Thread Subject: matrix generation

Hi,
I want to generate a 6x4 matrix of positive integers(max value:50) such that the difference between the 3rd element and 1st element in each row is 2 and difference of 2nd and 4th row element in all rows is 3.
Any help?

jack j wrote:
> Hi, I want to generate a 6x4 matrix of positive integers(max
> value:50) such that the difference between the 3rd element and 1st
> element in each row is 2 and difference of 2nd and 4th row element in
> all rows is 3. Any help?

Generate (say) first and second columns then add 2 and 3, respectively.
    You can ensure the condition by either rejecting greater>50 values
and regenerating that pair or restrict the original rng range to <48 so
sum can exceed 50.

--

"jack j" <12mailjack@gmail.com> wrote in message <hd6uef$j43$1@fred.mathworks.com>...
> Hi,
> I want to generate a 6x4 matrix of positive integers(max value:50) such that the difference between the 3rd element and 1st element in each row is 2 and difference of 2nd and 4th row element in all rows is 3.
> Any help?

Hi,

Does it fit you?

% jack.m 2009-11-08
%%%%%%%%%%%%
A = zeros(6,4);
A(:,1) = ceil(48*rand(6,1));
A(:,3) = A(:,1)+2;
A(:,4) = ceil(47*rand(6,1)+3);
A(:,2) = A(:,4)-3

Mira

dpb <none@non.net> wrote in message <hd6utr$jgr$2@news.eternal-september.org>...
> jack j wrote:
> > Hi, I want to generate a 6x4 matrix of positive integers(max
> > value:50) such that the difference between the 3rd element and 1st
> > element in each row is 2 and difference of 2nd and 4th row element in
> > all rows is 3. Any help?
>
> Generate (say) first and second columns then add 2 and 3, respectively.
> You can ensure the condition by either rejecting greater>50 values
> and regenerating that pair or restrict the original rng range to <48 so
> sum can exceed 50.
>
> --

Did u mean a random number generation in a 'for' loop or a 6x2 random matrix generation and then a concatenation?

"Miroslav Balda" <miroslav.nospam@balda.cz> wrote in message <hd6vnm$7u3$1@fred.mathworks.com>...
> "jack j" <12mailjack@gmail.com> wrote in message <hd6uef$j43$1@fred.mathworks.com>...
> > Hi,
> > I want to generate a 6x4 matrix of positive integers(max value:50) such that the difference between the 3rd element and 1st element in each row is 2 and difference of 2nd and 4th row element in all rows is 3.
> > Any help?
>
> Hi,
>
> Does it fit you?
>
> % jack.m 2009-11-08
> %%%%%%%%%%%%
> A = zeros(6,4);
> A(:,1) = ceil(48*rand(6,1));
> A(:,3) = A(:,1)+2;
> A(:,4) = ceil(47*rand(6,1)+3);
> A(:,2) = A(:,4)-3
>
> Mira

Thankx Mira, it works....and thanks dpb......

jack j wrote:
> "Miroslav Balda" <miroslav.nospam@balda.cz> wrote in message <hd6vnm$7u3$1@fred.mathworks.com>...
>> "jack j" <12mailjack@gmail.com> wrote in message <hd6uef$j43$1@fred.mathworks.com>...
>>> Hi,
>>> I want to generate a 6x4 matrix of positive integers(max value:50) such that the difference between the 3rd element and 1st element in each row is 2 and difference of 2nd and 4th row element in all rows is 3.
>>> Any help?
>> Hi,
>>
>> Does it fit you?
>>
>> % jack.m 2009-11-08
>> %%%%%%%%%%%%
>> A = zeros(6,4);
>> A(:,1) = ceil(48*rand(6,1));
>> A(:,3) = A(:,1)+2;
>> A(:,4) = ceil(47*rand(6,1)+3);
>> A(:,2) = A(:,4)-3
>>
>> Mira
>
> Thankx Mira, it works....and thanks dpb......

Yes, precisely what I suggested altho he chose 1 and 4 as the two
independent and did a little finer pruning on the upper bounds than a
single limit.

