Skip to content

Determine exact ordinary Toeplitz generator counts for tensor Gaussian kernels and their inverses - #7584

Open
CubeJerry wants to merge 1 commit into
CrocSwap:mainfrom
CubeJerry:research/tensor-gaussian-displacement-rank-20261114
Open

CubeJerry wants to merge 1 commit into
CrocSwap:mainfrom
CubeJerry:research/tensor-gaussian-displacement-rank-20261114

Conversation

@CubeJerry

Copy link
Copy Markdown
Contributor

Contribution and significance

Flattening a separable Gaussian tensor does not preserve a short ordinary Toeplitz generator representation. For normalized shifted factors of dimensions n_i>=2, total dimension T and fastest axis f, both the tensor and its inverse have exact ordinary Sylvester and Stein displacement rank 2T/n_f. This is also the exact minimum number of L(u)U(v) terms; the best lexicographic axis order gives 2T/max(n_i).

Explicit boundary minors, matching support bounds and inverse identities prove the result. This adds the flattened representation calculation to the retained shallow-inverse mechanism and the one-axis contracts of #5566/#5571/#6905. It does not repackage the serial factor-of-axes budget as a new result.

Verification

python3 -B scripts/research_package.py --package research/tensor-gaussian-displacement-rank

Recorded checks cover 48 generic forward cases, 47 generic inverse cases, 11 Gaussian tensors, 22 exact generator reconstructions, six axis orders and six scope controls. Rational elimination and independent Bareiss determinants corroborate PROOF.md; a separate proof review is included.

Scope and remaining obligations

Arbitrary coordinate permutations, approximate rank, paid gauges and the complete recovery budget are outside this theorem.

No new kappa: the package retains all five foundational assumptions and does not establish an unconditional multiplication theorem. Finite checks corroborate the scoped written proofs.

Provenance and credit

Base: e54523517d0f03e58352e647c403a5af3e525a57. Full PR draft, exact dependency pins, review and predecessor credit.

Prepared for Jerry Chen / @CubeJerry with substantial OpenAI assistance. Original source authors and predecessor results retain their credit. Model-assisted review is not human or formal-kernel acceptance.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant