end up taking the lead due to ‘luck of the draw’ in the<br>
minor variations we always see in execution times, since<br>
they will be amplified more in the total score compared to<br>
in the past.<br>
<br>
Hope this helps everyone!

Fri, 02 May 2008 08:24:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429895
Matthew Simoneau
Yes, our code wasn't robust to when the file was too big for<br>
MLint to handle and the queue was down for a bit. We're<br>
back up and running. Sorry for the delay of game.

Fri, 02 May 2008 15:22:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429946
Nicholas Howe
"Matthew Simoneau" <matthew@mathworks.com> wrote in message<br>
<fvdebf$p7p$1@fred.mathworks.com>...<br>
> Nicholas, it took me a while to figure this out because<br>
> there was a bug in the error reporting, but your code<br>
> contains TITLE a handle graphics command and on the<br>
> restricted list. Sorry for the confusion.<br>
<br>
Thanks for looking into this for me. I think I ended up<br>
removing that call to title soon afterwards.<br>
<br>
Another request, if anyone has time: does anyone have the<br>
matrix for Lucio's example problem? I'd be interested in<br>
trying my code on it. Thanks!

Fri, 02 May 2008 16:28:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429964
David
Thanks to the Matlab Contest Team for dreaming up a great wiring puzzle. It <br>
has been fun working out incremental improvements through Darkness and <br>
Twilight, especially with the wiringGUI and Lucio's analysis to visualize solutions.<br>
<br>
Judging from scores during Twilight, it seems that finding good bridges <br>
efficiently is what separated my Twilight algorithm from the rest of the pack.<br>
<br>
But now as we enter the Daylight phase, all my coding secrets will be revealed, <br>
... so roll up your sleeves ... and let the tweakfest begin ...<br>
<br>
Good luck to everyone for the rest of the contest!<br>
<br>
 David Jones

Fri, 02 May 2008 18:41:39 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#429987
Matthew Simoneau
Nick, I posted Lucio's test board to the blog:<br>
<br>
<a href="http://blogs.mathworks.com/contest/2008/05/01/sneakpeekleadingsolversatwork/#comment5149">http://blogs.mathworks.com/contest/2008/05/01/sneakpeekleadingsolversatwork/#comment5149</a>

Sat, 03 May 2008 13:39:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430071
the cyclist
"Helen Chen" <helen.chen@mathworks.com> wrote in message<br>
<fv82i0$nql$1@fred.mathworks.com>...<br>
> Just a reminder to let everyone know that the Spring Contest<br>
> launches tomorrow, Wednesday May 30th at high noon. Be<br>
> there or be square!<br>
> <br>
> See you then!<br>
> Helen<br>
<br>
It seems that much of the Statistics page is not being<br>
updated properly.

Sat, 03 May 2008 21:15:05 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430103
Helen Chen
"the cyclist" <thecyclist@letter.after.f.mail.com> wrote in<br>
message <fvhptn$kgm$1@fred.mathworks.com>...<br>
<br>
> It seems that much of the Statistics page is not being<br>
> updated properly.<br>
<br>
Thanks for the headsup. We'll take a look at this. <br>
<br>
Helen

Sat, 03 May 2008 23:53:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430110
Sergey
Here I want to provide Markus's post from previous contest:<br>
<br>
"As in all contests, some guys tend to ruin the contest by<br>
obfuscating their code, as DrSeuss does at the moment. <br>
<br>
In the contest rules under "Hacking" we read "we ask that<br>
you not abuse the system." I think this should also be valid<br>
for the annoying obfuscation of code.<br>
<br>
In the current leading code, I still find large portions of<br>
my twilight winning code. I insist that you at least do not<br>
obfuscate code that others have written!! If you want<br>
introduce a new variable, call it as you like, but leave the<br>
others as they are!<br>
<br>
Markus<br>
"<br>
<br>
SY

Mon, 05 May 2008 07:43:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430265
Alan Chalker
As I usually do, I've now posted a heavily commented version<br>
of the leading code so that those of you who aren't in the<br>
'thick of it' can have an opportunity to understand the<br>
alogrithms behind the solutions to this problem.<br>
<br>
The code is entry 47616 viewable at:<br>
<a href="http://www.mathworks.com/contest/wiring.cgi/view_submission.html?id=47616">http://www.mathworks.com/contest/wiring.cgi/view_submission.html?id=47616</a><br>
<br>
Hope this helps!

Mon, 05 May 2008 12:09:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430308
Sergey
Hi.<br>
Would it be possible to announce in advance if we are going <br>
to have (and when) late twilight, best result, 1000 <br>
character, generality competition? <br>
SY (Sergey)

Mon, 05 May 2008 16:37:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430380
David
When did (or will) the scoreneutral switch in test suite occur??<br>
<br>
 David Jones

Mon, 05 May 2008 17:15:08 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430386
Sergey
Anybody wants to play “best result” with 3 min time limit <br>
today?<br>
Let’s say between 6PM and 9PM ET?<br>
<br>
SY

Mon, 05 May 2008 17:26:04 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430390
srach
"Sergey " <ivssnn@yahoo.com> wrote in message<br>
<fvnfas$cmt$1@fred.mathworks.com>...<br>
> Anybody wants to play “best result” with 3 min time limit <br>
> today?<br>
> Let’s say between 6PM and 9PM ET?<br>
> <br>
> SY<br>
> <br>
<br>
Unfortunately, that's right in the middle of the night in<br>
Europe. But have fun anyway. :)<br>
<br>
<br>
Btw. the leading entry "does it shorten time" by Gwendolyn<br>
Fish was just a moment ago in back in the queue? Does this<br>
announce the swapping of the test suite?<br>
<br>
Regards,<br>
srach

Mon, 05 May 2008 17:45:06 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430393
Markus Buehren
Hi,<br>
<br>
I follow Sergey and would like to know when the next contest<br>
phases will end. I have this HUGE improvement and just wait<br>
to plug it in at the right time :)<br>
<br>
Markus

Mon, 05 May 2008 19:59:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430422
Matthew Simoneau
I ran into some bumps with the test suite swap, but we<br>
should be back in business.<br>
<br>
We're now running a "first to break 13000", which could take<br>
some time. I picked this target by figuring that most of<br>
the difference in results between the old and new test<br>
suites could be made up without major innovation, but there<br>
is always some risk that the target is too aggressive.<br>
<br>
The "best result" really clogged the queue last time, so<br>
instead we're bringing back a variation on the "1000<br>
Character Challenge". This time, however, we're be using<br>
node count rather than character count. Since the top entry<br>
is now around 4500 nodes, maybe 1000 node limit? Or even<br>
more restrictive? I like this one because it gives a chance<br>
for people to work with shorter code for a while, and now<br>
that we're using nodes instead of characters there won't be<br>
as much pressure to make the code illegible. We'll do this<br>
sometime tomorrow.<br>
<br>
Historically, we've announced the midcontest prizes as we<br>
go so we could react to whatever was going on. We've picked<br>
up some traditions, like the Sunday Push. I'll float the<br>
idea to the team of announcing at least most of them in<br>
advance next time.

Tue, 06 May 2008 17:36:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430626
Helen Chen
"Markus Buehren" <mb_matlab.REMOVE@gmxTHIS.de> wrote in<br>
message <fvnh32$q3s$1@fred.mathworks.com>...<br>
<br>
> I follow Sergey and would like to know when the next contest<br>
> phases will end. I have this HUGE improvement and just wait<br>
> to plug it in at the right time :)<br>
> <br>
<br>
Markus  <br>
<br>
Did you see Matt's note about the schedule on the contest<br>
blog ( <a href="http://blogs.mathworks.com/contest/">http://blogs.mathworks.com/contest/</a> )? The challenge<br>
to beat 13000 will remain open until someone beats it, but<br>
we have a concurrent second  the 1000 node challenge. <br>
<br>
Wouldn't it be interesting if someone won both challenges<br>
with the same submission? It would be another first for the<br>
MATLAB Contest!<br>
<br>
Helen

Tue, 06 May 2008 19:23:04 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430644
Markus Buehren
> Markus  <br>
> <br>
> Did you see Matt's note about the schedule on the contest<br>
> blog ( <a href="http://blogs.mathworks.com/contest/">http://blogs.mathworks.com/contest/</a> )? <br>
<br>
Sure, but the note was posted later than mine :)<br>
<br>
Markus

Tue, 06 May 2008 22:07:04 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430673
the cyclist
"the cyclist" <thecyclist@letter.after.f.mail.com> wrote in<br>
message <fvhptn$kgm$1@fred.mathworks.com>...<br>
> "Helen Chen" <helen.chen@mathworks.com> wrote in message<br>
> <fv82i0$nql$1@fred.mathworks.com>...<br>
> > Just a reminder to let everyone know that the Spring Contest<br>
> > launches tomorrow, Wednesday May 30th at high noon. Be<br>
> > there or be square!<br>
> > <br>
> > See you then!<br>
> > Helen<br>
> <br>
> It seems that much of the Statistics page is not being<br>
> updated properly.<br>
<br>
Stats page seems to be stuck again (Tuesday at 5:40 PM, just<br>
before end of 1000mode contest).

Wed, 07 May 2008 00:10:06 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430688
Markus Buehren
The statistics page is still down, so I parsed the results<br>
to find out who the winner of 1000nodeschallenge is. Here<br>
is the (inofficial) top ten:<br>
<br>
01. 15484.3796: Uss tc <br>
02. 15484.4924: node speed YC<br>
03. 15485.2384: M&M_262 MikeR<br>
04. 15485.3887: node speed YC<br>
05. 15485.5323: M&M_259 MikeR<br>
06. 15485.7220: super speed 2 YC<br>
07. 15485.8111: super speed YC<br>
08. 15485.8763: super speed 1 YC<br>
09. 15485.8969: aam333 Abhisek Ukil<br>
10. 15485.9271: node speed 3 YC<br>
<br>
It seems tc aka The Cyclist has won another contest prize.<br>
Well done! Check out which small tweak he used:<br>
<br>
<a href="http://www.mathworks.com/contest/wiring.cgi/diff.html?id1=48627&id2=48680">http://www.mathworks.com/contest/wiring.cgi/diff.html?id1=48627&id2=48680</a><br>
<br>
Yours<br>
Markus

Wed, 07 May 2008 00:11:04 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430690
Markus Buehren
Grmpf, just posting this I notice that the Statistics page<br>
is up again...

Wed, 07 May 2008 00:36:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430693
Helen Chen
"the cyclist" <thecyclist@letter.after.f.mail.com> wrote in<br>
message <fvqkq7$ae2$1@fred.mathworks.com>...<br>
> "the cyclist" <thecyclist@letter.after.f.mail.com> wrote in<br>
> message <fvhptn$kgm$1@fred.mathworks.com>...<br>
<br>
> Stats page seems to be stuck again (Tuesday at 5:40 PM, just<br>
> before end of 1000mode contest).<br>
<br>
I think it was quite backlogged with all the last minute<br>
postings before the 6 pm deadline as the queue has just<br>
cleared up. <br>
<br>
Helen

Wed, 07 May 2008 05:17:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430739
Alan Chalker
While I haven't had much time to compete in this contest,<br>
I've done some analysis and unfortunately believe that the<br>
13000 mark is going to be virtually impossible to beat. <br>
Here are some key points:<br>
<br>
As of midnight Tuesday, the leading entry has the following<br>
stats:<br>
Result: 134617<br>
Time: 40.0336<br>
Cyc: 21<br>
Nodes: 7289<br>
Score: 13508.84<br>
<br>
Because of the way the scoring formula works, the ONLY way<br>
to beat 13000 is to lower the result. Even if the time, cyc<br>
and nodes were reduced to 1 each, the score would only<br>
improve by 45 points to 13464. In order to reach 13000,<br>
the result needs to improve to ~129500, or by approximately<br>
5117 (which is a 3.8% improvement). <br>
<br>
That might not seem like much, but it actually is. Based<br>
upon the null move solvers submitted at the start of the<br>
contest, we know that there are a total of 359141 points in<br>
the original test suite. The current result is only 37% of<br>
the max value of the test boards. <br>
<br>
However, keep in mind that there is a nonzero lower limit<br>
to the score for a 'perfect solver'. While it's impossible<br>
for us to know what that is without examining the test<br>
suite, we can make some good guesses based upon the best<br>
results statistics. The best result so far is 133225,<br>
almost 1400 points better than the current leader (although<br>
it took 148 seconds to run), but still 3700 points away from<br>
13000, assuming it could run in less than 1/3 the time it<br>
currently does.<br>
<br>
Another way of looking at it is the % improvement column on<br>
the stats webpage. The score has only improved by ~4% since<br>
Sunday morning until right now. Thus in order to break the<br>
goal, in the next 12 hours entries need to outperform ~72<br>
hours worth of steady improvements.<br>
<br>
Examining the current leading solver against random<br>
individual boards in the sample test suite, I found that<br>
there might be room for slight improvement, however overall<br>
it's doing a really good job, as one might expect. <br>
<br>
Thus, while I'm sorry if this bursts anyone's bubbles,<br>
hopefully this will allow some competitors to refocus on<br>
just getting a top entry instead of trying to beat an<br>
arbitrary score. Good luck everyone for the remainder of<br>
the contest!