--

dpb <none@non.net> wrote in message <hd71ii$9sb$1@news.eternal-september.org>...
> jack j wrote:
> > "Miroslav Balda" <miroslav.nospam@balda.cz> wrote in message <hd6vnm$7u3$1@fred.mathworks.com>...
> >> "jack j" <12mailjack@gmail.com> wrote in message <hd6uef$j43$1@fred.mathworks.com>...
> >>> Hi,
> >>> I want to generate a 6x4 matrix of positive integers(max value:50) such that the difference between the 3rd element and 1st element in each row is 2 and difference of 2nd and 4th row element in all rows is 3.
> >>> Any help?
> >> Hi,
> >>
> >> Does it fit you?
> >>
> >> % jack.m 2009-11-08
> >> %%%%%%%%%%%%
> >> A = zeros(6,4);
> >> A(:,1) = ceil(48*rand(6,1));
> >> A(:,3) = A(:,1)+2;
> >> A(:,4) = ceil(47*rand(6,1)+3);
> >> A(:,2) = A(:,4)-3
> >>
> >> Mira
> >
> > Thankx Mira, it works....and thanks dpb......
>
> Yes, precisely what I suggested altho he chose 1 and 4 as the two
> independent and did a little finer pruning on the upper bounds than a
> single limit.
>
> --

But what if the condition is something like " (1st column element-2nd column element)+(3rd column element-4th column element)"for all rows should be equal to x???

In this case will such a defn work out?

> But what if the condition is something like " (1st column element-2nd column element)+(3rd column element-4th column element)"for all rows should be equal to x???
>
> In this case will such a defn work out?

If the condition is "(1st column element-2nd column element)+(3rd column element-4th column element)" should be a number greater than 'x', then also I doubt whether such a column by column definition will work.

Please help.....

jack j wrote:
...
> But what if the condition is something like " (1st column element-2nd
> column element)+(3rd column element-4th column element)"for all rows
> should be equal to x???
>
> In this case will such a defn work out?

Possibly depending on the size of the selection pool and value of x.
The rudimentary case for the above would be generate three independently
and solve for the fourth as the parentheses make no difference
numerically to the overall sum (and since you didn't state any other
condition).

You'll have the same problem of occasionally (or more than occasionally,
perhaps) having to resample; whether that's acceptable or not on the
required population statistics or not is another question you'd have to
judge.

--

dpb wrote:
> jack j wrote:
> ...
>> But what if the condition is something like " (1st column element-2nd
>> column element)+(3rd column element-4th column element)"for all rows
>> should be equal to x???
>>
>> In this case will such a defn work out?
>
> Possibly depending on the size of the selection pool and value of x. The
> rudimentary case for the above would be generate three independently and
> solve for the fourth as the parentheses make no difference numerically
> to the overall sum (and since you didn't state any other condition).
>
> You'll have the same problem of occasionally (or more than occasionally,
> perhaps) having to resample; whether that's acceptable or not on the
> required population statistics or not is another question you'd have to
> judge.

NOTA BENE: In the above I assumed the "for all rows" means each row
independently satisfies the criterion.

--

dpb <none@non.net> wrote in message <hd75l9$alt$2@news.eternal-september.org>...
> dpb wrote:
> > jack j wrote:
> > ...
> >> But what if the condition is something like " (1st column element-2nd
> >> column element)+(3rd column element-4th column element)"for all rows
> >> should be equal to x???
> >>
> >> In this case will such a defn work out?
> >
> > Possibly depending on the size of the selection pool and value of x. The
> > rudimentary case for the above would be generate three independently and
> > solve for the fourth as the parentheses make no difference numerically
> > to the overall sum (and since you didn't state any other condition).
> >
> > You'll have the same problem of occasionally (or more than occasionally,
> > perhaps) having to resample; whether that's acceptable or not on the
> > required population statistics or not is another question you'd have to
> > judge.
>
> NOTA BENE: In the above I assumed the "for all rows" means each row
> independently satisfies the criterion.
>
> --

Yes, each row should independently satisfy the criterion.
What happens with the following condition?
(1st column element-2nd column element)^2+(3rd column element-4th column element)^2 should be greater than x.

jack j wrote:
...
> Yes, each row should independently satisfy the criterion.
> What happens with the following condition?
> (1st column element-2nd column element)^2+(3rd column element-4th
> column element)^2 should be greater than x.

What is this, 20 questions?

Work out the necessary algebra...

--