ProbXiv
Explorersign in
Problem archiveProblem record

Statement

Let T\mathscr{T} be the tensor space and define

T(R)={T∈T:rank⁡{T}≤R}.\mathscr{T}^{(R)}=\{\boldsymbol{\mathcal{T}}\in\mathscr{T}:\operatorname{rank}\{\boldsymbol{\mathcal{T}}\}\leq R\}.

For L>1L>1, let Zl=∥E[R,l]∥Z_l=\|\boldsymbol{\mathcal{E}}[R,l]\| and define

FL[β]=P(ZL≤sin⁡(β)ZL−1≤⋯≤(sin⁡(β))L−1Z1).F_L[\beta]=P\left(Z_L\leq\sin(\beta)Z_{L-1}\leq\cdots\leq(\sin(\beta))^{L-1}Z_1\right).

DCPD distributional convergence conjecture. There exists at least one absolutely continuous probability measure μ\mu on tensors T\boldsymbol{\mathcal{T}} in T(R)\mathscr{T}^{(R)} such that (i) for every ε∈(0,1]\varepsilon\in(0,1] and every l>1l>1, there exists β∈[0,π/2)\beta\in[0,\pi/2) with Fl[β]>1−εF_l[\beta]>1-\varepsilon, and (ii) for every l>1l>1, there exists β∈[0,π/2)\beta\in[0,\pi/2) such that (sin⁡(β))l−1(\sin(\beta))^{l-1} is a strictly monotonically decreasing sequence converging to 00. These conditions are intended to ensure that the residual norms tend to zero and that DCPD converges to an exact CP decomposition. The supplied text gives no resolution status beyond stating the conjecture.

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.