Rank: 3602 based on 30 downloads (last 30 days) and 3 files submitted
photo

Statovic

E-mail

Personal Profile:

 

Watch this Author's files

 

Files Posted by Statovic View all
Updated   File Tags Downloads
(last 30 days)
Comments Rating
28 Oct 2008 Generate sorted vectors of uniformly distributed variates Efficiently generate sorted p-vectors of variates drawn uniformly from [0,1] Author: Statovic random number generat..., statistics, simulation, mathematics, probability 9 3
  • 3.0
3.0 | 1 rating
24 Jul 2008 Exact Negative Log-likelihood of ARMA models via Kalman Filtering Computation of the exact negative log-likelihood of ARMA models using the Kalman Filter Author: Statovic statistics, probability, arma, kalman filter, loglikelihood 6 1
  • 3.0
3.0 | 1 rating
22 Jul 2008 Non-negative Garotte Implementation of Leo Breiman's non-negative garotte for linear regression Author: Statovic statistics, probability, regression, nonnegative garotte, linear regression, leo breiman 15 2
Comments and Ratings by Statovic
Updated File Comments Rating
28 Oct 2008 Generate sorted vectors of uniformly distributed variates Efficiently generate sorted p-vectors of variates drawn uniformly from [0,1] Author: Statovic

The important difference between sort(rand(...)) and this algorithm is that the sort(.) procedure is at best O(n log(n)) time complexity, while this algorithm is linear time, i.e. O(n). For short length vectors the sort(.) approach will be quicker (due to a more efficient implementation of the sort(.) function), but for large n the differences can be significant. As an example:

>> tic;sort(rand(1,10^7),2);toc
Elapsed time is 1.757046 seconds.

vs.

>> tic;genrandsorted(1,10^7);toc
Elapsed time is 0.531082 seconds.

I am guessing that if one implemented the Bentley & Saxe algorithm in C it would probably be comparable to sort(.) for smaller 'n' (and of course, faster again for large 'n').

Comments and Ratings on Statovic 's Files View all
Updated File Comment by Comments Rating
16 Sep 2009 Exact Negative Log-likelihood of ARMA models via Kalman Filtering Computation of the exact negative log-likelihood of ARMA models using the Kalman Filter Author: Statovic Fraunhofer IIS Alawieh

the program is well written, but i believe the state space representation is wrong.
Though the implemented part is as mentioned in the refernce paper still , I argue about the 'R' in arma_ConvertToSS

Taking the MA parameters directly is not the proper representation.
instead R=[ c1-a1;c2- a2-( c1-a1);.....];

' SPECTRAL ESTIMATION FOR NOISY SIGNALS OBSERVED THROUGH A LINEAR SYSTEM' Check this paper

15 Sep 2009 Exact Negative Log-likelihood of ARMA models via Kalman Filtering Computation of the exact negative log-likelihood of ARMA models using the Kalman Filter Author: Statovic Fraunhofer IIS Alawieh

28 Oct 2008 Generate sorted vectors of uniformly distributed variates Efficiently generate sorted p-vectors of variates drawn uniformly from [0,1] Author: Statovic Statovic

The important difference between sort(rand(...)) and this algorithm is that the sort(.) procedure is at best O(n log(n)) time complexity, while this algorithm is linear time, i.e. O(n). For short length vectors the sort(.) approach will be quicker (due to a more efficient implementation of the sort(.) function), but for large n the differences can be significant. As an example:

>> tic;sort(rand(1,10^7),2);toc
Elapsed time is 1.757046 seconds.

vs.

>> tic;genrandsorted(1,10^7);toc
Elapsed time is 0.531082 seconds.

I am guessing that if one implemented the Bentley & Saxe algorithm in C it would probably be comparable to sort(.) for smaller 'n' (and of course, faster again for large 'n').

28 Oct 2008 Generate sorted vectors of uniformly distributed variates Efficiently generate sorted p-vectors of variates drawn uniformly from [0,1] Author: Statovic Jos (10584)

I acknowledge that this is indeed a clever algorithm but in matlab
sort(rand(..)), as already suggested by Wolgang Schwanghart, is much to be prefered. Therefore this submission has only some educational value, for which the internal comments should be expanded.

28 Oct 2008 Generate sorted vectors of uniformly distributed variates Efficiently generate sorted p-vectors of variates drawn uniformly from [0,1] Author: Statovic Wolfgang Schwanghart

What makes the function different to the command
x = sort(rand(n,p),2);
? I don't have the cited paper at hand, so it might be good to explain the difference, if there are any.

Best regards,
Wolfgang

Contact us