Every idea extracted from recent arXiv mathematics papers — verified and unverified. Click an idea to open its full card; badges show the empirical verdict.
For a recurrent, state-space, implicit, or complex-valued neural network, partition the local input-output Jacobian into amplitude and phase channels and penalize excessive sensitivity in either channel. This transfers the paper's voltage-source stiffness mechanism to feature magnitude and phase, producing a stability monitor that can distinguish harmless amplitude sensitivity from destructive phase rotation.
Use low precision only for repeated neural-function evaluations and intermediate stage vectors of an explicit ODE solver, while keeping the current state, timestep scaling, and final weighted accumulation in higher precision. This targets neural ODEs and diffusion probability-flow samplers, where function evaluations dominate runtime but accumulated integration error can destabilize long trajectories.
Represent selected activations, weights, or optimizer accumulators as four floating-point limbs and evaluate products through tensor-core matrix multiplications encoding limb convolution. Retain the convolution components during reductions and renormalize only at block boundaries, avoiding branch-heavy multi-double arithmetic inside every multiply-add.
Replace step-by-step hidden-state storage in a latent ODE, state-space model, or world model with a polynomial trajectory represented independently on short time blocks. At the end of each block, pass the next hidden state by summing temporal coefficients, allowing training and inference to discard the completed block while retaining a mathematically exact block interface.
Construct a robust central region of each class or domain embedding cloud by intersecting halfspaces whose discarded cap mass is at most a prescribed fraction. Use this floating-body region to define prototypes or consistency targets, suppressing one-sided outliers without assuming Gaussian covariance structure. The centerpoint level 1/(d+1) provides a principled default depth parameter.
Prepend an adaptive Savitzky-Golay derivative bank to a temporal neural network. For each input channel and derivative order, select the local window by minimizing Stein's unbiased risk estimate, then concatenate the raw signal with the estimated derivatives. This supplies denoised velocity and acceleration features without requiring clean derivative targets or forcing the backbone to learn unstable finite-difference filters.
Construct a block of gradient, preconditioned-gradient, or Hessian-vector-product directions without performing full-dimensional Gram-Schmidt. Use a random sketch to orthogonalize the block cheaply, then solve a small generalized eigenproblem using the true parameter-space overlap matrix so the extracted curvature modes are accurate for the generated subspace. Use the selected curvature modes to form a damped or trust-region optimizer step.
For a coordinate network representing a field near a boundary or interface, factor the prediction as u(x)=h(x)v(x), where h is a known fractional-Hardy ground-state profile, and regularize v with a weighted nonlocal difference energy. Add the corresponding critical Hardy penalty to the loss so that the network spends capacity on the nonsingular residual v instead of relearning the boundary singularity.
Use the complex-conjugate palindromic coefficient that cancels the leading temporal phase defect of oscillatory modes. Implement complex arithmetic directly or use an exactly equivalent doubled-real state, then project the final state to its real component for real-valued prediction tasks.