MATLAB Coder Error - optimset Unrecognized parameter name: 'GradObj'

Hi,
I'm trying to generate C code of a function using MATLAB Coder but I am getting the following error when generating MEX:
Unrecognized parameter name: 'GradObj'
Full instruction is:
optionsNLP = optimset('GradObj', 'off', 'GradConstr', 'off', ...
'DerivativeCheck', 'off', 'Display', 'iter', 'TolX', 1e-9, ...
'TolFun', 1e-9, 'TolCon', 1e-9, 'MaxFunEval', 300, 'DiffMinChange', 1e-5);
It is working when I run the MATLAB script directly.
I am thinking that maybe it can not be generated because a Toolbox is required to use the GradObj parameter?
Does anyone knows a workaround if any?
Thanks.
Vadim

 采纳的回答

Hi Vadim,
For generating C code for fmincon (or lsqnonlin, etc.) via MATLAB Coder, we require that you use the optimoptions function to set options. See the requirements in this documentation page.
Also, an important note is that only the SQP algorithm of fmincon is supported for C code generation.
The equivalent code for your example is as follows
optionsNLP = optimoptions('fmincon','Algorithm','sqp','SpecifyObjectiveGradient', false, ...
'CheckGradients', false, 'Display', 'iter', 'StepTolerance', 1e-9, 'FunctionTolerance', 1e-9, ...
'ConstraintTolerance', 1e-9, 'MaxFunctionEvaluations', 300, 'DiffMinChange', 1e-5);

3 个评论

Hi Steve,
Thanks a lot for your answer.
Indeed I finally found documentation for this (https://fr.mathworks.com/help/optim/ug/code-generation-for-optimization.html#mw_95c0ee1d-e2bf-420d-84db-e224ee882325) and optimoptions seems the way to go.
Again thank you.
I just noticed a small typo in the code sample you provided:
optionsNLP = optimoptions('fmiuncon',...)
Should be:
optionsNLP = optimoptions('fmincon'
Good catch! Thanks for pointing that out.
I guess I should have tried the new "Run" button on my answer. :)

请先登录,再进行评论。

更多回答(0 个)

类别

帮助中心File Exchange 中查找有关 MATLAB Coder 的更多信息

产品

版本

R2020a

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by