Skip to main content

omnibias.jax

0.5.0rc1Release candidatePython >=3.10Apache-2.0

JAX activation derivatives, functional network jets, direct contractions, and parameter-differentiable numerical models.

import omnibias.jax

Use this backend with JAX arrays and functional parameter trees. The forward derivative calculation is distinct from the JAX transformation you use to differentiate a loss with respect to model parameters. A high spatial order does not by itself require nesting that many reverse-mode spatial transformations.

Enable JAX x64 before constructing arrays if your validation requires float64. Treat compilation and steady-state evaluation as separate benchmark phases. Derivative order, architecture, and other structural arguments may need to be static under jit; follow the selected function's contract rather than assuming every Python option is traceable.

Directional jets, mixed-partial jets, and direct Laplacian contractions have different output layouts. Keep the factorial convention explicit when moving between Taylor coefficients and derivatives. The JAX high-order-residual and Laplacian examples demonstrate value_and_grad through the chosen spatial path. Shared formulas across backends do not imply universal bitwise identity.

Module contract & mathematical conventions

JAX activation derivatives and Taylor jets for physics-informed models.

Shared polynomial coefficients define the derivative tower. Operator blocks, mixed jets, and PINN architectures reuse it without nested coordinate autodiff. Architecture and optimization helpers are available in their own modules.

Shared formulas do not imply universal floating-point identity. Verify dtype, shape, derivative-order support, and the stated numerical assumptions for the path you use.

Imports & aliases​

These names share a definition with another public module. Follow a name for its full contract, parameters, and examples.

Import from this moduleCanonical reference
AdaptivePackBankomnibias.jax.AdaptivePackBank
BankSpecomnibias.core.BankSpec
BirkhoffOMBUomnibias.jax.BirkhoffOMBU
BlockSpecomnibias.core.BlockSpec
DEQConfigomnibias.core.DEQConfig
DEQNotContractiveomnibias.core.DEQNotContractive
DEQResultomnibias.jax.DEQResult
DEQSolverUnknownomnibias.core.DEQSolverUnknown
GradientSecantomnibias.core.GradientSecant
JaxActivationSpecomnibias.jax.JaxActivationSpec
JetLineSearchConfigomnibias.core.JetLineSearchConfig
LineSearchResultomnibias.core.LineSearchResult
WeightLossJetSpecomnibias.core.WeightLossJetSpec
X64_HINTomnibias.jax.X64_HINT
affine_jetomnibias.jax.affine_jet
affine_jet_mvomnibias.jax.affine_jet_mv
antiderivative_jetomnibias.jax.antiderivative_jet
arrangement_w_blockomnibias.core.arrangement_w_block
bank_forwardomnibias.jax.bank_forward
bias_scanomnibias.jax.bias_scan
block_directionomnibias.jax.block_direction
block_exact_searchomnibias.jax.block_exact_search
block_exact_sweepomnibias.jax.block_exact_sweep
compose_jetomnibias.jax.compose_jet
compose_jet_mvomnibias.jax.compose_jet_mv
compose_jet_riccatiomnibias.jax.compose_jet_riccati
coulomb_potentialomnibias.jax.coulomb_potential
deq_du_dWomnibias.jax.deq_du_dW
deq_solveomnibias.jax.deq_solve
deq_vjpomnibias.jax.deq_vjp
derivative_jetomnibias.jax.derivative_jet
equivariant_scan_applyomnibias.jax.equivariant_scan_apply
get_activationomnibias.jax.get_activation
identity_jetomnibias.jax.identity_jet
init_bias_scanomnibias.jax.init_bias_scan
init_multipackomnibias.jax.init_multipack
init_pack_bankomnibias.jax.init_pack_bank
is_registeredomnibias.jax.is_registered
jet_attentionomnibias.jax.jet_attention
jet_expomnibias.jax.jet_exp
jet_gradientomnibias.jax.jet_gradient
jet_hessianomnibias.jax.jet_hessian
jet_line_searchomnibias.jax.jet_line_search
jet_line_search_on_rayomnibias.jax.jet_line_search_on_ray
jet_multiplyomnibias.jax.jet_multiply
jet_partialsomnibias.jax.jet_partials
jet_reciprocalomnibias.jax.jet_reciprocal
jet_softmaxomnibias.jax.jet_softmax
jet_to_toweromnibias.jax.jet_to_tower
last_linear_blockomnibias.core.last_linear_block
layer_jetomnibias.jax.layer_jet
layer_jet_mvomnibias.jax.layer_jet_mv
lhopital_ratioomnibias.jax.lhopital_ratio
limit_of_ratioomnibias.jax.limit_of_ratio
list_activationsomnibias.jax.list_activations
make_bo_forceomnibias.jax.make_bo_force
make_bo_hessianomnibias.jax.make_bo_hessian
make_local_energyomnibias.jax.make_local_energy
mlp_jetomnibias.jax.mlp_jet
mlp_jet_mvomnibias.jax.mlp_jet_mv
multipack_applyomnibias.jax.BirkhoffOMBU
multipack_responseomnibias.jax.multipack_response
neural_field_hessianomnibias.jax.neural_field_hessian
neural_field_laplacianomnibias.jax.neural_field_laplacian
neural_field_valueomnibias.jax.neural_field_value
neural_field_value_and_laplacianomnibias.jax.neural_field_value_and_laplacian
neural_field_value_grad_hessianomnibias.jax.neural_field_value_grad_hessian
neural_field_value_grad_laplacianomnibias.jax.neural_field_value_grad_laplacian
ombu_bias_blockomnibias.core.ombu_bias_block
one_layer_lossomnibias.jax.one_layer_loss
one_layer_loss_gradomnibias.jax.one_layer_loss_grad
one_layer_loss_hessianomnibias.jax.one_layer_loss_hessian
one_layer_loss_jetomnibias.jax.one_layer_loss_jet
one_layer_newton_directionomnibias.jax.one_layer_newton_direction
register_activationomnibias.jax.register_activation
removable_valueomnibias.jax.removable_value
require_x64omnibias.jax.require_x64
scan_responseomnibias.jax.scan_response
soft_argmax_offsetomnibias.jax.soft_argmax_offset
steerable_basisomnibias.jax.steerable_basis
tower_to_jetomnibias.jax.tower_to_jet
vibrational_frequenciesomnibias.jax.vibrational_frequencies
x64_enabledomnibias.jax.x64_enabled