Run JAGS on Hopper

Set up R and JAGS in your own account, run a small Bayesian model, and download the results. No administrator installation is needed for this workflow.

Tested on Hopper with R 4.5.3, JAGS 4.3.2, and jagsUI 1.6.3. Allow about 20–30 minutes for setup, plus any queue wait. Software downloads total roughly 410 MB if you also need Conda; allow several GB of disk quota for installed files and caches.

Commands marked Your computer run in macOS Terminal, Linux Terminal, or Windows PowerShell. Commands marked Hopper run after SSH login. The server name in a prompt such as [you@hopper ~]$ tells you where you are.

Quick start

If you already have Conda and a Hopper account, download the folder in Step 1, then run this from the extracted folder. Run bash setup.sh first only if you have not installed the JAGS environment.

Hopper

source ./activate.sh
bash submit.sh

On the first submission, enter your authorized Slurm account and partition when prompted. The helper remembers them in this project folder. Then check the job:

Hopper

bash status.sh

New to Hopper? Follow the steps below. Use your own account, copy one command block at a time, and stop if a command fails. Model fitting belongs in a scheduled job, not on the login node.

1. Connect and download

Connect to your college VPN. Open Terminal or PowerShell and replace YOUR_USERNAME with your actual Hopper username.

Your computer

ssh YOUR_USERNAME@hopper.mckenna.edu

Enter your password; nothing appears as you type.

Once you are logged in, download and extract the starter directly onto Hopper:

Hopper

mkdir -p "$HOME/jags-workshop"
cd "$HOME/jags-workshop"
curl -fL -O https://www.cs.hmc.edu/~ndodds/guides/hopper-jags/hopper-jags-starter.zip
unzip -n hopper-jags-starter.zip
cd hopper-jags-starter

unzip -n keeps any files already present. To use a later starter release, extract it into a new folder so it does not mix with your edited model. If unzip is unavailable, extract the ZIP on your computer and transfer the folder with scp.

Download starter ZIP 340 KB Same ZIP as the button at the top of this page.

2. Check your account

You log in to a login node to edit files and submit work. Slurm, the scheduler, sends jobs to compute nodes. A partition is a queue with resource and time limits. A Slurm account identifies your allocation; it may differ from your username or lab group.

Hopper

hostname
uname -m
printf 'Shell: %s\n' "$SHELL"
command -v conda
module list
sinfo -s
sacctmgr -nP show assoc where user="$USER" format=Account,Partition

Expect hopper in the hostname, x86_64 for the architecture, and a Bash shell such as /bin/bash. If you use another shell, start bash for these instructions. No output from command -v conda simply means Conda is not loaded. If an R module is listed, run module unload R; this guide uses the environment's R.

An account row such as standard|debug allows that account in debug. A row such as standard| has no partition-specific restriction in that association; other partition and QOS access rules can still apply. Use sinfo -s to see queue names and limits, and ask support if your authorized queue is unclear. An empty account listing also needs support.

Keep your account name handy for the first submission. The short demo uses two cores, 2 GB RAM, and five minutes. Hopper's observed debug limit is one hour; main allows up to three days, subject to account policy. Check your quota if space is limited.

3. Make Conda available

Choose the path that matches your account. A working Miniconda, Miniforge, or Anaconda installation is enough.

Conda already works

Hopper

conda --version
source "$(conda info --base)/etc/profile.d/conda.sh"

Continue to Step 4.

Conda is installed but not loaded

Look for an installation:

Hopper

ls -d "$HOME/miniconda3" "$HOME/miniforge3" "$HOME/anaconda3" 2>/dev/null

Load the one you found. For miniconda3, use:

Hopper

source "$HOME/miniconda3/etc/profile.d/conda.sh"
conda --version

Substitute miniforge3 or anaconda3 if needed, then continue to Step 4. If no installation exists, follow the next section.

Install Miniforge if needed

Run these Linux commands on Hopper, even if your own computer is a Mac or Windows machine. They download a versioned installer from the official Miniforge project and verify its checksum.

Hopper

mkdir -p "$HOME/software-installers"
cd "$HOME/software-installers"
curl -fL -O https://github.com/conda-forge/miniforge/releases/download/26.7.2-0/Miniforge3-26.7.2-0-Linux-x86_64.sh
curl -fL -O https://github.com/conda-forge/miniforge/releases/download/26.7.2-0/Miniforge3-26.7.2-0-Linux-x86_64.sh.sha256
sha256sum -c Miniforge3-26.7.2-0-Linux-x86_64.sh.sha256

Proceed only if the result is OK:

./Miniforge3-26.7.2-0-Linux-x86_64.sh: OK

Hopper

bash Miniforge3-26.7.2-0-Linux-x86_64.sh

Press Enter to review the license, type yes if you accept it, press Enter for the default location, then answer no to shell initialization. If the destination already exists, load that installation instead of replacing it.

When installation finishes and your shell prompt returns:

Hopper

source "$HOME/miniforge3/etc/profile.d/conda.sh"
conda --version
cd "$HOME/jags-workshop/hopper-jags-starter"

4. Install and activate JAGS

For a new environment, run the setup once:

Hopper

cd "$HOME/jags-workshop/hopper-jags-starter"
bash setup.sh

It installs R, JAGS, and their dependencies from conda-forge in $HOME/.conda/envs/jags, then installs jagsUI from CRAN. It accepts the Conda package plan automatically. If the environment already exists, skip setup and verify it below.

Everyone activates and verifies:

Hopper

source ./activate.sh
command -v R Rscript
Rscript --vanilla -e 'cat(R.version.string,"\n"); library(rjags); library(jagsUI); print(jags.version()); print(packageVersion("jagsUI"))'

R and Rscript should resolve inside .conda/envs/jags/bin. Look for R 4.5.3, JAGS 4.3.2, and jagsUI 1.6.3. The traceplot “masked” message is normal: two packages provide a function with that name.

