Every idea extracted from recent arXiv mathematics papers — verified and unverified. Click an idea to open its full card; badges show the empirical verdict.
Replace an unconstrained relative-position attention-bias table by a learnable two-dimensional Fourier representation, or regularize the bias toward a small Fourier ratio. The ratio favors coefficient concentration without depending on the absolute scale of the bias, allowing the trained bias to be reconstructed from a small number of dominant frequencies.
Compress a trained graph-attention model's edge-dependent logits or gates in the line-graph Fourier domain. Smooth edge values become concentrated in low-frequency coefficients, allowing low-frequency coefficients to retain more precision while high-frequency residuals use fewer bits or are discarded.
Replace one independently learned vector per graph edge with a truncated spectral expansion on the line graph. The model learns coefficients for low-frequency edge modes and reconstructs edge features before message passing, reducing parameters while imposing an inductive bias that incident edges should have correlated behavior.
Replace a dense learned linear operator on continuous or image features by a truncated Hermite projection expansion whose coefficients are directly regularized in a Schatten-p norm. The layer becomes a structured low-rank operator, while the radial Hermite-Laguerre correspondence provides an analytically tractable parameterization and an exact spectral penalty.
Use the paper's norm-regularized conic dualization to impose PSD or SOS-style certificate constraints during neural-network training without forming Schur-complement or second-order-cone liftings. A neural dynamics model can be trained jointly with a polynomial Lyapunov or energy certificate, while the certificate subproblem is solved through accelerated updates in equality-constraint dual variables.
Replace the linear state transition in a small recurrent or state-space module by a circulant matrix acting on a vector over a finite field. The hidden state then has only finitely many possible values and follows an exactly periodic orbit after at most \(q^n\) states, eliminating numerical drift on modular-counting and symbolic-memory tasks. A learned real-valued encoder and decoder can surround the discrete core, while the transition itself is fixed, searched, or trained with a…
Replace or augment an RNN or state-space model hidden state with coordinates on a bounded 3-step nilpotent group. The first layer stores ordinary features, the second layer stores pairwise commutator memory, and the third layer stores nested commutators that can preserve three-time dependencies invisible to first- and second-order summaries. Layered reduction keeps the state bounded while retaining the algebraic interaction structure.
Replace a dense token-mixing matrix in a sequence model with a fixed or learnable SBP derivative operator D=P^{-1}Q. The discrete integration-by-parts identity makes the interior mixing energy-neutral or boundary-dissipative, reducing exploding activations in deep residual stacks while preserving directional information along the sequence.
Train a square orthogonal neural mixer while maximizing its entrywise fourth-power concentration. When optimization reaches a non-permutation stationary configuration, explicitly test rank-two row or column rotations and take a rotation with positive exact second variation, using the paper's constructive saddle-escape mechanism.
Replace a wide activation vector or spatial feature field by a small set of weighted coordinates that preserves the p-norm of every activation in a learned low-dimensional subspace. Unlike ordinary pruning, the selection objective is uniform over the whole coefficient sphere, so the compressed representation is designed to preserve unseen linear combinations and not merely the training examples.
Use the paper's explicitly solved SU(2)-based extremal flow as a structured recurrent transition instead of learning an unconstrained dense recurrent matrix. The transition has only two scalar parameters, a radius/frequency r and phase phi, while its rotating coefficient pattern continuously mixes four real state coordinates and can be integrated with a norm-preserving Cayley transform.
Parameterize a cell-complex neural network by features on p-cells and derive lower-dimensional boundary features using the cellular boundary map over F2. For a 2D square complex, neighboring plaquette bits determine each link feature through XOR, reproducing the paper's exact gauge-law reconstruction and preventing the network from representing inconsistent open boundary configurations.
Replace activation-magnitude-based adaptive computation halting with a criterion based on the actual recurrent update and a local stability margin. The loop halts when the state change is small relative to state scale for several consecutive steps, avoiding pathological decisions when LayerNorm-driven dynamics cause the activation norm to collapse.
Decompose tensor-valued hidden states into invariant symmetric, alternating, and higher Young-symmetry channels before applying learned maps. This removes redundant tensor coordinates and prevents a neural layer from mixing incompatible representation types, yielding smaller equivariant modules with a cleaner inductive bias.
Replace independent coordinate rounding of a fixed-sum vector with nearest-point quantization in the projected integer lattice A_n^*. The quantized vector preserves the zero-sum constraint exactly, while the globally optimal rounding correction accounts for the aggregate residual induced by projection.
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.
Replace a large dense layer whose input and output dimensions factor into multiple modes by a TT-matrix whose parameters are stored as a chain of small cores. Periodically apply TT-SVD rounding to remove weak singular directions and keep the representation within a prescribed approximation error. This transfers the paper's central computational principle—perform tensor-product contractions directly in compressed form—to neural network layers.
Build a recurrent block as a fixed or learned ordering of local vertex foldings, mirroring the paper's identification of staircase solution maps with Coxeter elements of a folding group. Each folding changes one polygon coordinate by a rational cross-ratio completion while leaving all other coordinates unchanged. The resulting structured recurrence is reversible and can support constant-memory backpropagation by recomputing folds in reverse order.
Prune hidden units only after testing whether their nonlinear gate is task-visible and downstream-used. For ReLU, a unit is removable when its preactivation does not cross zero on the task patch or its outgoing weight column is zero; this is a more structural criterion than weight magnitude and can be applied during width search or post-training compression.
Represent a large positive semidefinite neural operator as the sum of two Kronecker products and regularize an efficiently computed upper bound on its largest eigenvalues. The bound controls not only the spectral norm but every top-k eigenvalue sum, allowing a tunable penalty on concentrated or unstable directions without constructing the exponentially larger operator.
Replace nominal hidden-unit count by the number of distinct realized ReLU kink hyperplanes, then regularize or prune this effective count. Neurons whose normalized affine boundaries coincide can be exactly merged by summing their canonical coefficients, reducing memory without changing the represented function and aligning the model's complexity measure with the theorem.
Represent each k-element object by a vector in dimension \(r=\binom{n-2(k-s)}{s}\), and use a PSD Gram matrix to encode the rule that pairs with intersection smaller than s have zero similarity while pairs with intersection at least s have nonzero similarity. Insert this representation into set encoders, graph neural networks, or overlap-aware attention instead of allocating one feature for every s-subset.
Replace a dense recurrent transition matrix with a periodic CMV-style product of alternating local 2x2 unitary cores. The transition is exactly norm-preserving, has O(n) trainable parameters under periodic tying, and can be applied through local factor operations rather than stored as an n-by-n matrix. Use turnover refactorization when changing the ordering or boundary connection of cores, enabling a compact cyclic unitary state-space layer.
Replace a dense weight matrix by a cross approximation built from selected rows and columns rather than by a conventional truncated SVD. Periodically refresh the selected indices using residual leverage scores, warm-starting from the previous factorization so that the compressed layer can track weight changes during fine-tuning.