Wed, 07 May 2008 06:37:02 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430749
Luigi Sorbara
Buddy .. stop the tweak bombing .. some of us would like to <br>
submit their solutions .. simply BRUTAL<br>
<br>
"Alan Chalker" <alancNOSPAM@osc.edu> wrote in message <br>
<fvre0f$j6$1@fred.mathworks.com>...<br>
> While I haven't had much time to compete in this contest,<br>
> I've done some analysis and unfortunately believe that the<br>
> 13000 mark is going to be virtually impossible to beat. <br>
> Here are some key points:<br>
> <br>
> As of midnight Tuesday, the leading entry has the <br>
following<br>
> stats:<br>
> Result: 134617<br>
> Time: 40.0336<br>
> Cyc: 21<br>
> Nodes: 7289<br>
> Score: 13508.84<br>
> <br>
> Because of the way the scoring formula works, the ONLY way<br>
> to beat 13000 is to lower the result. Even if the time, <br>
cyc<br>
> and nodes were reduced to 1 each, the score would only<br>
> improve by 45 points to 13464. In order to reach 13000,<br>
> the result needs to improve to ~129500, or by <br>
approximately<br>
> 5117 (which is a 3.8% improvement). <br>
> <br>
> That might not seem like much, but it actually is. Based<br>
> upon the null move solvers submitted at the start of the<br>
> contest, we know that there are a total of 359141 points <br>
in<br>
> the original test suite. The current result is only 37% <br>
of<br>
> the max value of the test boards. <br>
> <br>
> However, keep in mind that there is a nonzero lower limit<br>
> to the score for a 'perfect solver'. While it's <br>
impossible<br>
> for us to know what that is without examining the test<br>
> suite, we can make some good guesses based upon the best<br>
> results statistics. The best result so far is 133225,<br>
> almost 1400 points better than the current leader <br>
(although<br>
> it took 148 seconds to run), but still 3700 points away <br>
from<br>
> 13000, assuming it could run in less than 1/3 the time it<br>
> currently does.<br>
> <br>
> Another way of looking at it is the % improvement column <br>
on<br>
> the stats webpage. The score has only improved by ~4% <br>
since<br>
> Sunday morning until right now. Thus in order to break <br>
the<br>
> goal, in the next 12 hours entries need to outperform ~72<br>
> hours worth of steady improvements.<br>
> <br>
> Examining the current leading solver against random<br>
> individual boards in the sample test suite, I found that<br>
> there might be room for slight improvement, however <br>
overall<br>
> it's doing a really good job, as one might expect. <br>
> <br>
> Thus, while I'm sorry if this bursts anyone's bubbles,<br>
> hopefully this will allow some competitors to refocus on<br>
> just getting a top entry instead of trying to beat an<br>
> arbitrary score. Good luck everyone for the remainder of<br>
> the contest!<br>
> <br>
>

Wed, 07 May 2008 06:46:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430750
Jin
why bombing the queue? <br>
I sugguest adding some submission limits, such as 1 <br>
entry/per min/per ip.

Wed, 07 May 2008 06:49:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430751
Jin
why bombing? when is the end?

Wed, 07 May 2008 08:23:05 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430768
Markus Buehren
Alans bombing as well as his analysis both speak for adding<br>
a "captcha" to the submission site (which I think is even<br>
better than an IP limit). This would prevent the<br>
overoveroverspecification of the solver to the test suite<br>
which occurs due to the many parameter variations that are<br>
submitted. <br>
<br>
I have found some algorithmic improvement yesterday, which<br>
dramatically boosted the score on the test suite we are<br>
having at home. However, as the leading code is so<br>
overspecified to the actual test suite, my change even lead<br>
to a worse score on it. I think the parameter tweaking by<br>
numerous automated uploads takes away the room for<br>
algorithmic improvements.<br>
<br>
The most obvious sign for tweaking is the introduction of<br>
randomness into the algorithm. Not only that parameters of<br>
the code are varied, we even don't know which parameters<br>
that are! In my opinion the use of random generators (rand,<br>
randn, randperm etc.) should be forbidden.<br>
<br>
Any other thoughts?<br>
<br>
Markus

Wed, 07 May 2008 08:47:04 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430771
srach
> I have found some algorithmic improvement yesterday, which<br>
> dramatically boosted the score on the test suite we are<br>
> having at home. However, as the leading code is so<br>
> overspecified to the actual test suite, my change even lead<br>
> to a worse score on it. I think the parameter tweaking by<br>
> numerous automated uploads takes away the room for<br>
> algorithmic improvements.<br>
<br>
But sometimes such differences also come with different<br>
matlab versions or different machines. Earlier in the<br>
contest, I found a small change that led to an improvement<br>
of about 10 sec on my computer. I checked that on a<br>
different computer (with a more recent matlab version) and<br>
there this change led to a worsening of 10 sec. <br>
<br>
(Same was true at the end of the last contest, where the<br>
final tweak that I saved for the last seconds of the contest<br>
actually made the score worse, although it led to<br>
improvements on my system.)<br>
<br>
I second your thoughts about randomness. <br>
<br>
Regards,<br>
srach

Wed, 07 May 2008 08:47:04 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430773
srach
> I have found some algorithmic improvement yesterday, which<br>
> dramatically boosted the score on the test suite we are<br>
> having at home. However, as the leading code is so<br>
> overspecified to the actual test suite, my change even lead<br>
> to a worse score on it. I think the parameter tweaking by<br>
> numerous automated uploads takes away the room for<br>
> algorithmic improvements.<br>
<br>
But sometimes such differences also come with different<br>
matlab versions or different machines. Earlier in the<br>
contest, I found a small change that led to an improvement<br>
of about 10 sec on my computer. I checked that on a<br>
different computer (with a more recent matlab version) and<br>
there this change led to a worsening of 10 sec. <br>
<br>
(Same was true at the end of the last contest, where the<br>
final tweak that I saved for the last seconds of the contest<br>
actually made the score worse, although it led to<br>
improvements on my system.)<br>
<br>
I second your thoughts about randomness. <br>
<br>
Regards,<br>
srach

Wed, 07 May 2008 08:47:25 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430774
srach
> I have found some algorithmic improvement yesterday, which<br>
> dramatically boosted the score on the test suite we are<br>
> having at home. However, as the leading code is so<br>
> overspecified to the actual test suite, my change even lead<br>
> to a worse score on it. I think the parameter tweaking by<br>
> numerous automated uploads takes away the room for<br>
> algorithmic improvements.<br>
<br>
But sometimes such differences also come with different<br>
matlab versions or different machines. Earlier in the<br>
contest, I found a small change that led to an improvement<br>
of about 10 sec on my computer. I checked that on a<br>
different computer (with a more recent matlab version) and<br>
there this change led to a worsening of 10 sec. <br>
<br>
(Same was true at the end of the last contest, where the<br>
final tweak that I saved for the last seconds of the contest<br>
actually made the score worse, although it led to<br>
improvements on my system.)<br>
<br>
I second your thoughts about randomness. <br>
<br>
Regards,<br>
srach

Wed, 07 May 2008 09:31:05 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430778
OkinawaDolphin
> The most obvious sign for tweaking is the introduction of<br>
> randomness into the algorithm. Not only that parameters of<br>
> the code are varied, we even don't know which parameters<br>
> that are! In my opinion the use of random generators <br>
(rand,<br>
> randn, randperm etc.) should be forbidden.<br>
> <br>
> Any other thoughts?<br>
<br>
If randomness is forbidden, genetic algorithms, simulated <br>
annealing or even a random walk are forbidden, too. Only <br>
deterministic algorithms are allowed in this case, no <br>
matter how inefficient they might be.<br>
<br>
Tweaking can be prevented by removing or deactivating the <br>
functionality for reading and editing existing code. <br>
However, tweaking is allowed in this contest and it is done <br>
by people who understand the code others wrote.

Wed, 07 May 2008 09:46:38 +0000
How long does it Take?
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430782
OkinawaDolphin
I submitted my latest entry 30 minutes ago and it has still <br>
the stats "new". Is the server down?

Wed, 07 May 2008 10:18:03 +0000
Re: How long does it Take?
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430786
OkinawaDolphin
My entry does not show up even after an hour. Tweak bombing <br>
should indeed be forbidden and prevented from happening by <br>
technical measures.

Wed, 07 May 2008 11:38:03 +0000
Re: How long does it Take?
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430798
Helen Chen
"OkinawaDolphin " <OkinawaDolphin@Hotmail.com> wrote in<br>
message <fvrvkr$688$1@fred.mathworks.com>...<br>
> My entry does not show up even after an hour. Tweak bombing <br>
> should indeed be forbidden and prevented from happening by <br>
> technical measures.<br>
<br>
Sorry all, the queue did fall asleep, but is running now.<br>
Things should clear out in a bit. <br>
<br>
Helen

Wed, 07 May 2008 11:54:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430800
Matt Butts
Contest ends at 12PM Eastern today, correct?

