Architecture ideas

Research ideas extracted from mathematics papers, categorized as Architecture.

Unverified 2026

2-System Greedy Token Selection

Replace top-k token pruning by greedy maximization of a diversity-aware monotone submodular utility under a spacing or coverage constraint. The selector repeatedly chooses the feasible token with the largest marginal utility, avoiding the redundant-token failure mode of independent score ranking while inheriting a constant-factor approximation guarantee under the stated 2-system abstraction.

Useful6/10
Difficulty5/10
Novelty6/10
Paper: Efficient Discrete Position Design for Movable Antenna Systems: Low Complexity and Robustness arXiv:2608.07413
Unverified 2026

Maximal multiscale differential block

Replace a conventional feature-pyramid sum by a bounded multiscale differential transform. At each scale, subtract a blockwise conditional expectation from a local average, then combine these residuals with bounded coefficients. Add a penalty on the largest interval response so that contributions from adjacent scales cannot accumulate destructively or explosively.

Useful6/10
Difficulty5/10
Novelty6/10
Paper: Noncommutative maximal differential transforms associated to averaging operators arXiv:2608.07300
Unverified 2026

Near-Optimal Lanczos Spectral Layer

Implement f(A)b inside a neural network with a short Lanczos recurrence instead of an eigendecomposition or dense matrix-function operation. Use an SPD operator A such as a regularized graph Laplacian or feature covariance matrix, and choose the number of iterations by monitoring successive approximations. For Stieltjes functions, Lanczos is guaranteed to be close to the best vector in the same Krylov subspace.

Useful6/10
Difficulty5/10
Novelty6/10
Paper: Optimal near-optimality bounds for the Lanczos method for matrix functions arXiv:2608.07160
Unverified 2026

Monotone spectral activation

Replace an unconstrained matrix nonlinearity on small symmetric feature blocks with the isotropic spectral lift of a permutation-equivariant monotone map on eigenvalues. The layer remains orthogonally equivariant, while the paper's equivalence transfers a scalar inner-product monotonicity certificate from eigenvalue space to the full matrix space.

Useful6/10
Difficulty5/10
Novelty6/10
Paper: Monotonicity of isotropic tensor functions on the set of symmetric matrices: completing Rodney Hill's generalization of the Chandler Davis convexity theorem arXiv:2608.07087
Unverified 2026

Geodesic Low-Rank Latent Bottleneck

Replace a Euclidean low-rank latent decoder with a geodesic factor decoder on a Riemannian manifold. A learned location α provides the component center, a small set of tangent loading vectors V captures anisotropic variation, and latent coefficients z generate curved manifold-valued features through the exponential map. Multiple such decoders can form a mixture-of-geodesic-experts layer for multimodal representations.

Useful6/10
Difficulty5/10
Novelty7/10
Paper: Mixture of Geodesic Factor Analyzers on Riemannian Homogeneous Spaces arXiv:2608.06971
Unverified 2026

Wasserstein-Gated Cached Equilibrium Adapter

Add a small constrained equilibrium layer whose response depends on the current neural state and recent exogenous history, then cache responses keyed by a learned history embedding. For a new history, reuse a cached response only when an empirical Wasserstein distance to the cached history is below a threshold; otherwise run a few inner optimization iterations. The paper's local Holder and trajectory-stability results motivate graceful degradation rather than catastrophic errors for nearby…

Useful6/10
Difficulty5/10
Novelty6/10
Paper: Stability of Differential Stochastic Variational Inequalities with History-Dependent Responses and Transfer Learning arXiv:2608.06923
Unverified 2026

Submodular Max-Delay Batch Timer

Replace a fixed maximum-wait timeout in dynamic neural inference batching with a service-aware timer. The timer estimates the marginal reduction in computation or communication cost from adding another request and delays service only while that saving justifies the additional oldest-request delay.

Useful6/10
Difficulty4/10
Novelty6/10
Paper: Online Multi-Level Aggregation with Per-Batch Maximum Delay arXiv:2608.06796
Unverified 2026

Buffered partition-of-unity gating

Replace an unconstrained spatial gate or interpolation kernel by a compactly supported function whose translates under a lattice exactly sum to one. Impose zero products between translates under a second lattice, so active gates do not collide; thresholding a positive superlevel set then provides a nonzero separation margin and predictable sparse computation.

