Why does MATLAB crash when I run code using IBM's MATLAB-CPLEX Connector?

Why does MATLAB crash when I run code using IBM's MATLAB-CPLEX Connector?

Case 1:

[0] 0x000000018000d360 matrix::detail::noninlined::mx_array_api::try_destroy_container at b:\matlab\src\matrix\array2.cpp:2807
[1] 0x000000018000bcdf intrusive_ptr_release at b:\matlab\src\matrix\undoc.cpp:1940
[2] 0x0000000018be40e3 MathWorks::lxe::TypeConstructorMxarrayBase::DestroyValue at b:\matlab\src\lxe\types\typemxarray.cpp:132
[3] 0x0000000018be8944 MathWorks::lxe::TypeConstructorXvalue::ResetValue at b:\matlab\src\lxe\types\typexvalue.cpp:47
[4] 0x0000000016a6ce6f MathWorks::lxe::UniqueIrPointerTypeConstructor::ResetValue at b:\matlab\src\lxe\m_lxe\type\typeirpointer.cpp:208
[5] 0x0000000018beb5c7 MathWorks::lxe::TypeConstructorTuple::ResetValue at b:\matlab\src\lxe\types\typetuple.cpp:346

Case 2:

[  0] 0x00000000fe7fb2c3                              bin\win64\libmwfl.dll+00045763 foundation::core::diag::thread_context::unspecified_bool+00000051
[  1] 0x00000000fe7f9288                              bin\win64\libmwfl.dll+00037512 foundation::core::diag::stacktrace_base::capture+00000024
[  2] 0x00000000fe7ff09f                              bin\win64\libmwfl.dll+00061599 foundation::core::diag::symbols::getSymbolAddress+00015039
[  3] 0x00000000fe7fdcf8                              bin\win64\libmwfl.dll+00056568 foundation::core::diag::symbols::getSymbolAddress+00010008
[  4] 0x00000000fe8035bb                              bin\win64\libmwfl.dll+00079291 foundation::core::diag::terminate+00000059
[  5] 0x00000000041835fc                  bin\win64\libmwfoundation_usm.dll+00013820 foundation::usm::getErrorHandler+00000060
[  6] 0x000000000419a3cf                  bin\win64\libmwfoundation_usm.dll+00107471 foundation::usm::Detail<foundation::usm::scope::Mvm>::findOrFail+00000319
[  7] 0x0000000004547b53                                bin\win64\libut.dll+00293715 utGetInterruptStateData+00000083
[  8] 0x0000000004548129                                bin\win64\libut.dll+00295209 utIsInterruptPending+00000009
[  9] 0x000007fef7b131bd C:\Program Files\IBM\ILOG\CPLEX_Studio128\cplex\matlab\x64_win64\cplexlink1280.mexw64+00012733 mexFunction+00002125

Answer 1

[MathWorks Support Team]

This issue is related to incompatibilities between the CPLEX MEX functions and certain versions of MATLAB. Note that the MATLAB connector from CPLEX is tested for the available versions of CPLEX at development time. As such, each version of CPLEX is only guaranteed to be compatible with a subset of MATLAB releases. However, currently, IBM's MATLAB-CPLEX Connector is no longer supported after MATLAB R2019b, therefore any release of MATLAB from R2020a onwards is not tested by the MathWorks development team and we cannot guarantee the compatibility of any unsupported versions.

Therefore please only use supported versions of IBM ILOG CPLEX Optimization Studio and MATLAB.

 

The latest version of MATLAB and CPLEX that are compatible together is MATLAB R2019b with IBM ILOG CPLEX Optimization Studio Version 12.10.0.0. You can generate a detailed CPLEX compatibility report from IBM's website at the link below:

https://www.ibm.com/software/reports/compatibility/clarity/prereqsForProduct.html

The link brings you to the Related software for a specific product page on IBM's website:

On this page enter "IBM ILOG CPLEX Optimization Studio" into the Full or partial product name field, and select "IBM ILOG CPLEX Optimization Studio" from the search results. Then select the version you want to check the compatibility of, and pressing submit will make the compatibility table appear to show the supported versions of MATLAB you can use with the chosen CPLEX version. 

 

For example, see below the compatibility table generated for IBM ILOG CPLEX Optimization Studio 12.10.0.0, which shows MATLAB is supported from R2017a up to and including the R2019b release:

 

If you want to install an older release of MATLAB such that you are using a release that supports your version of CPLEX, then log in to your MathWorks account on our website and navigate to the Downloads page, where you can then select the relevant release to download.