MATLAB mex interface to the HiGHS optimization library.
HiGHSMEX is not a part of the official HiGHS distribution.
HiGHSMEX is built with HiGHS v1.11.0.
For 64-bit Windows and macOS users the pre-compiled mex files highsmex.mexw64 and highsmex.mexmaca64 are provided so you do not have to install HiGHS on your system or, compile the mex file.
Thanks to Ray Zimmerman for providing help with the compilation of the mex file on macOS platform.
- Download or clone the HiGHS code. The highsmex.cpp file needs to be compiled with C++20 switch, hence it would be advisable to compile HiGHS with C++20 switch. To do so modify the CMakeLists.txt of HiGHS by changing the line
set(CMAKE_CXX_STANDARD 11)toset(CMAKE_CXX_STANDARD 20). Install HiGHS as described here. This should create a static library named highs e.g., highs.lib on Windows.
Note for macOS users Your cmake command should look like the following
cmake -S <path-to-source> -B <path-to-build> -DBUILD_SHARED_LIBS=OFF -DZLIB=OFF -DCMAKE_OSX_DEPLOYMENT_TARGET=13.4 - Open script named make_highsmex.m in MATLAB and specify the inputs. Then execute the script. This should create a mex file named highsmex.mex* e.g., highsmex.mexw64 on 64-bit Windows.
I do not have access to MATLAB on a Linux system so I cannot build the mex file myself. The above instructions should work on Linux system also. I am willing to provide assistance to anyone who wants to build the mex file on Linux.
HiGHS documentation is available here. A MATLAB function named callhighs is the interface to the HiGHS library. Run help callhighs on MATLAB command prompt to see the help on the input and output parameters of the callhighs function. Also, see the MATLAB script example_callhighs.m for various examples of usage.
Caution: Do not call the highsmex function directly, and only use the callhighs function.
HiGHSMEX provides access to almost all the capabilities of HiGHS library except the following
- Reading problem data from a model file.
- Setting names for the rows and columns of the model, or setting name for the objective.
- Advanced features described here.
- HiGHS callbacks.
Newer releases of MATLAB from R2024 include HiGHS as the LP and MILP solver. The advantages of HiGHSMEX are
- It can be used by MATLAB users with versions older than R2024.
- It can be used by MATLAB users with no access to MATLAB's Optimization Toolbox.
- It is quick and easy to integrate the new releases of HiGHS as they become available.
- Most of the features of HiGHS are available including QP, multi-objective LP, and, hot-starting.
- Provides access to all the options of HiGHS.
HiGHSMEX is covered by the MIT license.