The helper stores your Conda path in conda-base.txt. After each new login, return to the folder and run source ./activate.sh. Scheduled jobs use the same helper.

5. Submit your first model

Hopper

bash submit.sh

The first time, enter your Slurm account and an authorized partition—usually debug for this short test. After Slurm accepts the job, the helper saves these values in .slurm-settings and the job ID in last-job-id.txt. They survive reconnects. Later submissions show the saved account and partition before submitting.

Hopper

bash status.sh

A pending job is waiting for resources; a running job needs more time. Run the status command again later. Success includes both the PASS line and a completed job with exit code 0:0:

PASS: parallel JAGS sampling completed successfully.
JobID             State ExitCode
366462        COMPLETED      0:0
366462.batch  COMPLETED      0:0

This is an excerpt from a successful run; your job ID will differ. The job writes hello-fit.rds, hello-summary.csv, hello-diagnostics.pdf, and hello-session.txt. Email your job ID and PASS line to ndodds@g.hmc.edu.

To change queues, use bash submit.sh --partition main if you are authorized for main. To enter account and partition again, use bash submit.sh --reset. Resource overrides are not saved: each run starts with the defaults in run.sbatch unless you supply options.

Finish one run before submitting another in the same folder. Re-running replaces the named result files; use project copies to keep experiments separate.

6. Read the results

Open another terminal on your computer and change into a folder where you want the results. Replace YOUR_USERNAME:

Your computer

scp 'YOUR_USERNAME@hopper.mckenna.edu:~/jags-workshop/hopper-jags-starter/hello-*' .

Open the CSV and PDF. The reference PDF shows the same example.

Try a small change

Hopper

nano hello.R

Change y <- c(1.8, 2.1, 2.0, 1.9, 2.2) to y <- c(2.8, 3.1, 3.0, 2.9, 3.2). Save with Control + O, then Enter; exit with Control + X.

Run bash submit.sh and bash status.sh again. The expected mean becomes 15 / 5.01 ≈ 2.994; the check updates automatically. Restore the original observations when finished.

7. Run your research model

Start in a separate project folder. The destination my-project should not already exist; choose a new name if it does.

Hopper

cd "$HOME/jags-workshop"
cp -r hopper-jags-starter my-project
cd my-project
cp research-template.R analysis.R
source ./activate.sh

Edit analysis.R to load your data and JAGS model. The template intentionally stops until you replace its example. It shows how to use the Slurm CPU allocation, save results, and record R session information automatically. If we are working through your model together, bring your actual script, a small dataset, your package list, and any custom JAGS modules.

For research runs, use main or another authorized longer partition rather than inheriting debug's one-hour limit. This example requests four cores, 4 GB RAM, and two hours; adjust to your model and account limits.

Hopper

bash submit.sh analysis.R --partition main --cpus-per-task 4 --mem 4G --time 02:00:00

Hopper

bash status.sh

The copied folder retains your saved account, and a successful submission saves main as this project's partition. The helper supports --account, --partition, --cpus-per-task, --mem, --time, and --job-name; use bash submit.sh --help for details. Options work as --name value or --name=value.

Files are read relative to the project folder. Rscript --vanilla does not restore a saved workspace, so explicitly load all data and packages. Keep parallel workers within the allocated CPU count. Start with a small model, inspect resource use with status.sh, and evaluate mixing, effective sample size, Rhat, and model-specific diagnostics before interpreting results.

8. Keep a reproducibility record

Once the real model runs, save its environment alongside the code, input-data version, seeds, and job settings:

Hopper

source ./activate.sh
conda list --explicit > conda-linux-64-explicit.txt
conda env export --no-builds > environment-record.yml
Rscript --vanilla -e 'write.csv(installed.packages()[,c("Package","Version","LibPath")], "r-packages.csv", row.names=FALSE); writeLines(capture.output(sessionInfo()), "session-info.txt")'

Conda exports do not include a complete recipe for the CRAN-installed jagsUI. Keep install-jagsUI.R and r-packages.csv too. The installer selects jagsUI 1.6.3 from current CRAN or, if necessary, CRAN's archive of older versions.

Recreate the environment

Run this from a folder containing the export and installer files. The destination must be a new environment. This command accepts the exported package plan automatically:

Hopper

conda create --yes --prefix "$HOME/.conda/envs/jags-reproduced" --file conda-linux-64-explicit.txt
conda activate "$HOME/.conda/envs/jags-reproduced"
Rscript --vanilla install-jagsUI.R

This export-and-reinstall path was tested on Hopper, including a successful scheduled demo in the rebuilt environment.

To test that environment with the helpers, run export JAGS_ENV_PREFIX="$HOME/.conda/envs/jags-reproduced" before activation and submission. Run unset JAGS_ENV_PREFIX to return to the default. These exports target compatible Hopper Linux systems, not your Mac or Windows R installation.

Troubleshooting

Email ndodds@g.hmc.edu or hpc-help@its.cmc.edu with your command, full error, job ID, and log. Never send a password. Cancel your own unwanted job with scancel followed by its actual job ID.

Validation and references

The example, fresh installation, Miniforge initialization, generated plots, and downloaded starter have been tested on Hopper. The current details are in VALIDATION.txt inside the ZIP. Each person's access and research-specific dependencies still need their own check.

The Miniforge installer was tested in batch mode; the guide uses its interactive prompts. The automatic CRAN archive fallback has not been exercised because version 1.6.3 is currently available on CRAN.

• Miniforge installation
• Conda environments and exports
• jagsUI documentation
• Slurm submissions and account associations