Fix conda environment issues.

README.md

@@ -19,19 +19,25 @@ A Python package for running hybrid machine learning/molecular mechanics (ML/MM)
 
 ## Installation
 
-First, create a conda environment with all of the required dependencies:
+First, create a Conda environment with all required dependencies:
 
 ```bash
 conda env create -f environment.yaml
 conda activate fes-ml
 ```
 
-Finally, install `fes-ml` in interactive mode within the activated environment:
+Next, install `fes-ml` in editable mode within the activated environment:
 
 ```bash
 pip install -e .
 ```
 
+For GPU functionality, you must have the appropriate CUDA drivers installed on your host system. On HPC login nodes without GPU drivers, set the `CONDA_OVERRIDE_CUDA` environment variable to the desired CUDA version before creating the Conda environment. For example:
+
+```bash
+CONDA_OVERRIDE_CUDA=12.6 conda env create -f environment.yaml
+```
+
 ## Alchemical Modifications
 
 The following alchemical transformations are supported in fes-ml:

environment.yaml

@@ -1,30 +1,31 @@
-name: fes-ml-aev
+name: fes-ml
 
 channels:
   - conda-forge
-  - openbiosim/label/emle
+  - openbiosim
 
 dependencies:
   - ambertools
   - compilers
-  - cudatoolkit=11.8
   - eigen
   - loguru
-  - openmm>=8.1
+  - openmm>=8.5
   - openmm-torch
-  - openmm-ml
+  - openmm-ml>=1.7
   - openmmforcefields
   - openff-toolkit
   - openff-interchange
   - nnpops
   - pip
   - pybind11
-  - pytorch=*=*cuda*
+  - pytorch
   - python
   - pyyaml
   - sire
   - torchani
   - pygit2
+  - setuptools<82.0
+  - pymbar
   - pip:
     - git+https://github.com/chemle/emle-engine.git
     - coloredlogs