Wed, 07 May 2008 12:10:05 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430803
Nathan
Hi all<br>
<br>
In my opinion, wholesale tweakbombing, on a scale large<br>
enough to monopolise the queue for hours, is contrary to the<br>
spirit of the contest. Arguably, it's a form of hacking <br>
"Entries that compromise the contest machinery are no longer<br>
allowed". There is probably no technical solution that would<br>
prevent it (and anyway it's unfair to expect the Mathworks<br>
team to work as enforcers). However, I agree with Markus<br>
that a captcha would be a workable and useful measure. <br>
<br>
In principle, I would like to see random functions outlawed<br>
 however, competitors could easily circumvent it with<br>
handwritten random number generators (they wouldn't have to<br>
very good ones) which would lead to less intelligible code. <br>
<br>
In general I'm in favour of a light touch with rules and<br>
regulations. New restrictions will act as targets to be<br>
hacked and evaded. Fun and a social spirit should stay at<br>
the centre of the contest. <br>
<br>
Nathan<br>
<br>
PS Although Alan defeated the online diff function, CSdiff<br>
on windows (or presumably a unix diff) will reveal what he<br>
is up to :)<br>
<br>
"Markus Buehren" <mb_matlab.REMOVE@gmxTHIS.de> wrote in<br>
message <fvrot9$gsi$1@fred.mathworks.com>...<br>
> Alans bombing as well as his analysis both speak for adding<br>
> a "captcha" to the submission site (which I think is even<br>
> better than an IP limit). This would prevent the<br>
> overoveroverspecification of the solver to the test suite<br>
> which occurs due to the many parameter variations that are<br>
> submitted. <br>
> <br>
> I have found some algorithmic improvement yesterday, which<br>
> dramatically boosted the score on the test suite we are<br>
> having at home. However, as the leading code is so<br>
> overspecified to the actual test suite, my change even lead<br>
> to a worse score on it. I think the parameter tweaking by<br>
> numerous automated uploads takes away the room for<br>
> algorithmic improvements.<br>
> <br>
> The most obvious sign for tweaking is the introduction of<br>
> randomness into the algorithm. Not only that parameters of<br>
> the code are varied, we even don't know which parameters<br>
> that are! In my opinion the use of random generators (rand,<br>
> randn, randperm etc.) should be forbidden.<br>
> <br>
> Any other thoughts?<br>
> <br>
> Markus<br>
>

Wed, 07 May 2008 13:24:04 +0000
Re: How long does it Take?
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430817
srach
"Helen Chen" <helen.chen@mathworks.com> wrote in message<br>
<fvs4ar$dl5$1@fred.mathworks.com>...<br>
> "OkinawaDolphin " <OkinawaDolphin@Hotmail.com> wrote in<br>
> message <fvrvkr$688$1@fred.mathworks.com>...<br>
> > My entry does not show up even after an hour. Tweak bombing <br>
> > should indeed be forbidden and prevented from happening by <br>
> > technical measures.<br>
> <br>
> Sorry all, the queue did fall asleep, but is running now.<br>
> Things should clear out in a bit. <br>
> <br>
> Helen<br>
<br>
Is the queue sleeping again? <br>
<br>
Best,<br>
srach

Wed, 07 May 2008 14:13:04 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430824
Markus Buehren
> If randomness is forbidden, genetic algorithms, simulated <br>
> annealing or even a random walk are forbidden, too. Only <br>
> deterministic algorithms are allowed in this case, no <br>
> matter how inefficient they might be.<br>
<br>
In my view, you compare two different things here: Genetic<br>
algorithms and such are tools for finding an optimal<br>
parameter set. The solver algorithm itself is then used for<br>
solving the problem, using the parameters found in a step<br>
before. <br>
<br>
If you look at the entries using randomness in this contest,<br>
there is not the slightest sign that something like a<br>
genetic algorithm is applied there.<br>
<br>
> Tweaking can be prevented by removing or deactivating the <br>
> functionality for reading and editing existing code. <br>
<br>
No, there are simple ways to upload submissions<br>
automatically. You can even do it simply from within Matlab!<br>
I know which Matlab function one can use for it, but I won't<br>
tell it to anyone now :)<br>
<br>
> However, tweaking is allowed in this contest and it is done <br>
> by people who understand the code others wrote. <br>
<br>
No!! For tweaking, you don't need any clue about how the<br>
code itself works! Just play around a bit with the<br>
parameters, let one solver run more than once and select the<br>
best result, or combine two different solvers and choose the<br>
better result and sometimes you will find an improvement.<br>
You don't need to understand the algorithms themselfes for that.<br>
<br>
Markus

Wed, 07 May 2008 14:18:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430825
Luigi Sorbara
Thank you Nathan, well said. I would assume the majority <br>
of people participate in this contest out of interest and <br>
challenge (and maybe the opportunity to work for <br>
matlab :) ). I have limited time in the evening once I <br>
get home from work and do my husbandly chores .. to <br>
actually participate. There is nothing more frustrating <br>
then waiting over 40 minutes to see how a submission <br>
faired whilst looking at hundreds of tweak bomb entries. <br>
Helen .. any chance we can extend the contest one more <br>
night for us true lovers of the game :)<br>
<br>
"Nathan " <nathoqXXX@yahooXXX.com> wrote in message <br>
<fvs66t$8l4$1@fred.mathworks.com>...<br>
> Hi all<br>
> <br>
> In my opinion, wholesale tweakbombing, on a scale large<br>
> enough to monopolise the queue for hours, is contrary to <br>
the<br>
> spirit of the contest. Arguably, it's a form of hacking <br>
> "Entries that compromise the contest machinery are no <br>
longer<br>
> allowed". There is probably no technical solution that <br>
would<br>
> prevent it (and anyway it's unfair to expect the <br>
Mathworks<br>
> team to work as enforcers). However, I agree with Markus<br>
> that a captcha would be a workable and useful measure. <br>
> <br>
> In principle, I would like to see random functions <br>
outlawed<br>
>  however, competitors could easily circumvent it with<br>
> handwritten random number generators (they wouldn't have <br>
to<br>
> very good ones) which would lead to less intelligible <br>
code. <br>
> <br>
> In general I'm in favour of a light touch with rules and<br>
> regulations. New restrictions will act as targets to be<br>
> hacked and evaded. Fun and a social spirit should stay at<br>
> the centre of the contest. <br>
> <br>
> Nathan<br>
> <br>
> PS Although Alan defeated the online diff function, <br>
CSdiff<br>
> on windows (or presumably a unix diff) will reveal what <br>
he<br>
> is up to :)<br>
> <br>
> "Markus Buehren" <mb_matlab.REMOVE@gmxTHIS.de> wrote in<br>
> message <fvrot9$gsi$1@fred.mathworks.com>...<br>
> > Alans bombing as well as his analysis both speak for <br>
adding<br>
> > a "captcha" to the submission site (which I think is <br>
even<br>
> > better than an IP limit). This would prevent the<br>
> > overoveroverspecification of the solver to the test <br>
suite<br>
> > which occurs due to the many parameter variations that <br>
are<br>
> > submitted. <br>
> > <br>
> > I have found some algorithmic improvement yesterday, <br>
which<br>
> > dramatically boosted the score on the test suite we are<br>
> > having at home. However, as the leading code is so<br>
> > overspecified to the actual test suite, my change <br>
even lead<br>
> > to a worse score on it. I think the parameter tweaking <br>
by<br>
> > numerous automated uploads takes away the room for<br>
> > algorithmic improvements.<br>
> > <br>
> > The most obvious sign for tweaking is the introduction <br>
of<br>
> > randomness into the algorithm. Not only that <br>
parameters of<br>
> > the code are varied, we even don't know which <br>
parameters<br>
> > that are! In my opinion the use of random generators <br>
(rand,<br>
> > randn, randperm etc.) should be forbidden.<br>
> > <br>
> > Any other thoughts?<br>
> > <br>
> > Markus<br>
> > <br>
>

Wed, 07 May 2008 14:41:21 +0000
Re: How long does it Take?
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430830
Helen Chen
"srach " <s.rach_NOSPAM_@jacobsuniversity.de> wrote in<br>
message <fvsahk$jks$1@fred.mathworks.com>...<br>
> <br>
> Is the queue sleeping again? <br>
> <br>
Sorry, I was in a meeting, but yes, they are asleep again. <br>
Matt's taking a look and we should have them back up in a bit. <br>
<br>
Helen

Wed, 07 May 2008 14:55:06 +0000
Re: How long does it Take?
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430831
Matthew Simoneau
Sorry, the queue was down for about two hours this morning,<br>
but we're back up and running now.

Wed, 07 May 2008 15:24:03 +0000
Re: How long does it Take?
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430835
Alan Chalker
I'd like to take a moment and respond to a couple of points.<br>
<br>
Yes, I tweak bombed last night in an attempt to find some<br>
optimal parameters for the various constants in the code. I<br>
deliberately waited until the wee hours of the morning to do<br>
this since the majority of the leading players seem to<br>
compete during USA daylight hours. I'm sorry if this<br>
inconvenienced some other players though.<br>
<br>
I also only submitted enough entries to tie up the queue for<br>
a couple hours, and during a time when there weren't any<br>
other 'active' subprizes people were competing for, trying<br>
to be considerate of the fact that the contest end is<br>
nearing and people want to get feedback on their codes.<br>
Since I used an automated script (written in MATLAB of<br>
course;) to do this, I (or another player with less<br>
integrity) could easily tie up the queue for the remainder<br>
of the contest.<br>
<br>
Please note that the rules explicitly ALLOW tweak bombing:<br>
"Tuning the entry to the contest test suite via tweak<br>
bombing or other techniques is still allowed, but we ask<br>
that you not overwhelm the queue." However, how does one<br>
define 'overwhelming' the queue? This has been a reoccuring<br>
issue in past contests and there have been various<br>
suggestions, such as CAPTCHAs or random seeds to the RAND<br>
function, which I think should be strongly considered.<br>
<br>
Generally I'm not in favor of this type of 'style of play',<br>
however I saw what I thought was an opportunity to be<br>
involved in a way I haven't been involved before and thought<br>
I'd try it out. As an aside, since the queue is now at the<br>
point where new entries aren't going to run until after the<br>
contest closes, I intend to make one last stab at this in<br>
the off chance I hit upon just the right parameters.

Wed, 07 May 2008 16:04:03 +0000
Re: How long does it Take?
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430841
srach
The queue is closed. Good luck to all of you!<br>
<br>
Best,<br>
srach

Wed, 07 May 2008 16:05:06 +0000
Re: How long does it Take?
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430842
Nathan
Hi Alan<br>
<br>
I would consider the queue "overwhelmed" when it's blocked<br>
for several hours! There are many "leading players", and<br>
nonleading ones too, who do their work at times other than<br>
American daylight. <br>
<br>
ooops... while writing this I nearly forgot to post my<br>
lastminute tweaks, erm, innovations...<br>
<br>
Good luck<br>
<br>
Nathan<br>
<br>
"Alan Chalker" <alancNOSPAM@osc.edu> wrote in message<br>
<fvshij$ipo$1@fred.mathworks.com>...<br>
> I'd like to take a moment and respond to a couple of points.<br>
> <br>
> Yes, I tweak bombed last night in an attempt to find some<br>
> optimal parameters for the various constants in the code. I<br>
> deliberately waited until the wee hours of the morning to do<br>
> this since the majority of the leading players seem to<br>
> compete during USA daylight hours. I'm sorry if this<br>
> inconvenienced some other players though.<br>
> <br>
> I also only submitted enough entries to tie up the queue for<br>
> a couple hours, and during a time when there weren't any<br>
> other 'active' subprizes people were competing for, trying<br>
> to be considerate of the fact that the contest end is<br>
> nearing and people want to get feedback on their codes.<br>
> Since I used an automated script (written in MATLAB of<br>
> course;) to do this, I (or another player with less<br>
> integrity) could easily tie up the queue for the remainder<br>
> of the contest.<br>
> <br>
> Please note that the rules explicitly ALLOW tweak bombing:<br>
> "Tuning the entry to the contest test suite via tweak<br>
> bombing or other techniques is still allowed, but we ask<br>
> that you not overwhelm the queue." However, how does one<br>
> define 'overwhelming' the queue? This has been a reoccuring<br>
> issue in past contests and there have been various<br>
> suggestions, such as CAPTCHAs or random seeds to the RAND<br>
> function, which I think should be strongly considered.<br>
> <br>
> Generally I'm not in favor of this type of 'style of play',<br>
> however I saw what I thought was an opportunity to be<br>
> involved in a way I haven't been involved before and thought<br>
> I'd try it out. As an aside, since the queue is now at the<br>
> point where new entries aren't going to run until after the<br>
> contest closes, I intend to make one last stab at this in<br>
> the off chance I hit upon just the right parameters.<br>
> <br>
> <br>
>

Wed, 07 May 2008 16:06:05 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430843
the cyclist
"Helen Chen" <helen.chen@mathworks.com> wrote in message<br>
<fv82i0$nql$1@fred.mathworks.com>...<br>
> Just a reminder to let everyone know that the Spring Contest<br>
> launches tomorrow, Wednesday May 30th at high noon. Be<br>
> there or be square!<br>
> <br>
> See you then!<br>
> Helen<br>
<br>
Thanks for running another fun contest. Not actually having<br>
access to MATLAB anymore took some fun out of it. 8/ <br>
Didn't stop me from tweaking here and there, though.

