How-to guides

Each guide starts from a specific task and uses the stable public API. Examples are independent between pages; run a page's Python blocks in order.

• Define signatures and layouts
• Convert between compact and canonical storage
• Work with leading dimensions and broadcasting
• Use pairwise operations
• Plan and reuse products
• Use unary and grade operations
• Compute norms, energies, and signed forms correctly
• Reflect vectors and multivectors
• Project, reject, and invert safely
• Materialize a bivector exponential
• Apply a bivector-generated action without materializing the rotor
• Compose explicit rotor operations
• Put planned operations inside nn.Module
• Compile Clifford computations
• Optimize Clifford parameters
• Set resource limits
• Understand and control planned execution
• Provide a custom executor
• Analyze spectra, commutators, and transformations
• Use experimental analysis