Merge pull request #1 from jose-d/main

remove multiprocessing from readme, introduce INSTALL.md to formalize what to do to install the code.

INSTALL.md

@@ -0,0 +1,36 @@
+# Manual Installation of EMRI_MC using Python venv
+
+## Runtime environment preparation
+
+Python - code was tested with #FIXME! python3.11. Make sure development package for given python version is installed too. Eg. at EL-like linux distros, `dnf install python3.11-devel`.
+
+### Ensure CUDA is present
+
+eg. at cluster environments, load CUDA toolkit. [^1]
+
+```
+module load CUDA/12.2.0
+```
+
+### Create venv
+
+...and install recent `pip`..
+
+```
+python3.11 -m venv ~/.venvs/EMRI_MC
+pip install --upgrade pip
+```
+
+### Activate venv and install code dependencies
+
+`source ~/.venvs/EMRI_MC/bin/activate`
+
+#WIP, so far exact list of dependencies in unclear.
+
+note: install step `Running setup.py install for cupy ...` can take long time.
+
+`pip install cupy emcee seaborn matplotlib pandas numpy emcee IPython ipykernel scipy h5py`
+
+# Notes
+
+[^1]: CUDA Toolkit - https://developer.nvidia.com/cuda-toolkit

README.md

@@ -8,16 +8,9 @@ On top of Python's usual **numpy** and **scipy** libraries, to run the code one
 
 1. **cuda/cupy**: https://nvidia.github.io/cuda-python/install.html
 
-
-
 2. **emcee**: https://emcee.readthedocs.io/en/stable/user/install/
 
-
-
-3. **multiprocessing**: https://pypi.org/project/multiprocessing/
-
-
-4. **corner** or **seaborn** or anything similar: https://corner.readthedocs.io/en/latest/install/  https://seaborn.pydata.org/installing.html
+3. **corner** or **seaborn** or anything similar: https://corner.readthedocs.io/en/latest/install/  https://seaborn.pydata.org/installing.html
 
 The cuda/cupy library provides the GPU functionality, emcee is the library with the MCMC sampler, multiprocessing is needed for parallelisation, and the corner/seaborn for producing the corner plots.