Wed, 07 May 2008 16:26:05 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430848
Yi Cao
How about using an antispam tech, such the one used for <br>
File Exchange review comments to prevent automatic <br>
submissions, i.e. a user has to copy certain characters <br>
shown on the screen before he/she can submit a code. This <br>
has been used in File Exchange, hence should not be <br>
difficult for TMW to implement in the contest page.<br>
<br>
Yi

Wed, 07 May 2008 16:49:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430851
David
"the cyclist" <thecyclist@letter.after.f.mail.com> wrote in<br>
message <fvsk1d$m84$1@fred.mathworks.com>...<br>
> "Helen Chen" <helen.chen@mathworks.com> wrote in message<br>
> <fv82i0$nql$1@fred.mathworks.com>...<br>
> > Just a reminder to let everyone know that the Spring Contest<br>
> > launches tomorrow, Wednesday May 30th at high noon. Be<br>
> > there or be square!<br>
> > <br>
> > See you then!<br>
> > Helen<br>
> <br>
> Thanks for running another fun contest. Not actually having<br>
> access to MATLAB anymore took some fun out of it. 8/ <br>
> Didn't stop me from tweaking here and there, though.<br>
<br>
Congratulations to Tim for winning the 1000 node prize, ...<br>
without actually having access to MATLAB !<br>
<br>
 David Jones

Wed, 07 May 2008 17:29:04 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430860
Yi Cao
I strongly agree that using random function will discourage <br>
algorithmic development because a welltuned random code <br>
will significantly disadvantage any new algorithm, which <br>
has not been tuned for the specific test suite. <br>
<br>
As Markus said, using random functions here is total <br>
different from the use in a GA type algorithm, where the <br>
statistical nature of a random function is used to get the <br>
optimal solution. <br>
<br>
However, here, the random function is used just like other <br>
parameters to be tuned specificly for the test suite. By <br>
doing so, when altering an algorithm, the number of random <br>
function calls will change as well, hence the result is <br>
completely unpredictable.<br>
<br>
The main problem with random function is that it <br>
has 'memory', so that previours use of a random function <br>
will affect the follows. As Nathan mentioned, disallowing <br>
random function will not solve the problem because a user <br>
can develop his own. <br>
<br>
However, I would like to suggest to use a random function <br>
in runcontest function to make tuning random parameters <br>
imposible. For example, in runcontest function, a line of <br>
rand('state',sum(100*clock)) can make any intention to tune <br>
a random parameter impossible because even with the same <br>
parameters, the results could be totally different if a <br>
random function is used in a code. <br>
<br>
Yi

Wed, 07 May 2008 17:58:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430869
Steve Hoelzer
"Yi Cao" <y.cao@cranfield.ac.uk> wrote in message<br>
<fvsl6t$5db$1@fred.mathworks.com>...<br>
> How about using an antispam tech, such the one used for <br>
> File Exchange review comments to prevent automatic <br>
> submissions, i.e. a user has to copy certain characters <br>
> shown on the screen before he/she can submit a code. This <br>
> has been used in File Exchange, hence should not be <br>
> difficult for TMW to implement in the contest page.<br>
> <br>
> Yi<br>
<br>
Excellent idea! It wouldn't eliminate tweak bombing, but at<br>
least it would make it harder.

Wed, 07 May 2008 18:01:13 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430870
Steve Hoelzer
"Yi Cao" <y.cao@cranfield.ac.uk> wrote in message<br>
<fvsot0$gr9$1@fred.mathworks.com>...<br>
> However, I would like to suggest to use a random function <br>
> in runcontest function to make tuning random parameters <br>
> imposible. For example, in runcontest function, a line of <br>
> rand('state',sum(100*clock)) can make any intention to tune <br>
> a random parameter impossible because even with the same <br>
> parameters, the results could be totally different if a <br>
> random function is used in a code. <br>
> <br>
> Yi<br>
<br>
No. The same code must give the same score every time it<br>
runs. Otherwise, a new type of random variable "tuning" is<br>
possible: submitting the same code multiple times. I too<br>
wish that magic number tuning was not a part of the contest,<br>
but I don't think it can be programmatically enforced.<br>
<br>
Steve

Wed, 07 May 2008 18:18:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430875
Steve Hoelzer
"Jin " <xx@xx.xx> wrote in message<br>
<fvrj7b$kf6$1@fred.mathworks.com>...<br>
> why bombing the queue? <br>
> I sugguest adding some submission limits, such as 1 <br>
> entry/per min/per ip.<br>
> <br>
<br>
I think that would be an excellent limitation on<br>
submissions. One per minute seems fast enough for tweak<br>
bombers to do what they do, but the spacing should help to<br>
keep the queue shorter (not "overwhelmed").<br>
<br>
Steve

Wed, 07 May 2008 18:34:02 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430878
David
It would be great if the MATLAB Contest Team would let us know their plans for <br>
the end of the contest, ... when the queue is expected to reopen, ... when the <br>
deadline for the end of the contest will be, ... and so on.

Wed, 07 May 2008 18:40:22 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430879
srach
"David " <djones.nospam@remove.mcmaster.ca> wrote in message<br>
<fvssmq$7pp$1@fred.mathworks.com>...<br>
> It would be great if the MATLAB Contest Team would let us<br>
know their plans for <br>
> the end of the contest, ... when the queue is expected to<br>
reopen, ... when the <br>
> deadline for the end of the contest will be, ... and so on.<br>
> <br>
> <br>
<br>
The contest is over. The queue was closed at 12PM and now<br>
the remaining entries will be processed to find a winner. <br>
<br>
Regards,<br>
srach

Wed, 07 May 2008 18:54:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430882
Yi Cao
"Steve Hoelzer" <shoelzer@gmail.com> wrote in message <br>
<fvsqp9$8gg$1@fred.mathworks.com>...<br>
> No. The same code must give the same score every time it<br>
> runs. Otherwise, a new type of random variable "tuning" is<br>
> possible: submitting the same code multiple times. I too<br>
> wish that magic number tuning was not a part of the <br>
contest,<br>
> but I don't think it can be programmatically enforced.<br>
> <br>
> Steve<br>
<br>
However, currently, it is already possible to resubmit the <br>
same code several time to get the best score because the <br>
same code wont get the same result on timeing.<br>
<br>
Yi

Wed, 07 May 2008 19:14:20 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430884
Steve Hoelzer
"Yi Cao" <y.cao@cranfield.ac.uk> wrote in message<br>
<fvstsb$m4n$1@fred.mathworks.com>...<br>
> "Steve Hoelzer" <shoelzer@gmail.com> wrote in message <br>
> <fvsqp9$8gg$1@fred.mathworks.com>...<br>
> > No. The same code must give the same score every time it<br>
> > runs. Otherwise, a new type of random variable "tuning" is<br>
> > possible: submitting the same code multiple times. I too<br>
> > wish that magic number tuning was not a part of the <br>
> contest,<br>
> > but I don't think it can be programmatically enforced.<br>
> > <br>
> > Steve<br>
> <br>
> However, currently, it is already possible to resubmit the <br>
> same code several time to get the best score because the <br>
> same code wont get the same result on timeing.<br>
> <br>
> Yi<br>
<br>
True, but the timing variation is small and has little<br>
impact on score compared to varying the random number seed.<br>
<br>
That said, I would support a change to the scoring function<br>
such that run time is rounded to the nearest second. That<br>
should mostly eliminate the effect of timing variation.<br>
<br>
Steve

Wed, 07 May 2008 20:13:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430898
Nicholas Howe
There are many valid uses of random numbers besides genetic<br>
algorithms. Many quicksort implementations use randomness,<br>
for example. But any such use will leave the temptation to<br>
tweak the state of the random number generator so as to<br>
improve the result.<br>
<br>
I'd like to make another suggestion, in the spirit of<br>
friendly competition. Currently the Mathworks folks provide<br>
recognition to players who win various phases of the<br>
contest. But this doesn't always capture all the wonderful<br>
things that people have contributed to the solution. I'd<br>
like to suggest the idea of players recognizing other<br>
players who have achieved something notable during the<br>
contest, but perhaps didn't win a prize for it. I can't<br>
nominate anyone myself because I didn't follow the contest<br>
in daylight, but perhaps someone else can.

Wed, 07 May 2008 20:19:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430899
Markus Buehren
> How about using an antispam tech, such the one used for <br>
> File Exchange review comments to prevent automatic <br>
> submissions, i.e. a user has to copy certain characters <br>
> shown on the screen before he/she can submit a code. <br>
<br>
Thats exactly what is called a "CAPTCHA" in internet slang :)<br>
<br>
Markus

Wed, 07 May 2008 20:23:04 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430901
Yi Cao
"Steve Hoelzer" <shoelzer@gmail.com> wrote in message <br>
> <br>
> That said, I would support a change to the scoring <br>
function<br>
> such that run time is rounded to the nearest second. That<br>
> should mostly eliminate the effect of timing variation.<br>
> <br>
> Steve<br>
> <br>
<br>
Good point. I would like to support this proposal. It will <br>
discourage users to submit codes with trival or nil <br>
contributions.<br>
<br>
Yi

Wed, 07 May 2008 21:28:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430912
Alfonso NietoCastanon
<br>
on top of all the very good suggestions to keep the contest <br>
engaging and minimally frustrating for everyone I would <br>
also suggest reviving some form of generality prize at the <br>
end of the contest to keep the interest of those developing <br>
nonoverfitting solutions and alternative algorithm <br>
strategies alive<br>
<br>
That said it has been lots of fun, congrats to the matlab <br>
team for their good work!

Wed, 07 May 2008 22:10:05 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430915
Steve Hoelzer
"Alfonso NietoCastanon" <alfnie@gmail.com> wrote in message<br>
<fvt6t3$cpq$1@fred.mathworks.com>...<br>
> <br>
> on top of all the very good suggestions to keep the contest <br>
> engaging and minimally frustrating for everyone I would <br>
> also suggest reviving some form of generality prize at the <br>
> end of the contest to keep the interest of those developing <br>
> nonoverfitting solutions and alternative algorithm <br>
> strategies alive<br>
<br>
I like the idea of a generality prize, but I don't think<br>
it's valid to have one after daylight. There are so many<br>
parameter tweaked entries that one of them is bound to be<br>
overfitted to the generality test set. The darkness stage is<br>
probably as close as you can get to a generality contest.

