more detailed README

README.md

@@ -1,6 +1,8 @@
 # coding-project-template
 
-This repository template is useful for developing Python packages with pre-commit autoformatting.
+This repository template is useful for developing research-oriented coding projects as Python packages.
+
+## Citation
 
 To cite the work that this code is associated with, use:
 
@@ -13,28 +15,74 @@ To cite the work that this code is associated with, use:
 }
 ```
 
-## Installation
+## Quickstart
 
-### Option #1: Conda install
+Clone the repo:
 
-To install via [Conda](https://docs.conda.io/), do:
+```sh
+git clone git@github.com:eringrant/coding-project-template.git
+cd coding-project-template/
+```
 
-```bash
-git clone git@github.com:eringrant/TODO_package_name.git
-cd TODO_package_name
-conda env create --file environment.yml
+To install a Conda environment with the requisite packages **on CPU**:
+
+```sh
+conda env create --file environment-cpu.yml
 ```
 
-The Conda environment can then be activated via `conda activate TODO_package_name`.
+To test that the package can be imported successfully:
 
-### Option: pip install
+```sh
+conda activate TODO_package_name
+python -c "import TODO_package_name"
+```
 
-To install via [pip](https://pip.pypa.io/), do:
+## Setup
 
-```bash
-git clone git@github.com:eringrant/TODO_package_name.git
-cd TODO_package_name
-pip install -e .
+### Installing packages
+
+Any of the following installation methods will allow you to activate the environment with
+`conda activate TODO_package_name`.
+
+#### Option #1: Conda install (CPU-only)
+
+To install via [Mamba](https://mamba.readthedocs.io/) (recommended)
+or [Conda](https://docs.conda.io/), do:
+
+```sh
+conda env create --file environment-cpu.yml
+```
+
+#### Option #2: Conda install on GPU on a local machine
+
+To install via [Mamba](https://mamba.readthedocs.io/) (recommended)
+or [Conda](https://docs.conda.io/) with GPU support, do:
+
+```sh
+conda env create --file environment-gpu.yml
+```
+
+#### Option #3: Conda install on GPU on a SLURM cluster
+
+If working on the head node of a SLURM cluster, you will need
+to create a GPU-compatible environment on a compute node with an available GPU via:
+
+```sh
+srun --partition=gpu --gres=gpu:1 conda env create -f environment-gpu.yml
+```
+
+Note that you may have to adapt the `partition` name to the available partitions on your cluster;
+run `sinfo -s` to display details about partitions.
+
+### Storage locations
+
+To avoid using up too much disk space in your home directory, make sure to point some packages to
+alternative locations by adding the following to a shell configuration file, such as
+`~/.bashrc` or `~/.zshrc`. (Remember to replace the ellipsis `...` with your desired locations!)
+
+```sh
+# Transient output.
+export SCRATCH_HOME="..."
 ```
 
 ## Devtools