Changing the default tolerance for fmincon

I am doing a computation using fmincon, and it says tolfun=1e-06 and options.TolCon = 1e-06. I am trying to have a tolerance of 10^-20 instead.

I guess my question is,how do I change the tolerance for fmincon?

Answer 1

[Walter Roberson]

options = optimset('TolCol', 1e-6);
fmincon(....., options)

[Del]

Thanks! It worked perfect

[Rahul Bhadani]

This answer is no longer valid in 2020, with R2020a version.

Answer 2

[Brian B]

You can pass an options structure to fmincon. You create the structure using optimset.

[Ehsan Khorsandnejad]

When I use this code fmincon(....., options), I get the following error:

Error using optimfcnchk (line 91) NONLCON must be a function.

Error in fmincon (line 410) confcn = optimfcnchk(NONLCON,'fmincon',length(varargin),funValCheck,flags.gradconst,false,true);

Error in Fitting_EIS_RQ_RC (line 39) x_final = fmincon(f,x_initial,[],[],[],[],[0,0,0,0,0,0,0,0],[10^4,10^4,10^4,10^4,10^4,10^4,10^4,1], options);

[Steven Lord]

You missed specifying [] for the nonlcon input between ub and the options structure.

x_final = fmincon(f,x_initial,... % fun, x0
    [],[],[],[], ... % A, b, Aeq, beq
    [0,0,0,0,0,0,0,0], ... % lb
    [10^4,10^4,10^4,10^4,10^4,10^4,10^4,1], ... % ub
    options); % this should be nonlcon, options

Make that last line:

    [], options);

Answer 3

[Alan Weiss]

Seting a tolerance of 1e-20 is the same as disabling that tolerance, or setting it to 0. See the documentation of tolerances.

Alan Weiss

MATLAB mathematical toolbox documentation