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 full PSD self-attention with a pivoted Cholesky/Nyström approximation whose landmarks are sampled from the unexplained diagonal mass. Tokens with large residual self-similarity are more likely to become landmarks, so the rank budget is spent on difficult regions rather than uniformly selected tokens.
Choose the neural operator's input-history length from the measured correlation time of the unresolved closure signal produced by coarse-graining. This avoids under-memory, which causes systematic closure error, and over-memory, which increases attention cost and can destabilize training. The same diagnostic can drive adaptive memory truncation across physical regimes.
Replace dense cross-attention weights with a balanced transport plan whose nonzero query-key edges are maintained by a multiscale active-set procedure. Solve the coarse token-group problem first, lift its support to the fine token grid, add only edges indicated by local cost or marginal residuals, and warm-start the fine problem from the lifted plan. This should provide a principled sparse attention pattern rather than fixing a global top-k pattern before seeing the transport solution.
Replace dense attention between tokens on opposite sides of a one-dimensional boundary or segment split with a dyadic low-rank approximation of a Cauchy/Hankel distance kernel. Each distance-scale block uses O(log(1/\varepsilon)) features, and the number of active scales grows only logarithmically with context length after discarding a narrow boundary layer. This is especially suitable for a relative-position attention branch or state-space-like long-range branch, rather than arbitrary…
Compress a transformer KV cache by selecting actual past tokens whose key or hidden-state columns form a stable basis for all cached tokens. Instead of retaining tokens with the largest attention scores or leverage scores independently, compute rank-revealing pivoting of the leading right-singular-vector matrix and retain its pivot columns, then evaluate attention using the representatives plus an optional low-cost residual correction.
Replace dense query-key attention with an adaptive cross approximation constructed from selected query and key pivot tokens. At each rank, choose the pivot pair that removes large estimated residual energy, update the residual by a rank-1 cross correction, and stop when the residual estimate reaches a target tolerance. The resulting factorization computes approximate attention using a small number of landmark interactions while adapting to the actual token distribution.
Replace ordinary low-precision multiply-add accumulation in selected neural-network reductions with a two-word floating-point accumulator updated by the paper's branch-free DW-FMA network. The high word retains the main sum and the low word stores the rounding residual, improving cancellation behavior without the control-flow divergence of conditional compensated summation.
Replace a dense graph-attention or token-mixing matrix by a resolvent-like interaction operator and truncate it to graph neighborhoods whose radius is selected from an estimated spectral gap. Unlike fixed-window sparse attention, the sparsity level is tied to a measurable stability parameter and has an explicit exponential tail criterion.
Replace a dense coordinate-kernel interaction among N points by an orthogonal samplet transform with a sparse detail-detail matrix and a small polynomial branch. Detail basis vectors have vanishing moments, so smooth low-frequency behavior is represented by a few polynomial coefficients while localized residual interactions become sparse in the transformed domain.
Insert a sparse heavy-tailed nonlocal mixing operator into a residual sequence, graph, or spatial network so that information can traverse distant positions without stacking many local layers. Use the critical tail exponent s = 1/2, whose truncated first moment grows logarithmically and predicts an effective propagation distance proportional to depth times log depth rather than merely depth.
Replace a sparse graph layer's separate edge transformations with one joint low-rank factorization of all transformations entering each target node. For target node i, concatenate the neighbor matrices horizontally, project all neighbor features into a shared low-dimensional receiving basis, and reconstruct one output; retain the self transformation exactly. This can reduce edge-parameter storage and message-passing FLOPs when the incoming block row has rapidly decaying singular values.
Approximate a dense symmetric interaction matrix in a neural layer by \(\widehat A=C\widehat M C^{\top}\), but compute the small core \(\widehat M\) from a two-sided sketched least-squares fit rather than from the landmark principal submatrix. This preserves signed or indefinite directions and avoids exploding outputs caused by an almost-singular \(A(I,I)\).
Construct a sparse attention variant whose positional interaction kernel is piecewise polynomial rather than an arbitrary softmax score. Store key-value content in exact monomial moments within a spatial or learned-coordinate box hierarchy, then evaluate all queries from those moments without materializing the query-by-key matrix.
Replace a global attention truncation rule with a per-query local-tail budget. For each query token, retain nearby or high-priority keys until the estimated discarded interaction strength is below a target epsilon; this uses the paper's central distinction between local tail mass and the extensive norm of the discarded operator. The resulting attention pattern can allocate long-range computation only to tokens whose local tail is large.
Make Laplacian latency attention hardware-friendly by quantizing the exponential decay and row normalization to powers of two. Exponentials and divisions can then be implemented with lookup tables and integer shifts, enabling a low-bit attention kernel for neuromorphic and edge accelerators.
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 \(N^{-1}\).
Construct a symmetric token affinity matrix and approximate only its spectrally outlying token-mixing modes using a small set of sampled landmark columns. The resulting low-rank operator replaces an \(O(n^2)\) dense mixer by two skinny matrix multiplications, while the paper's residual guarantee predicts that large-magnitude global interaction modes are preserved.
Replace a dense token or state-mixing matrix with an inverse-capacitance operator whose couplings decay with graph distance, while introducing trainable heterogeneous diagonal capacitances to break spatial symmetries. The layer is cheap because the capacitance matrix is sparse and banded, but its inverse produces global responses with controllable locality.
Replace dense grid tokens or global spectral features with coefficients of compactly supported kernels centered on a nested hierarchy of spatial points. Encode an input field into coarse-to-fine coefficients, apply a neural map to those coefficients, and decode the predicted coefficients at arbitrary query locations; the contribution from each level provides an explicit multiscale output decomposition.
Replace a dense weak pairwise interaction over tokens or graph nodes by an analytically resummed Gaussian quadratic term plus explicitly evaluated corrections containing only subsets of size at most R. Choose R from the target approximation error and interaction scale instead of fixing the interaction order heuristically. This is intended for dense global attention, graph interactions, or latent-variable energy models with individually small couplings.
Replace pointwise pair interactions between mesh cells by quadrature of the interaction kernel over the full Cartesian product of the two cells. Decompose each cell pair into convex-hull pieces and apply a Duffy-like radial transformation so the coincidence singularity is confined to one quadrature coordinate, allowing fixed Gauss-Jacobi or adaptive quadrature to produce smooth, low-variance interaction features.