Wed, 07 May 2008 23:11:04 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430919
Alfonso NietoCastanon
"Steve Hoelzer" <shoelzer@gmail.com> wrote in message <br>
<fvt9bt$ai7$1@fred.mathworks.com>...<br>
> I like the idea of a generality prize, but I don't think<br>
> it's valid to have one after daylight. There are so many<br>
> parameter tweaked entries that one of them is bound to be<br>
> overfitted to the generality test set. The darkness stage <br>
is<br>
> probably as close as you can get to a generality contest.<br>
<br>
I see your point, nevertheless I would think that the <br>
previously used solution of limiting generality submissions <br>
to a subset of the last day entries (renamed to indicate <br>
that they go for the generality prize) should do the job <br>
together with people somehow limiting themselves as to the <br>
number of entries anyone would send for a "generality" <br>
prize without too much shame...<br>
<br>
And of course, while there are no perfect solutions I know <br>
at least for me (I like the algorithm development part more <br>
than the tweeking) this would make an interesting challenge <br>
to work towards (and one day of darkness is so little time <br>
to loose with this...)<br>
<br>
Alfonso

Wed, 07 May 2008 23:40:19 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430924
David
"Steve Hoelzer" <shoelzer@gmail.com> wrote in message <br>
> I like the idea of a generality prize, but I don't think<br>
> it's valid to have one after daylight.<br>
<br>
<br>
I have a proposal to enhance the "generality" of the leading/winning entries <br>
in the MATLAB Contest ...<br>
<br>
The idea is a variation on what is already done with two test suites.<br>
Instead of switching out the test suites in SEQUENCE, ...<br>
I think you could run them in PARALLEL,<br>
by using one large test suite with two subsets (A and B)<br>
that are statistically similar.<br>
(The statistical similarity would be important, ...)<br>
<br>
Then the RESULT when scoring an entry would be<br>
the MAX of RESULT_A and RESULT_B.<br>
<br>
Any tweaking that leads to a better result in subset A<br>
would provide no benefit if there was still a poor result in subset B.<br>
<br>
  <br>
<br>
Here is a generalized version of this idea that I would actually recommend <br>
implementing ... (it uses the majority of the large test suite to assign the final <br>
score, so it might be more robust).<br>
<br>
Let's say we want a test suite containing 256 boards for the Wiring puzzle.<br>
The contest team carefully creates 4 classes of puzzles (maybe 64 of each)<br>
... we might call them: small, medium, large, extralarge,<br>
or they might be created on a continuum from small to extralarge.<br>
Whether they are handcrafted or computergenerated,<br>
there need to be 4 "equally difficult" examples of each "class" of puzzle.<br>
Randomly assign one of 4 colours (Red, Green, Blue, Yellow)<br>
to the boards in each class.<br>
<br>
A good "general" solution should get similar results<br>
for each of the 4 colours. ... An entry with lots of silly tweaks<br>
and overfitting may have a higher variance of results<br>
across the different colours.<br>
<br>
Now we create 4 overlapping subsets of our single test suite:<br>
<br>
A = Red, Green, Blue (but not Yellow) = 48 small, 48 medium, 48 large, 48 <br>
extralarge<br>
B = Red, Green, Yellow (but not Blue) = 48 small, 48 medium, etc.<br>
C = Red, Blue, Yellow (but not Green) = ...<br>
D = Green, Blue, Yellow (but not Red) = ...<br>
<br>
RESULT_A is the sum of the results from Red, Green, and Blue boards.<br>
... and so on.<br>
<br>
The final RESULT used to score an entry is the MAX of RESULT_A, RESULT_B, <br>
RESULT_C, RESULT_D<br>
<br>
General solutions (with low variance) will get full credit for their good "result" <br>
in each of the test subsets. Tweaky overfitting solutions (with high variance) <br>
will get punished according to their worst result in one of the subsets.<br>
<br>
The execution time would still be the time to process the entire test suite.<br>
<br>
I hope this idea makes sense to the Matlab Contest Team. If not, ... I could <br>
try to do a better job of explaining it. I think it would be pretty <br>
straightforward to test it on some entries from the Wiring Contest.<br>
<br>
Who knows, ... they could even use this idea to award a "Generality Prize" for <br>
the Wiring Contest ... perhaps by scoring those entries that were leading the <br>
competition at one time or another ...<br>
<br>
 David Jones

Thu, 08 May 2008 00:27:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430928
Jin
parameter tweaking is allowed, and using random is not a <br>
bad thing as well. But using many submissions to walk <br>
through the solution space is meaningless(because it's <br>
trivial if we have *enough* time). I have only 2 hours/day <br>
to watch the contest. I aslo do some my own codes,but I <br>
have no time to finish them. However, I think to hit 13000 <br>
is very possible. But all are high on using parameter <br>
tweaking(throgh many submissions) to get a special lucky <br>
one in the late contest.

Thu, 08 May 2008 05:22:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430957
Alan Chalker
Congrats on your win Stefan! It looks like tweaking at the<br>
end paid off for you. And thanks to the MATLAB contest team<br>
again for organizing another fun and exciting event.

Thu, 08 May 2008 05:43:02 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430958
Markus Buehren
> Here is a generalized version of this idea that I would<br>
actually recommend <br>
> implementing ... (it uses the majority of the large test<br>
suite to assign the final <br>
> score, so it might be more robust).<br>
> <br>
> Let's say we want a test suite containing 256 boards for<br>
the Wiring puzzle.<br>
> The contest team carefully creates 4 classes of puzzles<br>
(maybe 64 of each) ...<br>
<br>
<br>
Hi David,<br>
<br>
don't you think that when using your suggestion, tweak<br>
bombing would lead to minor steps approaching the optimal<br>
solution for the modified score computation? I think there<br>
would also be an overfitting, in a slightly different manner.<br>
<br>
Markus

Thu, 08 May 2008 05:45:06 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430959
Markus Buehren
> > That said, I would support a change to the scoring <br>
> function<br>
> > such that run time is rounded to the nearest second. That<br>
> > should mostly eliminate the effect of timing variation.<br>
<br>
I also support everything that would take the motivation out<br>
of tweaking. But also with the changed timing, a minor time<br>
improvement could lead to a jump from one quantization<br>
interval to the next...<br>
<br>
Markus

Thu, 08 May 2008 05:51:04 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430960
Markus Buehren
Congratulations to the grand prize, Stefan!<br>
<br>
I am glad the grand prize is going to Germany the second<br>
time in a row! ;)<br>
<br>
Markus

Thu, 08 May 2008 08:16:02 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430972
Yi Cao
"Markus Buehren" <mb_matlab.REMOVE@gmxTHIS.de> wrote in <br>
message <fvu4c8$jkj$1@fred.mathworks.com>...<br>
> Congratulations to the grand prize, Stefan!<br>
> <br>
> I am glad the grand prize is going to Germany the second<br>
> time in a row! ;)<br>
> <br>
> Markus<br>
<br>
Congratulations to Stefan! I am glad to see the final <br>
wnning entry was based on one of my final improvements <br>
(13th generation). <br>
<br>
I noticed you were able to pick up my "new trial" code. <br>
Unfortunately, I was not able to correctly estimate the <br>
impact of the improved speed on score. If I included 4 <br>
different board manipulations, it might win the prize.<br>
<br>
I completely missed the darkness and twilight phases. In <br>
daylight phase, I was concentrated on improving the <br>
bridgepath function because I found it was the bottleneck <br>
for both complexity and speed. This has been proven <br>
productive although myself missed the 1000 node contest to <br>
Tim. :(<br>
<br>
The current winning code still has many rooms to improve. <br>
If we had time, the 13000 goal would be achievable.<br>
<br>
Anyway, it was another great fun to join the contest. <br>
Thanks to TMW team for organizing the contest.<br>
<br>
See you all in six month time.<br>
<br>
Yi

Thu, 08 May 2008 09:10:10 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#430987
Nathan
Thanks again Mathworks!<br>
<br>
Congratulations to Stefan, Tim and David. (Nice solution,<br>
David!) <br>
<br>
Nathan<br>
<br>
"Yi Cao" <y.cao@cranfield.ac.uk> wrote in message<br>
<fvucs2$d4v$1@fred.mathworks.com>...<br>
> "Markus Buehren" <mb_matlab.REMOVE@gmxTHIS.de> wrote in <br>
> message <fvu4c8$jkj$1@fred.mathworks.com>...<br>
> > Congratulations to the grand prize, Stefan!<br>
> > <br>
> > I am glad the grand prize is going to Germany the second<br>
> > time in a row! ;)<br>
> > <br>
> > Markus<br>
> <br>
> Congratulations to Stefan! I am glad to see the final <br>
> wnning entry was based on one of my final improvements <br>
> (13th generation). <br>
> <br>
> I noticed you were able to pick up my "new trial" code. <br>
> Unfortunately, I was not able to correctly estimate the <br>
> impact of the improved speed on score. If I included 4 <br>
> different board manipulations, it might win the prize.<br>
> <br>
> I completely missed the darkness and twilight phases. In <br>
> daylight phase, I was concentrated on improving the <br>
> bridgepath function because I found it was the bottleneck <br>
> for both complexity and speed. This has been proven <br>
> productive although myself missed the 1000 node contest to <br>
> Tim. :(<br>
> <br>
> The current winning code still has many rooms to improve. <br>
> If we had time, the 13000 goal would be achievable.<br>
> <br>
> Anyway, it was another great fun to join the contest. <br>
> Thanks to TMW team for organizing the contest.<br>
> <br>
> See you all in six month time.<br>
> <br>
> Yi

Thu, 08 May 2008 10:47:04 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#431005
srach
"Yi Cao" <y.cao@cranfield.ac.uk> wrote in message <br>
<snip> <br>
<br>
> Congratulations to Stefan! I am glad to see the final <br>
> wnning entry was based on one of my final improvements <br>
> (13th generation). <br>
<br>
Yes, the winning entry was indeed based on your 13th<br>
generation, and I appreciate your contribution, and the<br>
contributions of all others that improved that piece of code<br>
during the contest.<br>
<br>
<br>
> I noticed you were able to pick up my "new trial" code. <br>
> Unfortunately, I was not able to correctly estimate the <br>
> impact of the improved speed on score. If I included 4 <br>
> different board manipulations, it might win the prize.<br>
<br>
No, did not pick up your "new trial" code. (You introduced<br>
the board manipulations in "new trial 1" which was submitted<br>
11:55:46; the winning entry "tweaky bird and pushy cat 21"<br>
was already submitted 11:55:18.)<br>
<br>
Actually I came across the idea to trade some speed for<br>
accuracy already during the 1000 nodes contest, where<br>
basically the same modification led to about 4000 pts.<br>
improvement in results, while worsening the time by about 20<br>
ms (entry "....until you can call him a man.").<br>
<br>
For the final contest, I prepared lots of browser tabs and<br>
picked entries by random to apply this manipulation (and<br>
some different ones that were not quite successful). Then I<br>
waited the deadline to approach to submit all those entries.<br>
Unfortunately, my poker face broke to early and, thus, I<br>
started submitting to early. I was done with submitting<br>
already 11.56, leaving lots of time for others to pick up<br>
entries (see Alan's win in the best results.<br>
Congratulations, Alan!). Finally, the winning entry improved<br>
the score of "13th generation" by about 100 pts.<br>
<br>
Big thanks to the team at matlab central for organizing<br>
another great contest and all the participants for an<br>
entertaining week. <br>
<br>
See you all in Fall...<br>
<br>
Regards,<br>
Stefan aka srach

Thu, 08 May 2008 12:30:42 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#431026
Yi Cao
"srach " <s.rach_NOSPAM_@jacobsuniversity.de> wrote in <br>
message <fvuln8$m57$1@fred.mathworks.com>...<br>
> <br>
> No, did not pick up your "new trial" code. (You introduced<br>
> the board manipulations in "new trial 1" which was <br>
submitted<br>
> 11:55:46; the winning entry "tweaky bird and pushy cat 21"<br>
> was already submitted 11:55:18.)<br>
> <br>
<br>
No, I did not mean the winning entry but the entry <br>
of "tweaky bird and pushy cat 26", which was submitted at <br>
11:57:36. The "new trial" code combines all 4 search paths <br>
in "bridgepath_a" into a for loop. It becomes faster with <br>
reduced complexity to 18, but the original tuning <br>
parameters make the results worse. To fully use the <br>
improvement in speed, we have to inlcude more board <br>
manipulations. Observing all results it produced, I <br>
estimate we can have 4 board manipulations, which should <br>
significantly reduce the results.<br>
<br>
Yi

Thu, 08 May 2008 12:52:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#431032
srach
"Yi Cao" <y.cao@cranfield.ac.uk> wrote in message<br>
<fvurpi$2an$1@fred.mathworks.com>...<br>
> "srach " <s.rach_NOSPAM_@jacobsuniversity.de> wrote in <br>
> message <fvuln8$m57$1@fred.mathworks.com>...<br>
> > <br>
> > No, did not pick up your "new trial" code. (You introduced<br>
> > the board manipulations in "new trial 1" which was <br>
> submitted<br>
> > 11:55:46; the winning entry "tweaky bird and pushy cat 21"<br>
> > was already submitted 11:55:18.)<br>
> > <br>
> <br>
> No, I did not mean the winning entry but the entry <br>
> of "tweaky bird and pushy cat 26", which was submitted at <br>
> 11:57:36. The "new trial" code combines all 4 search paths <br>
> in "bridgepath_a" into a for loop. It becomes faster with <br>
> reduced complexity to 18, but the original tuning <br>
> parameters make the results worse. To fully use the <br>
> improvement in speed, we have to inlcude more board <br>
> manipulations. Observing all results it produced, I <br>
> estimate we can have 4 board manipulations, which should <br>
> significantly reduce the results.<br>
> <br>
> Yi<br>
<br>
You are, of course, correct; I did get you wrong. <br>
<br>
Regards,<br>
Stefan

