[scilab-Users] Solving multi-variable function

Fri Mar 14 20:06:13 CET 2008

Oh I see.  I was using fsolve incorrectly.  Thank you Eric.

Can you or someone give me an example of fsolve solving two or more variables with two or more equations?  This way I can use it correctly next time?  Thank you.

Jerry

----- Original Message ----
From: Eric Dubois <grocer.toolbox at gmail.com>
To: users at lists.scilab.org
Sent: Friday, March 14, 2008 1:53:06 AM
Subject: Re: [scilab-Users] Solving multi-variable function

Sorry for the misunderstanding.

To use fsolve, you need to have as many relations as variables, which is not the case with your test.

If you have less equations than variables, but your problem has one solution and only one (which is neither the case for your function test_fsolve), then use optim with the square of your initial function.

Éric.

PS: by the way the error message is not as speaking as it should be...

2008/3/14, Jerry Wang <jerrwang at yahoo.com>: 
Thanks Eric,

But what do you mean defining x0 inside vs outside?  The func_for_fsolve function is declaired as 
      function [y] = func_for_fsolve ( x0 )
and the caller provides the x0 as an vector into it...  
      [x_result,v,info]=fsolve([ret_age, mu_tree, delta_t, sigma_b],func_for_fsolve);

Should be okay right?

Also, are you able to run the simplified test_fsolve.sci and test_fsolve_caller.sce?  Do they work for you?  Or are you also getting error 98?
http://www.imagingspot.com/scilab/test_fsolve.sci.txt
http://www.imagingspot.com/scilab/test_fsolve_caller.sce.txt

Jerry

----- Original Message ----
From: Eric Dubois <grocer.toolbox at gmail.com>
To: users at lists.scilab.org

Sent: Thursday, March 13, 2008 2:16:49 PM
Subject: Re: [scilab-Users] Solving multi-variable function

I cannot run your function because you do not provide the function ABSPredictor. But, if your function func_for_fsolve.sci is exactly what is on your website, there are some oddities. In particular, you should not define x0 inside the function to maximise, but outside. 

Eric.

2008/3/13, Jerry Wang <jerrwang at yahoo.com>: 
Oh, sorry, somehow the website doesn't like the .sci extension.  Please see it here:
http://www.imagingspot.com/scilab/func_for_fsolve.txt

Yes, I provided initial values to the 4 variables.  The "caller" function has:

chdir('C:\MyScilab');
getf('ABSPredictor.sci')
getf('func_for_fsolve.sci');
tic();
ret_age=60;
mu_tree=36000;
delta_t=0.85;
sigma_b=0.50;
[x_result,v,info]=fsolve([ret_age, mu_tree, delta_t, sigma_b],func_for_fsolve);
time_length=toc()
beep();  

When debugging, the func_for_fsolve does fine for the first iteration.  Then it gives that error at the end of the first iteration.

Jerry

----- Original Message ----
From: Eric Dubois <grocer.toolbox at gmail.com>
To: users at lists.scilab.org

Sent: Thursday, March 13, 2008 12:46:42 PM
Subject: Re: [scilab-Users] Solving multi-variable function

Unfortunately, the function is not available at the provided address...

Did you provide values to ret_age, mu_tree, delta_t, sigma_b, in order to  feed the optimization program?

Eric.

2008/3/13, Jerry Wang <jerrwang at yahoo.com>: 
I did attempt to collapse the 4 variables into a single vector array but I received an error after fsolve reaches the end of the first iteration.  The error is:

!--error 98
 variable returned by scilab argument function is incorrect

--------------------------------------
I add the function into the scope via:
--------------------------------------
    getf('func_for_fsolve.sci');

--------------------------------------
I call the function via:
--------------------------------------
    [x_result,v,info]=fsolve([ret_age, mu_tree, delta_t, sigma_b],func_for_fsolve);

--------------------------------------
Inside func_for_solve I have:
--------------------------------------
function [y] = func_for_fsolve ( x0 )
...
ret_age = round(x0(1));
mu_tree = x0(2);
delta_t = x0(3);
sigma_b = x0(4);
...
[statistical_sustainable] = ABSPredictor(mc,now_age,ret_age,mu_tree,delta_t,sigma_b);
...
y = abs(statistical_sustainable(index)./100 - 0.90);
...
y = y + penalties;
disp('testing123');
disp('testing234');
endfunction

--------------------------------------
--------------------------------------

I tried testing the code by reducing the supplied argument x0 into 1 scalar variable instead of the 4 scalar vector, and the function call worked.  That's why I went ahead and assumed that fsolve takes only one variables.  Can you see what I am doing wrong that's causing me to receive:
!--error 98
 variable returned by scilab argument function is incorrect

Thank you!

Jerry

ps. the complete func_for_fsolve.sci is uploaded here for your viewing:
http://www.imagingspot.com/scilab/func_for_fsolve.sci

----- Original Message ----
From: Eric Dubois <grocer.toolbox at gmail.com>
To: users at lists.scilab.org
Sent: Thursday, March 13, 2008 5:41:21 AM
Subject: Re: [scilab-Users] Solving multi-variable function

Fsolve takes one variable input, but the input can be a vector of any size.

So, collapse your 4 variables into a single vector and adapt your fonction accordingly.

Eric.

2008/3/13, Jerry Wang <jerrwang at yahoo.com>: 
Hello Scilab-ers,

Can any one point me to the right direction?  I want to find the zero point of a function that has 4 variables.  I want scilab to solve for the values of the 4 variables that would give me a result = 0 for the answer.  I initially tried fsolve, but I discovered that fsolve only take one variable input.

Any thoughts?  Thanks!

Jerry

Be a better friend, newshound, and know-it-all with Yahoo! Mobile. Try it now.

Never miss a thing. Make Yahoo your homepage. 

Be a better friend, newshound, and know-it-all with Yahoo! Mobile. Try it now.

      ____________________________________________________________________________________
Looking for last minute shopping deals?  
Find them fast with Yahoo! Search.  http://tools.search.yahoo.com/newsearch/category.php?category=shopping
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.scilab.org/pipermail/users/attachments/20080314/aaae11a5/attachment.htm>