Replace a pixelwise density decoder with a decoder that predicts coefficients of a transport potential relative to a fixed reference density. The reconstructed density is the pushforward of the reference measure through a differentiable transport map, so positivity and total mass are structural properties rather than learned penalties.
Replace a conventional two-layer channel mixer in a reversible architecture with the tropicalization of two cluster mutations. For every pair of channels, the block applies sign-dependent integer shears and reflections, giving a cheap piecewise-linear transformation that is exactly invertible and requires no stored activations during backpropagation. Continuous trainable affine scale and mixing parameters can be placed around the fixed tropical core.
Use the paper's non-expansive spectral projection as a constraint-preserving optimizer step for trainable SPD matrices, including Mahalanobis metrics, covariance heads, graph kernels, and quantum density operators. After an SGD or AdamW proposal, symmetrize the matrix and clip every eigenvalue into \([\alpha,\beta]\).
Paper: Projected Riemannian Gradient Descent for the Bures-Wasserstein Barycenter: Dimension-Independent Linear Convergence at Unit Step SizearXiv:2609.03762
Replace isotropic variance control in a bottleneck or router with a spectral polarization penalty that drives each latent direction toward either variance 0 or variance 1. The intended result is an automatically selected active subspace: inactive coordinates can be pruned or quantized aggressively, while active coordinates retain information instead of being uniformly attenuated.
Replace slow primal-coordinate Gibbs sampling of a graph-structured Gaussian latent feature field with Gibbs sampling in the Fourier-dual factor graph. Use the primal convergence formula to determine when dual sampling is worthwhile, and map dual samples back to neural features through the graph incidence operator.
Represent a learned approximately Gaussian latent variable using total-degree Hermite coefficients instead of storing or transmitting all latent coordinates. Estimate the covariance defect relative to the unit Gaussian, choose the smallest Hermite degree whose theoretically predicted tail is below a target error, and train the encoder-decoder through the resulting differentiable spectral bottleneck. This is most appropriate for VAE latents, uncertainty embeddings, or intermediate features that…
Use the flat-torus covariance bound as a representation regularizer that controls the largest covariance eigenvalue while maintaining a prescribed total variance. This creates a directional anti-collapse constraint rather than only a scalar variance penalty, and can be applied to encoder outputs, VAE latents, or Transformer sequence representations.
Build a graph-structured binary latent layer whose local heat-bath probabilities are predicted by a neural network, while particle-exchange and refresh rates remain fixed. The learned probabilities change the stationary distribution and encode input-dependent conditioning, but the spectral invariance result predicts that they do not change the Markov-chain eigenvalues or relaxation modes. This provides a conditional sampler with a fixed, calibratable mixing budget instead of requiring a new…