Thu, 08 May 2008 16:32:04 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#431091
Helen Chen
"srach " <s.rach_NOSPAM_@jacobsuniversity.de> wrote in<br>
message <fvuln8$m57$1@fred.mathworks.com>...<br>
> For the final contest, I prepared lots of browser tabs and<br>
> picked entries by random to apply this manipulation (and<br>
> some different ones that were not quite successful). Then I<br>
> waited the deadline to approach to submit all those entries.<br>
<br>
Interesting approach, Stefan. This sounds like an overview<br>
of your test strategy would be a very interesting YouTube<br>
video for others to start planning their strategy for the<br>
Fall Contest. <br>
<br>
> Unfortunately, my poker face broke to early and, thus, I<br>
> started submitting to early. <br>
:)<br>
<br>
> Big thanks to the team at matlab central for organizing<br>
> another great contest and all the participants for an<br>
> entertaining week. <br>
> <br>
<br>
It was our pleasure! Thank you to everyone who<br>
participated. It was your actions, both with submissions and<br>
in discussion, that make our contests such fun. <br>
<br>
Helen

Thu, 08 May 2008 20:08:04 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#431136
OkinawaDolphin
The function runcontest displayed the overall results and<br>
time. However, statistics such as minimum, mean, variance<br>
and maximum are much more helpful for evaluating the<br>
performance of the solver than a sum only. Is it possible<br>
that the runcontest function of the next contest includes<br>
performance statistics?

Thu, 08 May 2008 23:05:07 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#431165
Fabio
Hi all,<br>
congratulations to the winners!<br>
It has been my first partecipation, it has been a very<br>
interesting experience, the problem was very interesting too<br>
and very funny!!<br>
I'm also glad that in the winning code there is a my<br>
(little) improvement (and vainly I obfuscate it :) ).<br>
<br>
The main speedup was to avoid to put the bridgepath function<br>
in the double nested loop in phase2. It was better to call<br>
only once a function newBridgepath (aka d4TZerDbiG in the<br>
winning entry) that finds a path for all the sources pins<br>
and all the targets concurrently. Initially the global<br>
speedup was of about 60% in time!!<br>
Unfortunately the change produced a function with a<br>
different behaviour and so it was impossible to apply the<br>
speedup to an "overtuned" entry and to get the same score again.<br>
<br>
An other speedup regards the functions phase1 and phase3: in<br>
such functions it's useless to call the function complexpath<br>
two times with the same value of b, so many computations<br>
could be saved, with an improvement of about 30% in time.<br>
Unfortunately I didn't succeed in using the saved time to<br>
win the contest. :(<br>
<br>
Regarding the suggestions about the rules of the contest, I<br>
think that:<br>
<br>
1) As many people say, randomness is unavoidable.<br>
2) Why don't hide the score and the code of an entry for the<br>
last 60 (or 30) minutes before a deadline? So it won't be<br>
necessary to submit a winningcandidate entry only 3 minutes<br>
before the end of the contest (with many risks!).<br>
3) To avoid the tweak bombing (I also submitted about 100<br>
entries in the last 5 minutes of the contest) it would be<br>
useful to use very large test suites, i.e. 2000 puzzles<br>
instead of 200, so the effect of randomness will be<br>
negligible. Moreover the efficiency will be crucial.<br>
4) (stupid idea) Why don't you organize also a "shortlife<br>
contest"? It's very difficult to find the time to follow the<br>
contest for a whole week. It would be very convenient to<br>
partecipate to a contest only during the weekend.<br>
<br>
A request:<br>
I'd ask the staff what is the initialization of the random<br>
generator. The behaviour of the random generator is very<br>
important to succeed in overfitting the puzzles and I want<br>
to see on my computer the score of some scripts that I<br>
mistakenly didn't submit. :((<br>
Fortunately these scripts are losing versus "tweaky bird and<br>
pushy cat 21" with many random seeds :)<br>
<br>
Thank you!!<br>
Goodbye and sorry for my bad english!!!

Fri, 09 May 2008 07:05:08 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#431201
OkinawaDolphin
<br>
> 4) (stupid idea) Why don't you organize also a "shortlife<br>
> contest"? It's very difficult to find the time to follow <br>
the<br>
> contest for a whole week. It would be very convenient to<br>
> partecipate to a contest only during the weekend.<br>
<br>
I suggest the opposite: The contest should last three or <br>
for weeks. The darkness and twilight should last a week <br>
respectively. Why?<br>
<br>
1. When daylight begins, many participants focus on <br>
tweaking. If you try to develop and optimize an algorithm <br>
during this time, the results are worse than average.<br>
<br>
2. The best programs were also the longest and most <br>
complicated. If functions are called "sunday" or "tweak", <br>
it is difficult to figure out what they mean. So tweaking <br>
is mainly done by parameter tuning instead of algorithm <br>
optimization.<br>
<br>
3. People who can work on their entries during the weekend <br>
only have also a chance to take part in the contest.<br>
<br>
By the way, I think that the explanation of the rules is <br>
misleading somehow. When pin values are one digit numbers, <br>
connectors are expensive compared to the pins.<br>
<br>
However, when the pin values are two digit numbers on a 15 <br>
* 17 board, they are actually cheap. Under this condition, <br>
connecting isolated islands is not a waste of time and <br>
material. The number of points saved exceeds the number of <br>
connectors often. <br>
<br>
By connecting distant islands, the board can get fragmented <br>
and it gets necessary to build bridges or to leave many <br>
pins unconnected. But this means that isolated island <br>
themselves should be avoided, not connecting them.

Fri, 09 May 2008 07:33:05 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#431204
Yi Cao
"OkinawaDolphin " <OkinawaDolphin@Hotmail.com> wrote in <br>
message <g00t34$kj5$1@fred.mathworks.com>...<br>
> <br>
> 2. The best programs were also the longest and most <br>
> complicated. If functions are called "sunday" or "tweak", <br>
> it is difficult to figure out what they mean. So tweaking <br>
> is mainly done by parameter tuning instead of algorithm <br>
> optimization.<br>
> <br>
<br>
The subday function name was given by me. I used it in my <br>
local computer to identify different version of code <br>
because it was developed on sunday. Basically, it is a <br>
solver. Since it was the first entry when it was submitted, <br>
it has been copied into allmost all entries. :)<br>
<br>
Yi

Fri, 09 May 2008 22:34:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#431307
Sergey
Congratulations to Stefan and all participants.<br>
<br>
Many thanks to organizers of great contest.<br>
<br>
SY (Sergey)<br>
<br>
P.S. Sadly:<br>
<br>
Any new even mildly new code modification has to be tweaked <br>
somehow to be competitive with previous highly tweaked <br>
code. The author has two choices: submit numerous <br>
modifications with “all possible” parameters or try to <br>
adjust parameters looking on code score. Second method has <br>
several significant disadvantages: somebody else may pick <br>
up code and find best parameter combination faster, author <br>
may not have enough time before deadline, queue may be <br>
stalled.<br>
So, here is my conclusion: I am writing my out “tweaking <br>
machine gun” and will use it next time if captcha is not <br>
implemented.

Sat, 10 May 2008 03:57:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#431319
Ned Gulley
"Sergey " <ivssnn@yahoo.com> wrote in message <br>
> Any new even mildly new code modification has to be tweaked <br>
> somehow to be competitive with previous highly tweaked <br>
> code. <br>
<br>
Hi Sergey:<br>
<br>
You are right. Introducing a new idea to compete with<br>
hypertweaked code is very difficult. We'd like to make it<br>
easier. Can you think of things we might do to the contest<br>
machinery to make it easier to introduce new algorithmic<br>
ideas? Based on all the feedback so far, we will probably<br>
introduce some kind of rate limiter (whether by CAPTCHA or<br>
otherwise). What are some other ideas? I've often wondered<br>
if people might flag new ideas explicitly with a message<br>
like "please help me tweak a new concept into first place."<br>
We tried a late stage twilight last time, but it didn't seem<br>
to help, and it wasn't much fun.<br>
<br>
Ned.<br>
Contest Team

Sat, 10 May 2008 06:35:04 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#431323
Yi Cao
"Ned Gulley" <gulley@mathworks.com> wrote in message > <br>
> Hi Sergey:<br>
> <br>
> You are right. Introducing a new idea to compete with<br>
> hypertweaked code is very difficult. We'd like to make it<br>
> easier. Can you think of things we might do to the contest<br>
> machinery to make it easier to introduce new algorithmic<br>
> ideas? Based on all the feedback so far, we will probably<br>
> introduce some kind of rate limiter (whether by CAPTCHA or<br>
> otherwise). What are some other ideas? I've often wondered<br>
> if people might flag new ideas explicitly with a message<br>
> like "please help me tweak a new concept into first <br>
place."<br>
> We tried a late stage twilight last time, but it didn't <br>
seem<br>
> to help, and it wasn't much fun.<br>
> <br>
> Ned.<br>
> Contest Team<br>
<br>
I think to encourage new idea, we have to introduce some <br>
generality index to be included in the overall score. When <br>
a new algorithm is submitted, even it may not be able to <br>
compete with current best code, but if it shows some <br>
advantage in generality, it might be pick up by someone <br>
else to tweak it into top. The generality index can come <br>
from another hidden testsuit or simply an average results <br>
by dropping different portion of cases in the testsuite <br>
(something like cross validation).<br>
<br>
Another suggestion, can we make the last 30 minutes or so <br>
as the twilight phase, or disable downloading an entry from <br>
the waiting queue? This will encourage people to submit <br>
their final code earlier.<br>
<br>
The node index and 1000 node contest were good. But, it <br>
seems that the weight of the node index was too small, no <br>
one cared about it. <br>
<br>
Yi

Sat, 10 May 2008 14:00:22 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#431355
Markus Buehren
> Based on all the feedback so far, we will probably<br>
> introduce some kind of rate limiter (whether by CAPTCHA or<br>
> otherwise).<br>
<br>
I am really glad to hear that :) A captcha will take most<br>
of the fun out of the annoying tweakbombing.<br>
<br>
> I've often wondered<br>
> if people might flag new ideas explicitly with a message<br>
> like "please help me tweak a new concept into first <br>
> place."<br>
<br>
Hmm, but in this way the contestant who had the idea will<br>
probably not be rewarded with a contest prize... <br>
<br>
> else to tweak it into top. The generality index can come <br>
> from another hidden testsuit or simply an average results <br>
> by dropping different portion of cases in the testsuite <br>
> (something like cross validation).<br>
<br>
Good idea, but as every submission must be rated under the<br>
same conditions, finally we will again have only one rating<br>
function outputting one score value. Entries can be tweaked<br>
against this score...<br>
<br>
> Another suggestion, can we make the last 30 minutes or so <br>
> as the twilight phase, or disable downloading an entry from <br>
> the waiting queue? This will encourage people to submit <br>
> their final code earlier.<br>
<br>
Good idea! Code in the queue should generally be hidden, and<br>
also all code submitted in the last hour or 30 minutes<br>
before a phase deadline, even if it was already processed.<br>
<br>
<br>
Further, I would still like to have some extension to the<br>
twilight phase. However, if someone does not have the time<br>
to develop his own code, he would not get an entry into the<br>
contest. So this is my suggestion for a compromise:<br>
<br>
The submissions of the ten best contestants (not ten best<br>
submissions!) after twilight should be published after 48<br>
hours. All other submitted code, as well as all new<br>
submissions during the "twilight 2" phase should be hidden<br>
until this phase ends. This way everyone can see some code<br>
examples, optimize them or build his own best combination of<br>
those submissions. <br>
<br>
Only drawback of this suggestion might be that people could<br>
submit obfuscated code to hide their ideas. However, I<br>
personally would be very proud to see my own code survive<br>
until the end of the contest, so I would not start<br>
obfuscating already during the first contest hours.<br>
<br>
<br>
Finally: Any idea about how to avoid obfuscation???<br>
<br>
<br>
Yours<br>
Markus

