Julia development Docker image for computational psychiatry

This repository provides a Julia-focused Docker image for use from Orca or VSCode.

The image is aimed at computational psychiatry and related modeling work, with package installation configured for:

• Bayesian statistics and probabilistic programming
• Message passing and variational inference
• Free energy principle / active inference adjacent workflows
• General Julia research and notebook-based development

Core development and analysis packages:

• IJulia
• Revise
• JuliaFormatter
• BenchmarkTools
• CSV
• DataFrames
• StatsBase
• StatsFuns
• Distributions
• KernelDensity
• StatsPlots
• CairoMakie

Bayesian statistics packages:

• Turing
• DynamicPPL
• MCMCChains
• AdvancedHMC
• Bijectors
• Optim
• BAT
• ApproxBayes
• GaussianProcesses
• ParetoSmooth
• ArviZ

Hierarchical / message-passing / active-inference adjacent packages:

• RxInfer
• ReactiveMP
• GraphPPL
• Rocket
• ForneyLab

The build also tries to install these research packages when available:

• GeneralizedFilters
• ActiveInference

docker build -t ghcr.io/ykunisato/ccp-j:latest .

docker run --rm -it \
  -v "$(pwd):/workspaces/ccp-j" \
  ghcr.io/ykunisato/ccp-j:latest

This image is intended to be published at ghcr.io/ykunisato/ccp-j.

The repository includes .devcontainer/devcontainer.json, so it can be opened directly as a development container from VSCode-compatible environments.

• The container starts as a non-root user named vscode.
• The working directory is /workspaces/ccp-j.
• JULIA_DEPOT_PATH is set to /usr/local/julia-depot.
• Package installation is best-effort during image build so amd64 and arm64 builds are less likely to fail on architecture-specific dependencies.
• If you want to pin exact package versions later, add a project-specific Project.toml and Manifest.toml.