使用MATLAB代码创建Python应用遇LD_LIBRARY_PATH及导入错误如何解决？

Hey there, let's work through this LD_LIBRARY_PATH issue you're hitting when trying to import your Matrices_Clifford module. I've run into similar headaches with MATLAB-Python integrations before, so here are some targeted fixes to try:

• Verify MATLAB Runtime Libraries Are in LD_LIBRARY_PATH
  First, make sure your system knows where to find the MATLAB Runtime libraries. The path usually looks something like /usr/local/MATLAB/MATLAB_Runtime/R202Xx/runtime/glnxa64 (adjust the version number to match your setup).

  • Check the current value of your path with:

    echo $LD_LIBRARY_PATH
    

  • If the MATLAB Runtime path isn't listed, add it temporarily to test:

    export LD_LIBRARY_PATH=/usr/local/MATLAB/MATLAB_Runtime/R202Xx/runtime/glnxa64:$LD_LIBRARY_PATH
    

  • Try importing Matrices_Clifford again. If this works, add the export line to your ~/.bashrc (or ~/.zshrc for zsh users) to make the change permanent.

• Check the Module's Generated setup.py Configuration
  Sometimes the auto-generated setup.py from MATLAB misses proper library path references. Open your setup.py and look for sections like library_dirs or runtime_library_dirs.

  • Ensure it includes the MATLAB Runtime library path. If not, add it manually:

    setup(
        name='Matrices_Clifford',
        version='1.0',
        description='MATLAB generated Python module',
        author='MathWorks',
        packages=['Matrices_Clifford'],
        package_dir={'Matrices_Clifford': 'Matrices_Clifford'},
        library_dirs=['/usr/local/MATLAB/MATLAB_Runtime/R202Xx/runtime/glnxa64'],
        runtime_library_dirs=['/usr/local/MATLAB/MATLAB_Runtime/R202Xx/runtime/glnxa64']
    )
    

  • Re-run python setup.py install after modifying, then test the import again.

• Use matlab.engine to Initialize the Runtime First
  Initializing the MATLAB engine before importing your custom module can resolve path conflicts by loading the runtime properly upfront. Add this to the top of your Python script:

  import matlab.engine
  eng = matlab.engine.start_matlab()
  import Matrices_Clifford
  

• Check for Conflicting Library Versions
  System libraries like libstdc++ or libgcc can clash with MATLAB Runtime's bundled versions. Run ldd on your module's shared library to spot issues:

  ldd Matrices_Clifford/*.so
  

  Look for lines marked not found or version mismatches. If you see conflicts, double-check that the MATLAB Runtime path comes first in LD_LIBRARY_PATH (as in the first step).

• Re-generate the Python Module from MATLAB
  If all else fails, there might have been a glitch during initial code generation. Go back to MATLAB, re-run deploytool or compiler.build.pythonPackage, making sure you select the correct Runtime version and target platform. Then re-install the newly generated module.

One of these steps should get you past the LD_LIBRARY_PATH error. Let me know if you hit any snags with specific steps!

内容的提问来源于stack exchange，提问作者David Armendariz