Sat, 10 May 2008 14:40:20 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#431358
Abhisek Ukil
First of all, hats off to Matlab team for organizing <br>
another exciting contest. and few ideas towards supporting <br>
generality:<br>
<br>
1. <br>
> Further, I would still like to have some extension to the<br>
> twilight phase. <br>
<br>
I agree with Markus, (don't know if you meant like this): <br>
but I would like to have the twilight extended at least <br>
half a day more.<br>
<br>
2. make the 1 Hr before the deadline time of say best <br>
result before 6,1000 node challenge, final day etc, <br>
TWILIGHT (no code download but score visible).<br>
<br>
3. A lot of tweaking is aimed at improving the best score <br>
by reducing computation time. In current format, a time <br>
improvement of the order of 0.0001 is enough to have new <br>
score improvement with same result. Same result possibly <br>
indicates no improvement from algorithm side. This could <br>
be changed after say Saturday push. reduce this order <br>
sequentially, Saturday 0.0001, Sunday 0.1 ... On sunday <br>
one can also round off the time to nearest integer. So, to <br>
improve score one has to really work on the algorithm to <br>
improve the "result". change the scoring coefficients <br>
after sunday to lay more weight on result than time.<br>
<br>
4. I like the idea of the "generality index" into the <br>
scoring equation. And around Sunday, change focus from <br>
just minimizing the score to somehow include the <br>
generality condition as well. Such that for an overfitted <br>
solution which exceeds the maximum "generality index" the <br>
score would be penalized evenif it reduces result or <br>
improves time on current set. Counting usage of random <br>
functions (if possible) might help as no. of use of random <br>
functions should be inversely related to the generality.<br>
<br>
Anyway just a few thoughts (maybe already quite a few <br>
lines), but open to debate and grow further ....<br>
<br>
BR,<br>
Abhisek

Sat, 10 May 2008 14:43:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#431359
Abhisek Ukil
First of all, hats off to Matlab team for organizing <br>
another exciting contest. and few ideas towards supporting <br>
generality:<br>
<br>
1. <br>
> Further, I would still like to have some extension to the<br>
> twilight phase. <br>
<br>
I agree with Markus, (don't know if you meant like this): <br>
but I would like to have the twilight extended at least <br>
half a day more.<br>
<br>
2. make the 1 Hr before the deadline time of say best <br>
result before 6,1000 node challenge, final day etc, <br>
TWILIGHT (no code download but score visible).<br>
<br>
3. A lot of tweaking is aimed at improving the best score <br>
by reducing computation time. In current format, a time <br>
improvement of the order of 0.0001 is enough to have new <br>
score improvement with same result. Same result possibly <br>
indicates no improvement from algorithm side. This could <br>
be changed after say Saturday push. reduce this order <br>
sequentially, Saturday 0.0001, Sunday 0.1 ... On sunday <br>
one can also round off the time to nearest integer. So, to <br>
improve score one has to really work on the algorithm to <br>
improve the "result". change the scoring coefficients <br>
after sunday to lay more weight on result than time.<br>
<br>
4. I like the idea of the "generality index" into the <br>
scoring equation. And around Sunday, change focus from <br>
just minimizing the score to somehow include the <br>
generality condition as well. Such that for an overfitted <br>
solution which exceeds the maximum "generality index" the <br>
score would be penalized evenif it reduces result or <br>
improves time on current set. Counting usage of random <br>
functions (if possible) might help as no. of use of random <br>
functions should be inversely related to the generality.<br>
<br>
Anyway just a few thoughts (maybe already quite a few <br>
lines), but open to debate and grow further ....<br>
<br>
BR,<br>
Abhisek

Sat, 10 May 2008 16:20:20 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#431369
Sergey
Seem to me, we are coming to some consensus. I would like <br>
to see last one hour before all deadlines to be twilight <br>
too. It will give some time for new code to be adjusted and <br>
make sure that the real author of good modification will <br>
get credit. However it does not solve completely problem of <br>
automatic submissions because they still may clog queue and <br>
deny everybody feedback from test, making it essentially <br>
dark.<br>
<br>
There were a lot of discussions about test set overfitting. <br>
I do not think that any suggested method of using subsets <br>
will really work; it will just change test set (or weighted <br>
combination of sets) to fit. <br>
Another idea is to make “generality” prize the main prize <br>
with limited number of submission from each participant. <br>
<br>
All this may be (?) less important with different kind of <br>
contest problem. I would say that Blackbox contest having <br>
its own problem (probing) was not under the danger of <br>
overfitting and unnecessary parameter tweaking.

Sat, 10 May 2008 17:49:04 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#431371
Alan Chalker
"Ned Gulley" <gulley@mathworks.com> wrote in message<br>
<g036ef$i42$1@fred.mathworks.com>...<br>
<br>
> <br>
> Hi Sergey:<br>
> <br>
> You are right. Introducing a new idea to compete with<br>
> hypertweaked code is very difficult. We'd like to make it<br>
> easier. Can you think of things we might do to the contest<br>
> machinery to make it easier to introduce new algorithmic<br>
> ideas? Based on all the feedback so far, we will probably<br>
> introduce some kind of rate limiter (whether by CAPTCHA or<br>
> otherwise). What are some other ideas? I've often wondered<br>
> if people might flag new ideas explicitly with a message<br>
> like "please help me tweak a new concept into first place."<br>
> We tried a late stage twilight last time, but it didn't seem<br>
> to help, and it wasn't much fun.<br>
> <br>
> Ned.<br>
> Contest Team<br>
<br>
<br>
Ned:<br>
<br>
I'd like to put forth a couple comments, suggestions:<br>
<br>
The 'latestage twilight' you tried previously wasn't<br>
really what everyone has been suggesting. I believe you did<br>
it right after the testsuite swap, whereas I believe<br>
everyone has more interest in it occurring at the VERY END,<br>
of the contest.<br>
<br>
I'm glad to hear you are seriously considering implementing<br>
a ratelimiter on the entries submissions. The main contest<br>
community has been asking for this for a long time and I<br>
guess my tweak bombing was the straw that finally broke the<br>
camel's back. I'm not sure what the reluctance has been in<br>
the past from the organizers to implement it, but I suspect<br>
it's just that there is a certain amount of technical effort<br>
you don't have time to implement. Perhaps you could look at<br>
the reCAPTCHA project<br>
(<a href="http://recaptcha.net/whyrecaptcha.html">http://recaptcha.net/whyrecaptcha.html</a>). It's a free<br>
implementation and general easy to utilize.<br>
<br>
I like the suggestion someone else made on having the queue<br>
entries not be visible until they are run. This, combined<br>
with a ratelimiter should significantly reduce the<br>
'freefor alls' we see at the contest deadlines.<br>
<br>
Another way to potentially test for generality is to have a<br>
second test suite that the codes are only run against maybe<br>
once a day. The testsuite swap doesn't seem to have<br>
reduced the tweaking too much, albeit caused just the<br>
opposite. I think the way to test for generality is to<br>
reduce, but not eliminate, the feedback you give us on the<br>
solvers against a secondary test suite. One suggestion would<br>
be to run at midnight each night the top x <br>
(50? 100?) solvers against an alternate testsuite and<br>
provide a 'generality' score. This would prevent someone<br>
from trying to 'bomb the queue' with a ton of entries, since<br>
only a handful would be checked for generality, and would<br>
add a new element for dedicated competitors where they have<br>
to closely examine the leading 'generality' solvers to<br>
understand why they are so much better. You might even mix<br>
up the alternate test suite each night.<br>
<br>
I really liked the new node challenge that replaced the<br>
character challenge. It virtually eliminated the<br>
obscufation that always occurs around that time of the contest.<br>
<br>
I missed having the midcontest analysis Lucio normally<br>
does. I always find that a very insightful part of the<br>
competition and while I understand it takes time, I hope you<br>
find a way to include it in future contest again. Perhaps<br>
you could outsource some of it to contest competitors like<br>
myself that are interested in contributing to the<br>
'metacontest'?<br>
<br>
I like the idea of randomly seeding the rand function<br>
within the runcontest routine for each time it runs (i.e.<br>
rand('twister', time*100)). That would definitely reduce<br>
the 'overfitting' we always see occuring. It wouldn't<br>
eliminate the 'tweak bombing', but some of these other<br>
suggestions would help with that.<br>
<br>
I also would like to echo the idea of reducing the time<br>
dependence on the score. One way to do that is to pick the<br>
scoring coefficients more appropriately to 'devalue' minor<br>
timing variations. Another way would be to just round the<br>
times to the nearest tenth second.<br>
<br>
I also like how you have provided a nice variety on the<br>
midgame contests. Sometimes you announce them way in<br>
advance and give lots of time to work on them, other times<br>
you 'spring them on us' and only provide a few hours. The<br>
varied end times also allow for competitors virtually<br>
anywhere in the world to be able to participated at least once.<br>
<br>
Thanks again for hosting another fun contest! I look<br>
forward to the fall version.

Sat, 10 May 2008 19:48:04 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#431380
srach
I do also second the idea of returning to twilight for the<br>
last hour of the contest. In addition, I would suggest to<br>
switch to a new testsuite for this last hour. This would<br>
give algorithmic improvements a better chance against highly<br>
tweaked code. However, one should abandon the idea that this<br>
last hour in twilight will leave time to optimize any code<br>
to the new testsuite, because the queue will fill quite<br>
quickly, I think. <br>
<br>
Regarding the tweak bombing, it is not "them" that are<br>
spamming, it's us. Thus, all we should need is a kind of<br>
agreement how many entries per person are considered normal<br>
participation, and at which number of entries spamming<br>
starts. Thus, if we agree on only submitting only about 100,<br>
200, or 400 entries, we don't need a CAPTCHA (still I would<br>
welcome one). Of course, one can still use different aliases<br>
for submissions, but we all are free not to do so.<br>
Nevertheless, the queue will still get pretty crowded close<br>
to deadlines. <br>
<br>
Same is true for obfuscating, we should simply don't do it.<br>
My impression of the current contest was, that obfuscation<br>
was less frequent compared to earlier contests, although it<br>
appeared to result in slightly faster runtimes. <br>
<br>
<br>
In this contest, I liked the 1000 node contest; it is much<br>
better than the 1000 character contest, because there is no<br>
need anymore to obfuscate in order to reach the character<br>
limit. <br>
<br>
Maybe it should be considered to extend the first twilight a<br>
little bit to allow more time for the development of<br>
different algorithms. <br>
<br>
Regards,<br>
Stefan aka srach

