Run A SciPy Container From Den

This tutorial builds a small scientific Python container from the local Den OCI base image. It is a practical starter for data and scientific users who want a repeatable conda stack without first creating a bootable VM appliance.

Build The Den Base Image

From the Den checkout:

ct run oci-image

This creates:

den:dev

Build The Job Container

Build the example image:

docker build \
  --platform linux/arm64 \
  -t den-scipy-job-example:local \
  examples/oci/scipy-job

Use --platform linux/amd64 instead when den:dev was built from the linux-64 Den target.

The example Dockerfile starts from den:dev, creates a conda environment under /opt/app, installs NumPy and SciPy from conda-forge, and copies a small analysis script into /srv/den-scipy-job/.

Run It

Run the job and print the result:

docker run --rm den-scipy-job-example:local

Write the result to a mounted directory:

mkdir -p build/example-output
docker run --rm \
  -e DEN_SCIPY_OUTPUT=/output/result.json \
  -v "$PWD/build/example-output:/output" \
  den-scipy-job-example:local

Inspect the output:

cat build/example-output/result.json

What This Proves

This proves that Den can be used as a conda-native base image for scientific containers. It does not prove multi-user HPC node integration, scheduler integration, GPU support, or shared filesystem policy.