Reputation: 11
I want to find constants a1, b1 which minimize
So I write following matlab code(nlp.m
):
function NLP
x0 = ones(2, 1);
[a1b1, sum1] = fmincon(@objfun, x0, [], [], [], [], [], [], [])
function sum1 = objfun(a1b1)
sum1 = sum(abs(a1b1(1) * x + a1b1(2) - y));
with x
and y
n-vector created in workspace
However, when I run nlp.m
I get the following error:
>> nlp
Undefined function or variable 'x'.
...
Seems that I can't get access to variables defined in workspace. So how can I deal with it?
Upvotes: 1
Views: 858
Reputation: 926
You can access variables in the base workspace directly from a script but not a function. To pass these variables to your objective function see http://www.mathworks.com/help/optim/ug/passing-extra-parameters.html.
So you could change your function to using a script like this:
% NLP - script
x0 = ones(2, 1);
f = @(a1b1)objfun(a1b1,x,y);
[a1b1, sum1] = fmincon(@f, x0)
And define your objective function in a separate function:
function sum1 = objfun(a1b1, x, y)
sum1 = sum(abs(a1b1(1) * x + a1b1(2) - y));
An other alternative would be to save the base workspace to a file and load this file in your main function. In this case you could update your function like this:
function nlp
% save workspace variable to file
evalin('base','save(''ws.mat'',''x'',''y'');')
x0 = ones(2, 1);
s=load('ws.mat');
x=s.x;y=s.y;
f = @(a1b1)objfun(a1b1, x, y);
[a1b1, sum1] = fmincon(@f, x0);
function sum1 = objfun(a1b1, x, y)
sum1 = sum(abs(a1b1(1) * x + a1b1(2) - y));
Upvotes: 1