Sun, 11 May 2008 07:22:04 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#431428
Yi Cao
"Steve Hoelzer" <shoelzer@gmail.com> wrote in message <br>
<fvsqp9$8gg$1@fred.mathworks.com>...<br>
> "Yi Cao" <y.cao@cranfield.ac.uk> wrote in message<br>
> <fvsot0$gr9$1@fred.mathworks.com>...<br>
> > However, I would like to suggest to use a random <br>
function <br>
> > in runcontest function to make tuning random parameters <br>
> > imposible. For example, in runcontest function, a line <br>
of <br>
> > rand('state',sum(100*clock)) can make any intention to <br>
tune <br>
> > a random parameter impossible because even with the <br>
same <br>
> > parameters, the results could be totally different if a <br>
> > random function is used in a code. <br>
> > <br>
> > Yi<br>
> <br>
> No. The same code must give the same score every time it<br>
> runs. Otherwise, a new type of random variable "tuning" is<br>
> possible: submitting the same code multiple times. I too<br>
> wish that magic number tuning was not a part of the <br>
contest,<br>
> but I don't think it can be programmatically enforced.<br>
> <br>
> Steve<br>
<br>
If most of us agree to discourage using random functions, <br>
then including a random seeds in runcontest is a solution. <br>
It wont affect entries which do not use any random function <br>
but will make results unpredictable if a random function is <br>
used. It is against "The same code must give the same score <br>
every time it runs". However, this is what the <br>
user "deserved" because he/she makes some decision <br>
randomly. OK, they can resubmit the same code several <br>
times, but seldom one of them can get to the top because <br>
they cannot tune parameters systematically. <br>
<br>
I support the idea to let the final stage of contests a <br>
twilight phase. It is also logical, a contest starting from <br>
darkness, then twilight, then full daylight, finally <br>
twilight again because it is going to finish. It does not <br>
need to be a full day, but 4 hours will be enough. If <br>
possible, I would like only those submitted during the last <br>
4 hours to be protected from download but not old entries. <br>
This will help us to search the complete list to find a <br>
good code to work with during the final stage.<br>
<br>
Yi

Mon, 12 May 2008 08:08:02 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#431512
Markus Buehren
Hi all,<br>
<br>
I post one idea here again in the hope for some comments.<br>
The idea was about an extension of the twilight phase:<br>
<br>
The submissions of the ten best contestants (not ten best<br>
submissions!) after twilight should be published after 48<br>
hours. All other submitted code, as well as all new<br>
submissions during the "twilight 2" phase (like another 24<br>
hours) should be hidden until this phase ends. This way<br>
everyone can see some code examples, optimize them or build<br>
his own best combination of those submissions. <br>
<br>
<br>
And another thing: I guess many contestants have more time<br>
for the contest in the weekend compared to weekdays.<br>
However, in the weekend most of the times there are no<br>
challenges ending where one could win a prize. Maybe the<br>
contest team could announce one or two challenges also for<br>
saturday and sunday before going into their deserved<br>
weekend? If a rate limiter like a captcha is implemented,<br>
hopefully there will be no queue or statistics crashes<br>
during that time...<br>
<br>
Markus

Mon, 12 May 2008 14:34:00 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#431569
Steve Hoelzer
Contest team, I haven't said it yet, so thanks for running<br>
another great contest! I'm already looking forward to the<br>
next with some of the changes that have been mentioned here.<br>
<br>
Speaking of those changes, here's my take on many of items<br>
that have been brought up...<br>
<br>
Ideas I like:<br>
1. Rate limiter.<br>
2. Longer twilight.<br>
3. 1000 nodes instead of 1000 characters. (Although<br>
maybe 1000 is too large. Maybe a 500 node challenge?)<br>
4. Can't view entries until they are scored.<br>
5. Midcontest and endofcontest analyses. I know it is<br>
a lot of work, but I always find them interesting and<br>
sometimes helpful in developing my own code.<br>
<br>
Ideas I'm not sure about:<br>
1. One hour twilight before all contest deadlines. This<br>
may cause people to only submit code right before deadlines,<br>
which would make the rest of daylight pretty boring. If<br>
twilight is only added at the very end of the contest, then<br>
the phases become darkness, dawn, daylight, and dusk.<br>
2. Random rand seed. Random numbers *can* have a place<br>
in legitimate entries, and I feel pretty strongly that an<br>
entry should give the same score every time it runs.<br>
However, this would eliminate one type of magic number tuning.<br>
3. Quantize run time in the scoring equation. This<br>
minimizes the impact of tiny speed tweaks and run time<br>
variation, but quantization boundaries will always be a problem.<br>
<br>
Ideas I'm against:<br>
1. Some other method of computing a generality score. No<br>
matter how you do it, tweaking can optimize entries for the<br>
new score computation.<br>
2. Reduce the weight of time in the score. It's more<br>
interesting to balance time vs. performance, rather than<br>
just go for performance.

Mon, 12 May 2008 14:34:02 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#431570
Steve Hoelzer
Contest team, I haven't said it yet, so thanks for running<br>
another great contest! I'm already looking forward to the<br>
next with some of the changes that have been mentioned here.<br>
<br>
Speaking of those changes, here's my take on many of items<br>
that have been brought up...<br>
<br>
Ideas I like:<br>
1. Rate limiter.<br>
2. Longer twilight.<br>
3. 1000 nodes instead of 1000 characters. (Although<br>
maybe 1000 is too large. Maybe a 500 node challenge?)<br>
4. Can't view entries until they are scored.<br>
5. Midcontest and endofcontest analyses. I know it is<br>
a lot of work, but I always find them interesting and<br>
sometimes helpful in developing my own code.<br>
<br>
Ideas I'm not sure about:<br>
1. One hour twilight before all contest deadlines. This<br>
may cause people to only submit code right before deadlines,<br>
which would make the rest of daylight pretty boring. If<br>
twilight is only added at the very end of the contest, then<br>
the phases become darkness, dawn, daylight, and dusk.<br>
2. Random rand seed. Random numbers *can* have a place<br>
in legitimate entries, and I feel pretty strongly that an<br>
entry should give the same score every time it runs.<br>
However, this would eliminate one type of magic number tuning.<br>
3. Quantize run time in the scoring equation. This<br>
minimizes the impact of tiny speed tweaks and run time<br>
variation, but quantization boundaries will always be a problem.<br>
<br>
Ideas I'm against:<br>
1. Some other method of computing a generality score. No<br>
matter how you do it, tweaking can optimize entries for the<br>
new score computation.<br>
2. Reduce the weight of time in the score. It's more<br>
interesting to balance time vs. performance, rather than<br>
just go for performance.

Tue, 13 May 2008 18:53:03 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#431878
Fabio
Hi all,<br>
<br>
I want to express an idea about the weight of time in the score.<br>
The quantization of time in the score can suffer of<br>
boundaries problem, as Steve Hoelzer wrote. A solution can<br>
be the following:<br>
<br>
1) If an entry obtains the *same* result of a previous entry<br>
and its CPU time is better than the one of the previous<br>
entry by less than 1s (or 0.XXXs), the score "collapses" to<br>
the one of the first entry (+ epsilon). In example, assume that:<br>
<br>
Score = Time + Result<br>
<br>
the following results will hold:<br>
<br>
Result 10, Time 10.5, Score: 20.5<br>
Result 10, Time 10.4, Score 20.5 + eps<br>
Result 10, Time 9.501, Score 20.5 + eps<br>
Result 10, Time 9.49, Score 19.49<br>
Result 9.9, Time 9.4, Score 19.3<br>
Result 10, Time 8.491, Score 19.49 + eps<br>
Result 10, Time 8.3, Score 18.3<br>
<br>
Thus small and accidental cpu time improvements will be ignored.<br>
<br>
Fabio

Wed, 14 May 2008 23:40:20 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#432101
Alan Chalker
Is it just me or is the contest page missing now? <br>
<a href="http://www.mathworks.com/contest/wiring/home.html">http://www.mathworks.com/contest/wiring/home.html</a><br>
redirects to a 'page not found'....

Thu, 15 May 2008 14:11:55 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#432256
Steven Lord
<br>
"Alan Chalker" <alancNOSPAM@osc.edu> wrote in message <br>
news:g0ft93$plv$1@fred.mathworks.com...<br>
> Is it just me or is the contest page missing now?<br>
> <a href="http://www.mathworks.com/contest/wiring/home.html">http://www.mathworks.com/contest/wiring/home.html</a><br>
> redirects to a 'page not found'....<br>
<br>
I'm not sure what the status was last night, but it seems to be up now.<br>
<br>
 <br>
Steve Lord<br>
slord@mathworks.com

Thu, 15 May 2008 18:34:01 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#432327
David
I want to add a few comments to what Steve Hoelzer has<br>
already said about the suggested changes for the Matlab<br>
contest ...<br>
<br>
"Steve Hoelzer" <shoelzer@gmail.com> wrote in message<br>
<g09kgq$ldb$1@fred.mathworks.com>...<br>
<br>
> Ideas (Steve) liked:<br>
><br>
> 4. Can't view entries until they are scored.<br>
<br>
I would suggest a variation on this idea.<br>
<br>
During most of the Daylight phase, allow entries in the<br>
queue to be viewed (even before they are scored). I think<br>
this adds to the interest and excitement by allowing all<br>
competitors to learn what various tricks people are trying,<br>
and most of the time there is no benefit for waiting until<br>
the entries are scored. ... and I wouldn't like to see any<br>
incentive for people clogging the queue, just to delay<br>
others from seeing their notyetscored entry further down<br>
the queue.<br>
<br>
During the final hour of the contest, go into a kind of<br>
"Late Twilight" where entries can't be viewed until they are<br>
scored. This will enable competitors to submit what they<br>
think might be a winning improvement, without needing to<br>
wait until the last 5 minutes, and without needing to submit<br>
100 variations of essentially the same thing.<br>
<br>
I also think many people have real life constraints on their<br>
schedules (at work or home, depending on time zone) that<br>
makes it difficult for them to be at their computer<br>
precisely moments before the deadline to trigger some<br>
specialized software to submit 500 entries at a rate of 10<br>
per second, or to click submit on several dozen Firefox<br>
browser tabs they set up in advance ...<br>
<br>
<br>
<br>
<br>
> 5. Midcontest and endofcontest analyses.<br>
<br>
<br>
I agree these are really valuable and worth the effort. <br>
They help competitors "catch up" and understand how the key<br>
algorithms form Darkness/Twilight are working, ... thus<br>
increasing more effective participation in the later stages<br>
of the contest. They also have lasting educational value<br>
for students who want to how to use Matlab to solve problems<br>
similar to the Contest puzzles.<br>
<br>
> <br>
> Ideas Steve was not sure about:<br>
> 1. One hour twilight before all contest deadlines.<br>
<br>
I agree we should keep the frenzy and freeforall<br>
atmosphere for all the midcontest deadlines, ... but the<br>
one hour twilight before the final deadline does makes sense.<br>
<br>
<br>
> 2. Random rand seed.<br>
<br>
I think the repeatability of scoring the entries is<br>
critically important, so it would be a mistake to vary the<br>
random seed.<br>
<br>
Randomization does have a legitimate role in many realistic<br>
algorithms. In the Wiring Contest, there were often several<br>
"equivalent" choices for different wire paths. An effective<br>
strategy was to have a very fast algorithm so you could<br>
solve the problem two (or more) times, choosing different<br>
"random" paths each time, and then selecting the better<br>
solution.<br>
<br>
We don't know the actual test suite, and we don't know the<br>
actual random number seed. Both are part of the same<br>
phenomenon of "overfitting" in the leading entries.

Thu, 15 May 2008 18:41:04 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#432329
David
I would like to add another suggestion to the mix ...<br>
<br>
<br>
If people are generally happy with the idea of a "Late<br>
Twilight" phase in the closing hours of the Matlab Contest,<br>
it would make sense to give a prize, ... let's call it the<br>
"Dusk prize", since the sun would be setting on Daylight.<br>
<br>
This would reward a person who submits the leading entry "in<br>
the clear" during Daylight.<br>
<br>
After that, during the "Late Twilight", all entries would<br>
remain visible once they are scored, but entries sitting in<br>
the queue would not be visible.<br>
<br>
 David Jones

Sun, 25 May 2008 18:57:02 +0000
Re: MATLAB Central Spring Contest
http://www.mathworks.com/matlabcentral/newsreader/view_thread/168479#433931
Yi Cao
I have developed a postcontest solver to break the 13000 <br>
barrier. It is able to reduce results to below 130000 in <br>
about 30 seconds. Complexity is reduced to 10 and length in <br>
node to 2778.<br>
<br>
The code has been submitted to Matlab File Exchange. It has <br>
many tunable parameters for those who wish to optimize the <br>
code to play with. Please let me know your findings.<br>
<br>
Yi