ProbXiv
Explorersign in
Problem archiveProblem record

Statement

Let [L]={1,…,L}[L]=\{1,\ldots,L\} and [L′]={1,…,L′}[L']=\{1,\ldots,L'\}. Let W([L],[L′])\mathcal W([L],[L']) denote the set of channels from [L][L] to [L′][L'], let δi\delta_i be the point mass at ii, and let P(K)([L])\mathfrak P^{(K)}([L]) denote the set of probability distributions on [L]K[L]^K invariant under coordinate permutations. For a channel WW, write W⊗KW^{\otimes K} for its KK-fold tensor power. Under the hypotheses that W(δi)≠W(δj)W(\delta_i)\neq W(\delta_j) for all distinct i,j∈[L]i,j\in[L], the map W⊗KW^{\otimes K} is considered as a map from P(K)([L])\mathfrak P^{(K)}([L]) to P([L′])\mathfrak P([L']).

Optimality conjecture. If K<L−1K<L-1, then there is always a channel W∈W([L],[L′])W\in\mathcal W([L],[L']) satisfying W(δi)≠W(δj)W(\delta_i)\neq W(\delta_j) for all distinct i,j∈[L]i,j\in[L] such that W⊗KW^{\otimes K} is not invertible as a map from P(K)([L])\mathfrak P^{(K)}([L]) to P([L′]\mathfrak P([L']).

The preceding theorem shows that K≥L−1K\geq L-1 is sufficient, so this conjecture would make the bound optimal. It asserts that below this threshold one can always find pairwise distinct input images for which tensoring KK times still loses information.

Record

Comments

Nothing has been published against this problem yet, and nothing has been said about it.

Sign in with an institutional address to take part in the discussion. Reading every thread stays open to everyone.

Sign in

Solve with an agent

Open the statement in a chat, with the problem and the ground rules already written into the prompt.

This opens a third-party site. Nothing is posted back to ProbXiv and nothing you write there is recorded here — what a model gives you is an attempt, which a person still has to check.