1. Quick start

This is the shortest path from a fresh machine to your first result.

  1. Install LatticeMind from the Nanoacademic package index, on Python 3.10 through 3.14. Its dependencies come from public PyPI, so pip needs both indexes:

    python -m pip install \
  --index-url https://pypi.nanoacademic.ca/simple/ \
  --extra-index-url https://pypi.org/simple/ \
  lattice-mind

  2. Install your LatticeMind license. This is LatticeMind’s own license, not a RESCU one. Put exactly one *.lic or *.license file in ~/.nanoacademic/LATTICEMIND. On Windows that expands to C:\Users\<username>\.nanoacademic\LATTICEMIND. For any other location, set:

    LATTICEMIND_LICENSE_FILE=/full/path/to/latticemind.lic

  3. Answer two questions. Which model to use, and where RESCU is:

    latticemind setup

    OpenAI is the default and asks for its API key; Claude and Gemini API keys are offered too, and /model inside a conversation lists every provider, including Qwen and a local Ollama. RESCU is offered where LatticeMind finds it. The answers are saved to ~/.latticemind/.env for the terminal and the browser alike, and a bare latticemind with nothing configured asks the same two questions itself. Setting the variables by hand still works — see Connecting your AI model.

  4. Make a folder for the work and start in it. The folder you start in is the project:

    mkdir silicon-bands
cd silicon-bands
latticemind

Then type what you want:

Do a simple SCF and band structure calculation for silicon.

LatticeMind builds the structure, writes the RESCU inputs and validates them, then shows you exactly what it would run and where, and asks:

1. yes   2. yes, always   3. no

Answer with a number or in words — yes, go ahead, run it, or no, cancel. Nothing reaches the solver until you say so. If you only want the inputs for now, ask for that: “prepare the inputs but do not run anything.”

While it works, a single status line keeps you posted, and you can type a correction at any time. ctrl-c offers to stop, to leave the work running, or to keep watching.

Tip

Type /examples inside the app for ready-to-run prompts, or /help for the short list of commands. Everything else is plain language.

Prefer a browser? Launch the web interface instead:

latticemind-web

It opens at http://127.0.0.1:7865 by default.

Note

Install the LatticeMind license from step 2 before starting LatticeMind. To run a project, also configure the selected solver executable and its own license on the computer that executes it: RESCU uses a RESCU license, and NanoDCAL uses a NanoDCAL license plus the NeutralAtomDatabase. See Licenses and solver executables.

1.1. Where do results go?

Into the folder you started in. LatticeMind keeps a hidden .latticemind/ directory there for its records, and writes the solver inputs, pseudopotentials/, results/ and the figures beside it. See Projects & the workspace for the layout.

For scripted or automated use, you can send that output somewhere else:

export LATTICEMIND_ARTIFACTS_DIR="/path/to/outputs"

1.2. Coming back

Run latticemind in the same folder again and the conversation resumes, with every result still in place. /new starts a fresh conversation there.

1.3. What next?