YALMIP have some issue sending plog to global solver

[JackJack]

Dear Professor Johan 

This program cannot work with Baron, but it can work well with bmibnb.

x = sdpvar(1);

y = sdpvar(1);

t = sdpvar(1);

obj = -t;

cons = [];

cons = [cons, 100 <= x <= 200];

cons = [cons, 300 <= y <= 400];

cons = [cons, t <= plog(y, x + y)];

optsB = sdpsettings('solver','baron', ...

'verbose',1, ...

'savesolveroutput',1, ...

'savesolverinput',0);

tic;

solB = optimize(cons, obj, optsB); The error is as follows: Optimization Status: 9 (Unknown problem (learn to debug) (Unknown problem (learn to debug) (Error using baron
There was an error evaluating a user supplied function. Please examine the below error and correct your function.

Error using "plog" (line 55)
PLOG called with weird argument

Error Trace:
- Error in "@(x)[1*x(3)-plog(x(4))]" (line 0)
- Error in "baron" (line 0)
- Error in "baron" (line 0)
- Error in "baron" (line 0)
- Error in "callbaron" (line 78)
- Error in "solvesdp" (line 423)
- Error in "optimize" (line 31)
- Error in "your_program" (line 100)
)))

Is there any way to overcome this?

[Johan Löfberg]

Right, I will have to clean away those. Until then, simply don't use plog but write the function explicitly. I've raised a bug to remember

https://github.com/yalmip/YALMIP/issues/1525

[Johan Löfberg]

Fixed in develop

On Thursday, 23 October 2025 at 04:46:47 UTC+2 JackJack wrote:

[JackJack]

I hope that this fix also fix the problem with the following 

1. exp, pexp
2. log, log2, log10, slog, plog
3. entropy, logsumexp, kullbackleibler

[Johan Löfberg]

kullbackleibler and crossentropy added (not tested though as my license was broken)

[JackJack]

Well after applying your fix 

clc;

clear all;

x = sdpvar(1);

y = sdpvar(1);

t = sdpvar(1);

obj = -t; % maximize t

cons = [];

cons = [cons, 100 <= x <= 200];

cons = [cons, 300 <= y <= 400];

cons = [cons, t <= plog(y, x + y)];

optsB = sdpsettings('solver','baron', ...

'verbose',1, ...

'savesolveroutput',1, ...

'savesolverinput',0,'CplexLibName','cplex12100.dll',...

'filekp',1);

tic;

solB = optimize(cons, obj, optsB);

elapsedTime = toc;

if solB.problem == 0

% Extract and display values

x_opt = value(x);

y_opt = value(y);

t_opt = value(t);

obj_opt = value(obj); % this is -t_opt

fprintf('Optimal solution found.\n');

fprintf('x* = %.10f\n', x_opt);

fprintf('y* = %.10f\n', y_opt);

fprintf('t* = %.10f\n', t_opt);

fprintf('Objective value (-t) = %.10f\n', obj_opt);

fprintf('Elapsed time = %.4f seconds\n', elapsedTime);

else

fprintf('Solver reported an issue: %s\n', yalmiperror(solB.problem));

end I run this program and got this strange output z =

    '(x(4)*log(x(5)/x(4)))'

Not sure why thought ?

[Johan Löfberg]

missing ; on line 163 in callbaron