Optimization Toolbox with MacBook Pro M1 silicon

I am writing a code to solve a Mixed-integer linear problem using the function 'intlinprog'.

When I run the following piece of code:

options = optimoptions('intlinprog', 'MaxTime',MaxTime, 'ObjectiveCutOff', 400,  'HeuristicsMaxNodes', 250,'CutMaxIterations', 50,'MaxNodes', 10^10, 'CutGeneration', 'advanced', 'IntegerPreprocess', 'advanced', 'Heuristics', 'advanced');
[sol, fval, exitflag, condition, lambda] = solve(Prob,'options',options);

Matlab gets stuck on this:

Solving problem using intlinprog.

LP: Optimal objective value is 166.066812.

and it doesn't start with branch and cut phase as usual.

Then after the 'MaxTime' has elapsed I get the following message:

"No feasible folution found."

"Intlinprog stopped because no point satisfies the constraint."

I tried to run the same code using both my old MacBook Air and Matlab Online and it worked fine, finding a lot of feasible solutions.

Has anyone experienced something similar?

What should I do? Should I unistall/reinstall the Optimization toolbox? Could it be a Rosetta 2 connected issue?

Thank you in advance!

[Giovanni Puccetti]

Experienced a similar problem with intlinprog. Have you fixed the issue in the meanwhile?

On Apple M1 it gets stuck, on Apple Intel based and online works finely (but probably not as fast as it could on M1).

[lorenzo giusti]

Fixed after updating to Matlab 2021...

[Joseph Mueller]

I am experiencing the same problem. Small mixed-integer problems run fine and are solved with intlinprog in under a second on Intel-based Mac (particularly using R2021b), but they take upwards of 30 seconds to complete when using the new Macbook Pro with M1 chip (tested on both R2021b and R2022a).

Answer 1

[YI CHEN]

Neither can I. The same code failed on my Mac with the M2 chip, when it was built and implemented successfully on my Mac with the Intel chip.