PyMC is a new open source Probabilistic Programming framework written in Python that uses Theano to compute gradients via automatic differentiation as well as compile probabilistic programs on-the-fly to C for increased speed. Contrary to other Probabilistic Programming languages, PyMC3 allows model specification directly in Python code.
Allocate an interactive session and run the program. Sample session:
[user@biowulf]$ sinteractive -c 16 --mem 40g salloc.exe: Pending job allocation 46116226 salloc.exe: job 46116226 queued and waiting for resources salloc.exe: job 46116226 has been allocated resources salloc.exe: Granted job allocation 46116226 salloc.exe: Waiting for resource configuration salloc.exe: Nodes cn3144 are ready for job [user@cn3144 ]$ module load pymc/3 [+] Loading singularity on cn3344 [+] Loading pymc 3.11.5 on cn3344 [user@cn3144 ]$ python-pymc Python 3.9.13 | packaged by conda-forge | (main, May 27 2022, 16:56:21) [GCC 10.3.0] on linux Type "help", "copyright", "credits" or "license" for more information. >>> import pymc3 as pm >>> import theano [user@cn3144 ]$ module load pymc/4 [+] Loading singularity on cn3344 [+] Loading pymc 4.1.3 on cn3344 [user@cn3144 ]$ python-pymc Python 3.10.5 | packaged by conda-forge | (main, Jun 14 2022, 07:04:59) [GCC 10.3.0] on linux Type "help", "copyright", "credits" or "license" for more information. >>> import pymc as pm >>> import aesara [user@cn3144 ]$ module load pymc/5 [+] Loading singularity on cn3344 [+] Loading pymc 5 on cn3344 [user@cn3144 ]$ python-pymc Python 3.10.5 | packaged by conda-forge | (main, Jun 14 2022, 07:04:59) [GCC 10.3.0] on linux Type "help", "copyright", "credits" or "license" for more information. >>> import pymc as pm >>> import jax [user@cn3144 ]$ exit salloc.exe: Relinquishing job allocation 46116226
Run pymc4 with JAX:
[user@biowulf ]$sinteractive --gres=gpu:p100:1 [user@cn3144 ]$ module load pymc/4 [+] Loading singularity on cn3344 [+] Loading pymc 4.1.3 on cn3344 [user@cn3144 ]$ python-pymc Python 3.10.5 | packaged by conda-forge | (main, Jun 14 2022, 07:04:59) [GCC 10.3.0] on linux Type "help", "copyright", "credits" or "license" for more information. >>> import pymc.sampling_jax as pmjax [user@cn3144 ]$ exit
Run pymc with jupyter notebook:
[user@biowulf ]$sinteractive -c 16 --mem 40g --tunnel
salloc.exe: Pending job allocation 46116226
salloc.exe: job 46116226 queued and waiting for resources
salloc.exe: job 46116226 has been allocated resources
salloc.exe: Granted job allocation 46116226
salloc.exe: Waiting for resource configuration
salloc.exe: Nodes cn3144 are ready for job
Created 1 generic SSH tunnel(s) from this compute node to
biowulf for your use at port numbers defined
in the $PORTn ($PORT1, ...) environment variables.
Please create a SSH tunnel from your workstation to these ports on biowulf.
On Linux/MacOS, open a terminal and run:
ssh -L 33327:localhost:33327 biowulf.nih.gov
For Windows instructions, see https://hpc.nih.gov/docs/tunneling
[user@cn3144]$ module load pymc/3
[user@cn3144]$ jupyter notebook --ip localhost --port $PORT1 --no-browser
[I 17:11:40.505 NotebookApp] Serving notebooks from local directory
[I 17:11:40.505 NotebookApp] Jupyter Notebook 6.4.10 is running at:
[I 17:11:40.505 NotebookApp] http://localhost:37859/?token=xxxxxxxx
[I 17:11:40.506 NotebookApp] or http://127.0.0.1:37859/?token=xxxxxxx
[I 17:11:40.506 NotebookApp] Use Control-C to stop this server and shut down all kernels (twice to skip confirmation).
[C 17:11:40.512 NotebookApp]
To access the notebook, open this file in a browser:
file:///home/apptest1/.local/share/jupyter/runtime/nbserver-29841-open.html
Or copy and paste one of these URLs:
http://localhost:37859/?token=xxxxxxx
or http://127.0.0.1:37859/?token=xxxxxxx
Then you can open a browser from your computer to connect to the jupyter notebook.
Create a python script (e.g. pymc-script.py). For example:
import pymc3 as pm
Create a batch input file (e.g. pymc.sh). For example:
#!/bin/bash module load pymc/3 python-pymc pymc-script.py
Submit this job using the Slurm sbatch command.
sbatch -c 16 --mem 40g pymc.sh