Speed up Matlab funtion optimset or fsolve

I repaired my post:

pop =   [   17.5489   22.0450   12.3356         0         0         0
   24.2207   24.4891   11.6109         0         0         0
   14.4916   24.0423    5.3751         0         0         0
   15.5933   24.2429    1.5084         0         0         0
   14.0768    8.4837   13.5116         0         0         0
   23.0225   18.9584   15.0561         0         0         0
   24.0094   14.7980    2.2642         0         0         0
   22.8346   24.8367    5.7287         0         0         0
   22.9392   15.0265   21.5263         0         0         0
   24.9098   23.5567    8.0397         0         0         0
   12.9304   18.0870    3.6067         0         0         0
   10.7218    4.9208   23.2223         0         0         0
   18.4719    6.1156   13.3655         0         0         0
    7.1474   10.3145   21.6322         0         0         0
   24.2926   18.5611   21.2992         0         0         0
    8.7695   14.3906    7.3888         0         0         0
    9.4825    2.9985    4.1883         0         0         0
   19.7259    9.3699   11.3208         0         0         0
   14.6069    8.1301   14.1943         0         0         0
   21.7199    8.1151   22.3485         0         0         0
    1.0794   18.6137   16.8374         0         0         0
   13.7829    6.0216   14.2601         0         0         0
   15.9794   17.3477   10.1006         0         0         0
   16.5683   18.8737   10.2242         0         0         0
   16.5552    3.8948   22.9116         0         0         0
   21.0076   17.0693   19.1147         0         0         0
   12.7140    2.1515   14.9522         0         0         0
    7.5796    4.3793    7.7026         0         0         0
   13.7626   24.1572   23.8614         0         0         0
    9.5798    9.9038   14.7426         0         0         0
   13.1142   22.8105   23.3993         0         0         0
    4.4876   14.6915   10.0169         0         0         0
    3.1052   14.1913   21.2961         0         0         0
   15.0551    7.6167   12.4830         0         0         0
   17.8893    8.8950    3.4193         0         0         0
    2.4890    1.1493   16.6434         0         0         0
   14.8805   23.2007   18.4197         0         0         0
   22.6164   20.7104   20.8279         0         0         0
    4.4673   12.8200   24.1456         0         0         0
   16.2761   18.2880   15.7975         0         0         0
   24.6292    7.2568    4.6912         0         0         0
   18.3555   10.0734    6.3775         0         0         0
    5.0947    5.8278   10.4251         0         0         0
   14.9085    4.0396   10.6858         0         0         0
   10.4061   17.2763   24.1169         0         0         0
    2.7613   12.6563   22.6465         0         0         0
    3.6709   10.5564   10.4706         0         0         0
    6.5381    8.2180    4.5643         0         0         0
   10.1283    4.4854   10.8909         0         0         0
   14.1282   10.2230    5.9243         0         0         0
   23.5585    2.1127    6.2529         0         0         0
   24.1545    2.5714    8.1056         0         0         0
   11.5353   16.8609    4.1955         0         0         0
    7.1352   21.6823   10.7806         0         0         0
   16.5840   16.9111    3.5680         0         0         0
   22.2396    9.2008   17.2345         0         0         0
   18.5981   13.9947    7.1893         0         0         0
    2.9765   24.6996   12.7910         0         0         0
   13.8668    2.4006   21.3071         0         0         0
    1.8152    6.2962    1.5340         0         0         0
    2.3862   11.2142   15.6256         0         0         0
    1.4202   23.4801    6.8175         0         0         0
    4.9021    9.4990    8.9636         0         0         0
   21.0888   18.8429   13.4654         0         0         0
   10.5702   21.0224   19.9158         0         0         0
   19.2113   16.6863    7.6894         0         0         0
    5.9662    5.3120    8.8289         0         0         0
   12.5199    5.2563    5.1281         0         0         0
   17.0951   24.5651   11.9476         0         0         0
   17.1268   16.9841   18.6717         0         0         0
   11.5547   23.8365    1.5220         0         0         0
    3.1174   15.5810   13.3654         0         0         0
    5.4717   16.0996   13.6953         0         0         0
   19.9857   17.1725    5.2883         0         0         0
    6.3686   12.0594   13.6691         0         0         0
    2.1712   11.9174    8.4521         0         0         0
   24.9662   17.5384    3.3057         0         0         0
   22.1810   24.8619   19.5151         0         0         0
    5.5853    7.9903   20.8547         0         0         0
   15.9705   16.3129   12.3720         0         0         0
    3.6243   16.5996   15.2950         0         0         0
    1.7450   16.6138   17.6058         0         0         0
   18.9411   15.2876    9.0251         0         0         0
   16.6245   17.9257    6.9308         0         0         0
    3.7184    4.2381    7.6155         0         0         0
   15.9711    2.9491    1.5460         0         0         0
   16.3059   22.6271    2.2782         0         0         0
    9.2693   24.7359   22.6225         0         0         0
    5.1539    7.8712    1.0588         0         0         0
   23.9903    3.0403    6.0633         0         0         0
    6.8119    9.3094    3.5283         0         0         0
   21.1938    5.9598   19.9863         0         0         0
   11.1118    4.7188   18.6718         0         0         0
   12.9967    1.6915   19.8843         0         0         0
   19.8283    1.0944   24.3935         0         0         0
    1.6343   13.8421    9.8628         0         0         0
   12.1101    9.7896   23.5194         0         0         0
    9.2871    8.9912    1.3104         0         0         0
   13.6724   18.1415    4.1001         0         0         0
    2.1903    2.8484   17.0288         0         0         0]
for m=1:100
        options=optimset('fsolve');
        f=@(x)henchung(x,pop,busdata,SpecG,m);
        [x,y,h]=fsolve(f,x0,options);
        result_line;
        pop(m,end) = Ploss*1000;
        pop(m,end-1)=Over;
        pop(m,end-2)=Delf;
        clear x y h;
    end;
    pop
I want to improve some way to reduce caculation time to run fsolve or optimset.

Answer 1

[Alan Weiss]

It is hard to know what you are asking, because you give code that is not in a loop, so we cannot see what you are really optimizing. As a first guess, I would say don't put the optimset call inside a loop, because you need only call it once. But this might be an irrelevant comment, because you did not show us your real problem, which includes a loop.

Alan Weiss

MATLAB mathematical toolbox documentation

[Quynh tran]

thanks for your contribution. My code is not problem. I repaired my code. Please check.

Answer 2

[John D'Errico]

There is no need to call optimset more than once, so why do you think you need to speed it up? Save the options structure, and be done with that.

As far as fsolve goes, everybody wants things to go faster. Solve fewer problems. Get a faster computer. Those are the easy solutions.

Harder is to solve multiple problems in a single call, using a sparse (block diagonal) Jacobian pattern. Since you can in theory solve several hundred problems at once that way, it can gain a bit. (I've typically seen the max gain come from around 100-200 batched subproblems.) It will require good starting values. It will require understanding of how to use options like 'JacobPattern'. It will require setting up the objective function to handle batches of sub-problems.

[Quynh tran]

my code just is an example. Because my file need to solve with 3000 times or more for pop code, not only 100 times. So I want to reduce time to caculate. Thaks.

[Steven Lord]

Since you didn't show us the source code for the henchung function, it will be impossible for us to offer any concrete suggestions beyond recommending that you profile your henchung function to identify bottlenecks then work to improve the performance of those sections of your code.

[Quynh tran]

thaks