I recently installed Python and pyxll (plus the required additional libraries) on a new computer, which raised a few problems with incompatible versions, so here is a summary of what worked (as of 26th September 2022):
Office should be installed and working before starting the Python installation. Then download and install Python Rel. 3.10.7 (must be the same bit number as Excel). Make sure that the options to install pip and tcl/tk and IDLE are selected.
When Python is installed the pip library installer can be used to install the rest of the required packages, including pyxll. To install pyxll (see here for more details) enter at a command line:
pip install pyxll
As a minimum, numpy, scipy and pandas should also be installed. For many of the applications published on this blog the numba jit compiler is also required:
pip install numpy
pip install scipy
pip install pandas
pip install numba
For applications requiring the solution of large sparse matrix equations the pypardiso library is recommended as being much faster than the sparse solvers included in scipy. This may now be installed simply with pip:
pip install pypardiso
For working with text based equations and units the following libraries are required:
pip install sympy
pip install pint
Note that sympy requires mpmath for multi-precision arithmetic, but this is now included in the pip installation.
For plotting graphics (including 3D graphs and animations) the following libraries are required. Note that the specific releases listed below are required.
pip install matplotlib==3.5.2
pip install plotly
pip install kaleido==0.1.0post
If earlier or later releases of matplotlib or kaleido have been installed, they can be replaced with the required release with:
pip install –upgrade matplotlib==3.5.2
The “–upgrade” command will work to either “upgrade” or “downgrade” the installed package.
Finally, the sectionproperties application, including all required additional libraries, may now be simply installed with:
pip install sectionproperties