Tardigrade Innovation

Research

Math & physics work

Core research threads: gravitational-wave quantum classification via Clifford geometric algebra, the SciLM multi-domain foundation model, and the physics-constrained ML tooling that connects them.

Gravitational-Wave Quantum Classifier

Preprint pending

Cl(3,0) geometric-algebra encoding + tree tensor network circuit

Encodes gravitational-wave polarizations (h+, h×) as a Cl(3,0) multivector, with the phase residual carried as the Hodge-dual third component, then classifies echo / no-echo ringdown signatures — a quantum-gravity proxy — with a Spin(3)-equivariant tree tensor network circuit. Targeting IBM ibm_kingston hardware.

Clifford algebraTensor networksQiskitQuantum gravity
Full spec →Source

SciLM

In progress

Multi-domain scientific foundation model, swarm-of-primitives architecture

A physics-constrained foundation model spanning gravitational-wave, condensed matter, and materials domains. Replaces mixture-of-experts routing with six composable mathematical primitives — conserved-current, response/Green's function, extremal/variational, spectral/eigenstructure, group-equivariant, and renormalization — that assemble into domain experts. Phase-1 falsification target: GW quasinormal-mode pole recovery and cascade damage scaling. Built in collaboration with Michael Lawler (Cornell/Binghamton).

PyTorchPhysics-informed MLJAX
Full spec →Source

Plasma Control: PINN + RL

Live

PINN-constrained PPO control agent for BOUT++ plasma simulation

A reinforcement-learning control agent (PPO) constrained by a physics-informed neural network, with Clifford geometric-algebra layers and a chalcogenide memristor reservoir computing backend, controlling BOUT++ plasma simulations in real time. Source published on GitHub; PyPI packaging in progress.

PPOPINNClifford algebraReservoir computingBOUT++
Full spec →Source

Amplituhedron Solver

In progress

Positive-geometry approach to scattering amplitude computation

A solver built around the amplituhedron formalism, computing scattering amplitudes from positive-geometry structure directly rather than summing Feynman diagrams. Shares a throughline with the Clifford/geometric-algebra work: physics encoded as geometric structure rather than as a sum over terms.

Positive geometryScattering amplitudesComputational algebra
Full spec →Source

Non-Euclidean Playground

In progress

Interactive explorer for hyperbolic and spherical geometry

An interactive visualization tool for building intuition in hyperbolic and spherical geometry — tessellations, geodesics, and curvature made explorable rather than only equational.

WebGLDifferential geometryInteractive visualization
Full spec →Source

Geometric AI Agents

In progress

Agent architectures with geometric-algebra state representations

Exploratory work on agent architectures that carry state as geometric-algebra multivectors rather than flat vectors, extending the same Clifford-algebra approach used in the GW classifier and plasma control agent into general agent design.

Clifford algebraAgent architecturesReinforcement learning
Full spec →Source

Surrogate Solvers

In progress

Fast learned surrogates for expensive physics simulations

Learned surrogate models standing in for expensive numerical solvers (field simulations, plasma dynamics) — trading a controlled amount of accuracy for orders-of-magnitude speedup, with error bounds tracked explicitly rather than ignored.

Surrogate modelingPhysics-informed ML
Full spec →Source

tardigrade-cad

Live

Physics-constrained CAD for magnet and coil design

JAX-based Biot–Savart field solver with optax-driven geometry optimization for designing coils and magnet assemblies against physical field constraints, not just geometric ones.

JAXOptaxBiot–Savart
Full spec →Source