Autonomous math → ML research

From mathematics papers
to tested ML ideas.

This pipeline reads new arXiv mathematics papers, turns them into concrete ideas for machine learning, and then actually tests them: first on a toy mechanism check, then on a real benchmark against a tuned baseline with 8 paired seeds. Verdicts are computed by deterministic code — not by the language model.

Papers analyzed
13457
Ideas extracted
3050
Experiments run
1427
Mechanism confirmed
701
Beat the baseline
178

Latest verified ideas

All verified →
✓✓ Beats tuned baseline 2026

Differentiable Physics-Equilibrium Projection

Use a neural network to predict an operating point or latent state, then pass it through a sparse differentiable implicit layer that solves governing nonlinear equilibrium equations. This replaces soft physics penalties with an exact or tightly solved equality projection and can be combined with primal-dual inequality handling and deterministic restoration.

Useful9/10
Difficulty7/10
Novelty5/10
Paper: UNION: A Unified AC-OPF Framework for Topology-Varying Real-Time Grid Operation arXiv:2608.25784
✓✓ Beats tuned baseline 2026

Local Characteristic Residual Gating

Transform local neural residuals into the Ripa model's characteristic coordinates before spatial aggregation, apply a mode-dependent gate based on neighboring characteristic jumps, and transform back. This lets the model damp oscillatory acoustic or equilibrium-mode corrections near discontinuities without globally smoothing every feature.

Useful6/10
Difficulty6/10
Novelty6/10
Paper: Fifth-Order Well-Balanced Path-Conservative A-WENO Scheme for the Ripa Model arXiv:2607.09293
✓✓ Beats tuned baseline 2026

Epoch-Frozen Masked Low-Rank Candidate Encoder

Add an online low-rank reconstruction layer before a recommender or ranking MLP. It estimates a latent action-feature subspace from masked candidate vectors, freezes that subspace for an epoch, reconstructs each partially observed candidate in the latent coordinates, and feeds only those coordinates to the predictor. The method is most promising when the ambient candidate dimension d is large but the effective rank m is small and missingness is not too severe.

Useful6/10
Difficulty5/10
Novelty7/10
Paper: Stochastic Linear Bandits with Partially Observed Actions arXiv:2607.08971
✓✓ Beats tuned baseline 2026

Invariant nonstandard residual blocks

Replace the usual explicit residual update with a nonstandard general-linear block containing several internal feature stages. The effective step is a positive denominator function rather than the raw depth step, allowing the block to take large nominal steps while damping the update and preserving bounded activations. This is most promising for deep residual MLPs, neural ODE discretizations, and state-space sequence models where exploding hidden states limit usable depth.

Useful6/10
Difficulty6/10
Novelty6/10
Paper: Some properties of high-order nonstandard multistep multistage methods arXiv:2607.08694
✓✓ Beats tuned baseline 2026

Contractive projected residual dynamics

Build a recurrent or continuous-depth block from a dissipative vector field and project every state derivative onto the tangent cone of a closed convex hidden-state set. Unlike ordinary clipping, tangent-cone projection removes only the outward component at the boundary and preserves admissible motion. Under the paper's maximal-dissipativity result, the continuous flow is nonexpansive in its initial state.

Useful6/10
Difficulty5/10
Novelty5/10
Paper: Projected incrementally scattering passive systems on closed convex sets arXiv:2607.08301
✓✓ Beats tuned baseline 2026

Finite-group relative message passing

Use the quotient group's generator classes as a finite relation vocabulary and tie message functions by group displacement instead of by individual graph edges. This creates a compact, exactly consistent relation-aware GNN that can recognize repeated local structure and transfer parameters across graph instances sharing the same Cayley geometry.

Useful6/10
Difficulty5/10
Novelty8/10
Paper: Minimal Isometric Embeddings of Graphs into Cayley Graphs of Finite Abelian Groups arXiv:2607.07920
✓✓ Beats tuned baseline 2026

Rank-Safe Variable-Projection Gauss-Newton

Separate a neural network into nonlinear hidden parameters and a linear output layer. Solve the output layer exactly by least squares, then update hidden parameters with a truncated-pseudoinverse Gauss-Newton step that discards numerically singular directions.