Useful6/10
Difficulty5/10
Novelty6/10
Paper: Tilings, packings, and the existence of Schwartz-class Gabor windows arXiv:2608.06679
Unverified 2026

Random-Batch Interaction Layer

Replace an all-pairs interaction or attention-like message field with an unbiased estimator obtained from a uniformly random partition into batches. Keep the partition fixed for \(\tau\) integration steps, then redraw it; the paper's error law predicts that too-long reuse causes a quadratic-in-\(\tau\) error accumulation, while finite-population error decreases as \(N^{-1}\).

Useful6/10
Difficulty4/10
Novelty5/10
Paper: Mean field error estimate of the random batch method for vortex blob dynamics for the 2D Navier--Stokes Equation arXiv:2608.06533
Unverified 2026

KPZ Directed-Polymer Attention

Replace independent Gaussian attention noise or unconstrained token routing with a directed-polymer path distribution over positions and layers. The router aggregates exponentially many monotone paths through temporally correlated random edge scores, producing heavy-tailed but spatially coherent routing and preventing attention from collapsing onto a single token. The paper's t^{2/3} wandering and t^{1/3} free-energy fluctuations become measurable diagnostics and tunable targets rather than…

Useful6/10
Difficulty5/10
Novelty7/10
Paper: KPZ Superdiffusion of Local Correlators in Diffusive Random Quantum Circuits arXiv:2608.06459
Unverified 2026

Invertible Fourier Surrogate for Periodic Sequence Modeling

Represent periodic input-output behavior using a compact real vector of Fourier coefficients and learn an invertible neural map from input coefficients to output coefficients. Inference then obtains the input representation for a desired periodic output by a single inverse pass instead of iterative optimization through a nonlinear forward model, while the Fourier representation reduces sequence dimensionality when high-rate signals are spectrally sparse.

Useful6/10
Difficulty6/10
Novelty4/10
Paper: Certified Feedforward Tracking for Unknown Nonlinear Systems via Invertible Neural Networks arXiv:2608.06419
Unverified 2026

Intersection Euler Interaction Token

Compute a compact multiscale interaction signature between colored point clouds and append it to a point-cloud or multimodal neural network as a learned interaction token. The signature captures separated, overlapping, and higher-order enclosing configurations while remaining invariant to rigid transformations.

Useful6/10
Difficulty5/10
Novelty7/10
Paper: The Intersection Euler Characteristic Profile: Euler Calculus and Stability for Topological Interaction of Ball Unions arXiv:2608.06180
Unverified 2026

LKJ Covariance for Variational Adapter Blocks

Use an LKJ correlation factor as the correlation component of a variational posterior over a compact adapter, LoRA factor, or Bayesian neural-network parameter block. The model learns marginal scales separately while the correlation matrix remains automatically positive semidefinite and unit-diagonal, avoiding unconstrained covariance matrices, invalid correlations, and fragile covariance decompositions.

Useful6/10
Difficulty6/10
Novelty6/10
Paper: Bartlett Couplings of the Onion and Vine LKJ Samplers arXiv:2608.06116
Unverified 2026

Half-Idleness Curvature Attention

Compute each graph edge's Lin–Lu–Yau curvature exactly from one p=1/2 Wasserstein problem, then use the resulting scalar as an edge bias or multiplicative gate in graph attention. Positive-curvature edges receive stronger message exchange while negatively curved edges are attenuated, giving the network a geometry-derived inductive bias rather than requiring the model to learn all edge importance from scratch.

Useful6/10
Difficulty5/10
Novelty5/10
Paper: Equivalence of Lin--Lu--Yau curvature and 1/2-Ollivier curvature on weighted graphs arXiv:2608.05939
Unverified 2026

Power-Law Volterra Memory

Add a causal memory branch whose weights are generated by the paper's power-type Volterra kernel rather than learned independently at every lag. Learn or softly constrain the exponents so the model can select rough short-memory behavior or smoother long-memory behavior while using only a few parameters. The branch can be implemented as a truncated causal convolution, a multiresolution approximation, or a recurrent state-space realization.

Useful6/10
Difficulty6/10
Novelty6/10
Paper: Small ball probabilities and Chung's law of the iterated logarithm for Gaussian Volterra processes with power-type kernels arXiv:2608.05679
Unverified 2026

Braess-aware graph rewiring

Use Kemeny’s constant as a diffusion-quality gate when adding shortcut edges or cliques to a graph used by a GNN. Candidate augmentations are accepted only when they reduce estimated average hitting time, preventing rewiring operations that superficially shorten paths but make the random walk mix more slowly.

