On 4-dimensional convex projective domains invariant by a lattice of $\mathrm{SL}_2 (\mathbb{R})$
arXiv:2607.07150
2026
Architecture
2 ideas extracted · analyzed Aug 30, 2026
What the math gives to ML
The paper gives an explicit irreducible representation of GL_2(R) on homogeneous binary polynomials and identifies a closed, convex, pointed cone of polynomials that are nonnegative on R^2. This can transfer to neural networks as an exactly equivariant feature block, replacing unconstrained channels with symmetric-power features whose transformation law is known analytically. The cone can also constrain energy, uncertainty, or attention-related features to remain nonnegative under all planar directions. The strongest initial experiment is a degree-four equivariant layer combined with a positive-semidefinite Gram parameterization, tested under held-out rotations and shears.
Ideas from this paper
Unverified
2026
Replace an unconstrained feature vector of size n+1 by the coefficients of a homogeneous degree-n binary polynomial and make the layer transform through the irreducible symmetric-power representation of GL_2(R). For n=4 this creates a five-channel equivariant feature block whose transformation law is exact rather than learned through augmentation.
Useful6/10
Difficulty5/10
Novelty6/10
Unverified
2026
Constrain selected degree-four feature blocks to represent globally nonnegative binary quartics using a positive-semidefinite Gram matrix. This gives a structured alternative to unconstrained activations for energy, uncertainty, density, or direction-dependent gating features that must remain nonnegative under every planar direction.
Useful5/10
Difficulty4/10
Novelty7/10