Useful6/10
Difficulty6/10
Novelty5/10
Paper: Structure-Guided Gauss-Newton Method: Linear Advection-Reaction Equation arXiv:2607.07506
✓✓ Beats tuned baseline 2026

Reversible Mealy Token Mixer

Replace a recurrent token-mixing operation with a finite-state carrier scan over binary or quantized token features. The local transition table is constrained to conserve a scalar token weight and to be bijective, making the mixer reversible, constant-memory, and less prone to activation drift than a generic recurrent layer.

Useful6/10
Difficulty5/10
Novelty8/10
Paper: Invariant Measures for Soliton Systems Generated by Mealy Automata arXiv:2607.06942

Highest usefulness

All ideas →
✓✓ Beats tuned baseline 2026

Differentiable Physics-Equilibrium Projection

Use a neural network to predict an operating point or latent state, then pass it through a sparse differentiable implicit layer that solves governing nonlinear equilibrium equations. This replaces soft physics penalties with an exact or tightly solved equality projection and can be combined with primal-dual inequality handling and deterministic restoration.

Useful9/10
Difficulty7/10
Novelty5/10
Paper: UNION: A Unified AC-OPF Framework for Topology-Varying Real-Time Grid Operation arXiv:2608.25784
Mechanism confirmed, baseline not beaten 2026

Robust CBF Safety Layer for Neural Policies

Attach a robust high-order control-barrier-function safety layer after a neural policy for a learned or known control-affine plant. The network proposes a nominal action, while a small online projection modifies it only enough to satisfy input bounds and barrier inequalities under an estimated disturbance and an explicit transient error bound.

Useful9/10
Difficulty5/10
Novelty5/10
Paper: Robust Safety Filtering for Input-Constrained Underactuated Linear Systems arXiv:2608.10872
Failed on benchmark 2026

Power-Preserving Formation GNN

Replace an unconstrained graph-message-passing block with a port-Hamiltonian layer whose edge interactions are generated by a skew-symmetric formation-matrix coupling and whose node damping is positive semidefinite. The layer can model relative graph structure while preventing unforced hidden-state energy growth, reducing exploding activations and oversmoothing caused by arbitrary repeated propagation.

Useful8/10
Difficulty5/10
Novelty6/10
Paper: Formation Matrix and Energy-based Control of Multi-Agent Systems arXiv:2609.04158
Mechanism failed 2026

Behavior-Gap Clustered Neural Controllers

Cluster recurrent modules or MoE experts by the geometry of their observed finite-horizon input-output behaviors rather than by parameter distance. Train one shared optimizer/controller or low-rank adapter per cluster while retaining module-specific parameters and routing. This should reduce control and optimizer overhead without merging modules whose temporal responses are dynamically incompatible.

Useful8/10
Difficulty5/10
Novelty8/10
Paper: Data-Based Clustering and Control of Similar Biological Systems arXiv:2609.03921
Mechanism failed 2026

Hamiltonian Horizon-Critical Optimizer

Replace a fixed first-order parameter update by a finite-horizon controlled local model for each important curvature mode of the network. The optimizer computes the Hamiltonian flow and its Riccati feedback gain; if the chosen horizon approaches a conjugate point, it shortens the horizon or increases control cost before the gain becomes singular. This converts the paper's finite-time transition into a measurable trust-region and scheduling mechanism for neural training.

Useful8/10
Difficulty6/10
Novelty8/10
Paper: Equivalence classes of finite-time transitions in optimal control and non-equilibrium relaxation arXiv:2609.03862
Mechanism confirmed, baseline not beaten 2026

Equivariant Shared-Mechanism World Model

Use the paper's families of local graph embeddings to identify repeated occurrences of the same causal substructure across time steps, environments, or entities. Feed every aligned occurrence through one shared transition mechanism and impose an explicit equivariance penalty under the symmetry group acting on occurrence indices, rather than learning an independent predictor for every context.

Useful8/10
Difficulty5/10
Novelty5/10
Paper: Symmetries and Causality: Causal Effect Identification Beyond IID Data arXiv:2609.03697

How verification works

1

Mechanism check

A cheap toy experiment tests whether the claimed mathematical phenomenon actually reproduces and yields the predicted quantitative effect. Only ideas that pass go further.

2

Benchmark vs. tuned baseline

The idea is implemented on a real (small) neural network task and compared against a tuned baseline — same training budget, 8 paired seeds, significance via a permutation test. The smallest detectable effect is reported too.

Deterministic verdict

The verdict — helps, hurts or no effect — is computed by test code from paired-seed statistics. Failures and non-results are published as well; that is the point.

AI-generated research hypotheses, automatically tested. Not peer-reviewed.