2. Detailed installation

This page covers a clean, isolated installation on each operating system. We strongly recommend installing into a dedicated environment so LatticeMind’s dependencies never collide with other Python projects.

2.1. Requirements

Python 3.10 through 3.14. The release provides cp310, cp311, cp312, cp313, and cp314 wheels for Linux x86-64, Windows AMD64, macOS Intel, and macOS Apple Silicon. pip picks the one matching your interpreter and architecture.
Internet access for the AI model (the calculations themselves can run offline) and for pip to reach the package indexes.
A valid LatticeMind license file. LatticeMind is licensed separately from RESCU and checks its own license before the CLI or Web UI starts, whichever solver is selected.
An AI provider API key, or a local Ollama model (see Connecting your AI model).
(To run calculations) the selected solver, installed and licensed on the computer that runs it, plus its licensed data. RESCU execution uses a RESCU license; NanoDCAL requires a NanoDCAL license and the NeutralAtomDatabase (see Licenses and solver executables).

2.2. Where LatticeMind comes from

LatticeMind is published on the Nanoacademic package index. Its dependencies come from public PyPI, so pip needs both indexes on every install and upgrade:

--index-url https://pypi.nanoacademic.ca/simple/
--extra-index-url https://pypi.org/simple/

The commands below include them. No credentials are required.

2.3. Option A — Conda / Miniconda (recommended)

Conda gives you a clean, reproducible Python environment on any platform.

conda create -n latticemind python=3.12
conda activate latticemind
python -m pip install \
  --index-url https://pypi.nanoacademic.ca/simple/ \
  --extra-index-url https://pypi.org/simple/ \
  lattice-mind

2.4. Option B — Python virtual environment

If you do not use Conda, a standard virtual environment works just as well.

Windows (PowerShell):

python -m venv .venv
.\.venv\Scripts\Activate.ps1
python -m pip install --upgrade pip
python -m pip install `
  --index-url https://pypi.nanoacademic.ca/simple/ `
  --extra-index-url https://pypi.org/simple/ `
  lattice-mind

macOS / Linux:

python3 -m venv .venv
source .venv/bin/activate
python -m pip install --upgrade pip
python -m pip install \
  --index-url https://pypi.nanoacademic.ca/simple/ \
  --extra-index-url https://pypi.org/simple/ \
  lattice-mind

2.5. Option C — from a wheel file

When Nanoacademic supplies a wheel directly, or the machine cannot reach the index, install the file instead. The offline package comes as one archive per platform (Linux, Windows or macOS), each holding the wheels for every supported Python version together with the documentation. The wheels are compiled and platform-specific, so both the cp310–cp314 tag and the platform tag must match your interpreter:

python -m pip install path/to/lattice_mind-<version>-<matching-tags>.whl

2.6. Verify the installation

After installation, these commands should be available on your PATH:

Command	Purpose
`latticemind`	Start a research conversation in the current folder.
`latticemind setup`	Choose the model and locate RESCU; `--show` prints what is ready.
`latticemind quick`	Start the one-shot pipeline.
`latticemind-web`	Open the web console in your browser.

Confirm which version the intended environment contains:

latticemind --version

Then install the LatticeMind license as described in The LatticeMind license before starting latticemind or latticemind-web.

2.7. What gets installed

The wheel installs everything LatticeMind needs to build, validate, and analyse calculations, including the post-processing libraries h5py (RESCU .mat and .h5 outputs), scipy (legacy MATLAB files and numerical helpers), and matplotlib (Python plots when RESCU’s own plotting does not cover a requested figure). It also bundles:

the RESCU documentation used for retrieval — built in and offline, no setup required (see Licenses and solver executables);
a runtime pseudopotential archive for turnkey RESCU execution;
the Web UI assets.

Note

You do not need MATLAB, a compiler, or any build tools to use LatticeMind. Those are only relevant if you build the package from source, which is a developer task.

Run latticemind setup to choose a model and locate RESCU in two questions, or continue to Connecting your AI model to connect your AI model by hand, then Licenses and solver executables to install the LatticeMind license and configure RESCU execution. If you plan to run NanoDCAL, that page also explains the NanoDCAL license and NeutralAtomDatabase setup.