9 projects
omnibias-geometry
Differential geometry on manifolds for omnibias: metric tensor, Christoffel symbols, covariant derivative, the Laplace-Beltrami operator, Riemann/Ricci/scalar curvature, geodesics, exterior calculus (d, wedge, Hodge star, codifferential), and the pullback metric of a learned chart (g = J^T h J) for learned manifolds. Built on the omnibias-fields substrate with cross-backend (torch + jax) parity.
omnibias-pinn
Physics-informed neural networks with closed-form n-th derivative operators on top of omnibias-core. Cross-backend (torch + jax) typed fields, ops, hard-conservation cages, prebuilt PDE residuals, and diagnostics.
omnibias-fields
Backend-agnostic field substrate for omnibias: the FieldState value object, the attribute-DSL views, the lazy sigma^(n) cache, and the cross-backend (torch + jax) closed-form differential-operator surface (gradient, divergence, curl, laplacian, hessian, jacobian, integration, inner products, Sobolev norms, tensor divergence, Wirtinger).
omnibias-keras
Keras 3 unified backend for omnibias: closed-form n-th derivative scalar operators (OMBU), operator-typed blocks, and drop-in cmbDense / cmbConv layers that run on TensorFlow, JAX, or PyTorch via keras.ops.
omnibias-ferminet
FermiNet bridge for omnibias: folx-compatible Laplacian adapters, restricted-depth Tier-2 FermiNet ansatz with closed-form Laplacian, and multiblock primitives for analytic nuclear Hessian / Born-Oppenheimer derivative work.
omnibias-jax
JAX backend for omnibias: closed-form n-th derivative activation kernels (sigmoid via Eulerian polynomials, tanh via Legendre, Gaussian via Hermite), neural-field Laplacian / Hessian primitives, and Born-Oppenheimer derivative tools for variational quantum Monte Carlo.
omnibias-torch
PyTorch backend for omnibias: trainable scalar operators (OMBU), operator-typed blocks, closed-form activation derivative kernels, and reference PINN / CmbNet / CvxLayer architectures.
omnibias-core
Numerically-stable closed-form n-th derivative forward-pass framework: pure-Python core with polynomial coefficient generators (Eulerian / Legendre / Hermite) and the backend-agnostic ActivationSpec protocol.
vardang-distributions
Gaussian distributions
