LatticeMind
Contents
Introduction
How LatticeMind works
Supported solvers
Key features
Who this manual is for
Installation
1. Quick start
1.1. Where do results go?
1.2. What next?
2. Detailed installation
2.1. Requirements
2.2. Option A — Conda / Miniconda (recommended)
2.3. Option B — Python virtual environment
2.4. Verify the installation
2.5. What gets installed
3. Connecting your AI model
3.1. What LatticeMind sends
3.2. Where LatticeMind looks for keys
3.3. Choosing a provider
3.4. Choosing models
3.5. Optional: Materials Project
3.6. Verifying it works
4. Licenses and solver executables
4.1. The RESCU license used at LatticeMind startup
4.2. Running RESCU calculations
4.3. Running NanoDCAL calculations
4.4. The built-in RESCU documentation
4.5. Do I have everything?
5. Cloud & headless deployment
5.1. Running the Web UI on a server
5.2. Cloud checklist
5.3. qBraid / Jupyter proxy quickstart
5.4. Working inside an existing folder
5.5. Choosing where calculations run
5.6. A safe first cloud calculation
6. Updating & troubleshooting
6.1. Updating LatticeMind
6.2. Troubleshooting
6.3. Before launching expensive jobs
6.4. Getting help
Getting started
1. The three interfaces
1.1. Starting LatticeMind
1.2. The Dashboard
1.3. The Web UI
1.4. The CLI
1.5. Which should I use?
2. Using LatticeMind from Codex, Claude, and shell agents
2.1. Let an agent discover the installation
2.2. Search LatticeMind and solver documentation
2.3. Inspect an existing project without changing it
2.4. Run one prompt without opening an interactive interface
2.5. Licenses and provider settings for one-shot work
2.6. Read the JSON response and exit code
3. Your first project
3.1. Step 1 — Start LatticeMind
3.2. Step 2 — Describe what you want
3.3. Step 3 — Look before you leap
3.4. Step 4 — Run it
3.5. Step 5 — See your results
3.6. What next?
4. How LatticeMind works
4.1. The pipeline
4.2. Self-healing and human-in-the-loop
4.3. One project, one solver
4.4. What you can rely on
5. Projects & the workspace
5.1. Where projects are stored
5.2. Managed vs. attached projects
5.3. Anatomy of a project folder
5.4. Resuming and organizing work
6. Interactive commands
6.1. The essentials
6.2. Working with projects
6.3. Controlling execution
6.4. Checking progress & results
6.5. Steering the plan
6.6. Where to go deeper
7. The Web UI tour
7.1. The dashboard at a glance
7.2. Launching a run
7.3. Settings
7.4. Execution controls
7.5. Watching a run
7.6. Recent projects & folders
7.7. Results
Basic tutorials
1. Silicon: SCF, band structure & DOS
1.1. The prompt
1.2. What LatticeMind does
1.3. The generated SCF input
1.4. Running it
1.5. The results
1.6. Inspecting the project
1.7. What you learned
2. Ferromagnetic nickel: a two-step spin DOS
2.1. The prompt
2.2. What LatticeMind does
2.3. The result
2.4. What you learned
3. Building alloys & supercells (SiGe)
3.1. The prompt
3.2. What LatticeMind does
3.3. Varying the alloy
3.4. What you learned
Advanced tutorials
1. Running on an HPC cluster
1.1. Before you start
1.2. Step 1 — Define the cluster profile
1.3. Step 2 — Select and test it
1.4. Step 3 — Build, preview, and submit
1.5. Step 4 — Track and fetch
1.6. Recovering an interrupted job
1.7. Safety and cost
2. NanoDCAL: electronic structure and transport (experimental)
2.1. Silicon band structure
2.2. Validate the band-gap interpretation
2.3. Spin-resolved transmission channels
2.4. How to read and validate the channel map
2.5. Going further: quantum transport
2.6. Three useful follow-up checks
2.7. What you learned
Reference
1. Settings & configuration reference
1.1. Where settings come from
1.2. AI provider & models
1.3. Reports & output location
1.4. Web UI
1.5. Solver commands, license & pseudopotentials
1.6. Parallel execution
1.7. Remote / HPC execution
1.8. Documentation retrieval
1.9. Structure sourcing
1.10. A complete
.env
example
2. Command-line & interactive commands
2.1. Console commands
2.2. Shell discovery and one-shot commands
2.3. Releasing a long calculation
2.4.
latticemind-web
options
2.5. Remote utility subcommands
2.6. Interactive command palette
2.7. Environment variables
3. Remote & HPC execution
3.1. Overview
3.2. Step 1 — Confirm plain SSH works
3.3. Step 2 — Describe the target
3.4. Step 3 — Select and test the profile
3.5. Step 4 — Launch and track jobs
3.6. Serial-only guardrails
3.7. Runtime profiles (deployment context)
4. Solvers & calculation types
4.1. RESCU calculation types
4.2. How the solver is chosen
4.3. Choosing between solvers
5. NanoDCAL support (experimental)
5.1. Configure NanoDCAL before the first run
5.2. Supported and tested calculation types
5.3. Publish-ready figures and numeric summaries
5.4. Execution options
Contact Us
LatticeMind
Index
Index