Useful6/10
Difficulty5/10
Novelty7/10
Paper: Kemeny's constant and Braess cliques in graphs arXiv:2608.04150
Unverified 2026

Contour-reduced parametric SSM

Replace repeated full-dimensional matrix-exponential or ODE solves in a conditioned continuous-time state-space layer with contour quadrature evaluated in a projection basis. The same reduced basis and contour nodes can serve many conditioning vectors, while shifted reduced resolvents provide a stable and differentiable approximation over a prescribed time window.

Useful6/10
Difficulty6/10
Novelty7/10
Paper: Contour integral methods and model order reduction for parametric linear control systems arXiv:2608.05363
Unverified 2026

Hitting-Time Adaptive Transformer Depth

Use attention-graph hitting times to identify tokens whose information has not mixed through the network, then route only those tokens through additional Transformer blocks. Tokens with fast reachability exit early, while slow or isolated tokens receive more computation.

Useful6/10
Difficulty6/10
Novelty7/10
Paper: Identifying slow relaxation in many-body quantum systems through state-graph geometry and state-graph heterogeneity arXiv:2608.05298
Unverified 2026

KAM-Stabilized Quasiperiodic Recurrent Memory

Construct a recurrent module with a phase variable and a transverse memory coordinate modeled on a perturbed twist map. Train the transverse state to lie on an invariant graph over the phase, while the phase follows an approximately irrational rigid rotation. A KAM-inspired graph correction and residual penalty should reduce long-horizon drift in recurrent prediction.

Useful6/10
Difficulty6/10
Novelty8/10
Paper: Persistence of invariant graphs for twist maps under analytic perturbations arXiv:2608.05239
Unverified 2026

Chain-Current Latent States

Replace an unconstrained geometric latent vector with a state consisting of discrete chain coefficients, a continuous current, and an integral-current curvature. Neural updates are projected through the differential-homology boundary operator, so learned states remain compatible with conservation and boundary structure on meshes or point clouds.

Useful6/10
Difficulty5/10
Novelty7/10
Paper: Differential Homology arXiv:2608.05048
Unverified 2026

Automaton-Gated Objective Switching

Augment a neural policy with deterministic DFA states for the task objective and safety constraint, then select among objective-specific policy heads using those states. Before either target is reached, execute a mixed policy; after one target is reached, switch permanently to the policy specialized for the remaining target.

Useful6/10
Difficulty4/10
Novelty5/10
Paper: Optimal Constrained sc-LTL Planning in MDPs via Switching Policies arXiv:2608.05021
Unverified 2026

Spectrahedral Obedience Layer

Insert a differentiable equilibrium layer between a neural payoff/state encoder and the final action recommendations. The layer parameterizes a joint recommendation object and enforces all unilateral-deviation inequalities as positive-semidefinite constraints, preventing the network from producing recommendations that agents have a strict incentive to disobey. A quantum-inspired density-matrix parameterization can model correlated recommendations using PSD matrices rather than factorized action…

Useful6/10
Difficulty6/10
Novelty7/10
Paper: Quantum Bayes Correlated Equilibrium and the Comparison of Quantum Information Structures in Games arXiv:2608.04973
Unverified 2026

PSD-Safe Bernstein Distance Kernel

Replace an unconstrained learnable distance-bias function in a graph neural network or distance-aware attention layer by a Bernstein approximation of a positive-definite circular kernel. The resulting kernel is a degree-n polynomial in normalized distance while preserving positive semidefiniteness of every finite Gram matrix on the circle, preventing training from producing an invalid covariance-like similarity structure.

Useful6/10
Difficulty4/10
Novelty7/10
Paper: Preservation of Positive-Definiteness by Bernstein Operators on the Circle arXiv:2608.04836
Unverified 2026

Sparse Weighing Mixer

Use enumerated weighing matrices as sparse orthogonal channel-mixing operators inside MLPs or residual blocks. Their ternary entries reduce multiplication to signed additions, while exact orthogonality prevents amplification or attenuation of feature norms; a trainable fixed-support version can recover expressivity without giving up computational sparsity.

Useful6/10
Difficulty5/10
Novelty6/10
Paper: Constructing Large Orthogonal Minimally Aliased Response Surface Designs Through Enumeration and Combination of Weighing Designs arXiv:2608.04814