I am using CVX Version 2.2, and Mosek solver (Version 9.1.9) is selected. But still getting below Warning, and also it is taking more time to run the code (taking very high time to provide optimal value).
CVX Warning:
Models involving “log_det” or other functions in the log, exp, and entropy
family are solved using an experimental successive approximation method.
This method is slower and less reliable than the method CVX employs for
other models. Please see the section of the user’s guide entitled
The successive approximation method
for more details about the approach, and for instructions on how to
suppress this warning message in the future.
I think this Warning can be suppressed by adding cvx_expert true command before cvx_begin quiet right? but this warning doesnot affects to the final result right?? Also, how to reduce time taken to run (taking very high run time)??
Also in the above link (7th command), I saw that “When you run the problem with log in CVX 2.2 with Mosek 9.1 installed, CVX should utilize Mosek’s native exponential cone capability, which should be the best option utilization of that capability”, but how to know that CVX is utilizing Mosek’s native exponential cone capability?? If I run cvx_solver command then Mosek 9.1.9 solver is default and selected, is it sufficient to check this or any other setup need to be done?? Thank you
Issuance of the warning when Mosek 9.x or 10.x is used is a known “bug”. Just ignore the warning. CVX does call Mosek to natively solve an exponential cone problem despite the warning to the contrary.
why it is taking more time to run the code (taking very high time to provide optimal value) even though mosek solver is selected, and how to reduce this run time??
How to solve the slower and less reliablity due to successive approximation method, even when Mosek is selected (provided in the CVX Warning)?
Above CVX Warning can be suppressed by adding cvx_expert true command before cvx_begin quiet right? but anyway this warning doesnot affects to our final result right?
How to know that CVX is utilizing Mosek’s native exponential cone capability? If I run cvx_solver command then Mosek 9.1.9 solver is default and selected, is it sufficient to check this or any other setup need to be done??
As I wrote before, ignore the warning. it is issued, but should not be issued, when Mosek 9.x or 10.x is used.
As for whether the run time can be improved, if you show a complete reproducible problem with all input data, and all CVX and Mosek output, perhaps one of the Mosek employees who post on this forum might be able to offer an assessment.
Thanks, but still not answered the following 2 queries:
CVX Warning:
Models involving “log_det” or other functions in the log, exp, and entropy
family are solved using an experimental successive approximation method.
This method is slower and less reliable than the method CVX employs for
other models. Please see the section of the user’s guide entitled
The successive approximation method
for more details about the approach, and for instructions on how to
suppress this warning message in the future.
Above CVX Warning can be suppressed by adding cvx_expert true command before cvx_begin quiet right? but anyway this warning doesnot affects to our final result right?
How to know that CVX is utilizing Mosek’s native exponential cone capability? If I run cvx_solver command then Mosek 9.1.9 solver is default and selected, is it sufficient to check this or any other setup need to be done??
