Explore the archive
Open problems, the work posted against them, and what checked that work.
problems
Let n denote the size of a triangulation, and let the average number of its nodes of degree d be divided by n, where d is an integer larger than 2. Polynomial degree-distribution conjecture. There is a probability measure p on the integers…
For k uniform random points Z_1,…,Z_k in a square of area k, prove Elen[span(Z_1,…,Z_k)]~ℓ k.
Let U_1,…,U_n be independent uniform random variables on [-1,1], and set X=1/n∑_i=1^n U_i. For -1<t<1, define the truncated mean m_X(t)=E[X| X>t]. Strengthened truncated-mean inequality. For every -1<t<1,…
How to design+ train NN to be adversarial robust? How to select training data to improve robustness?
Let τ(G) denote the relaxation time of the simple random walk on a connected regular graph G with n vertices. Aldous and Fill conjectured that, as n→∞, max_substackG connected and regular |V(G)|=nτ(G)≤ (1+o(1))3n^2/2π^2, and that…
Many open questions remain. Others: (1) Is the uniqueness theorem true for rough paths with p > 1 (2) Can one identify the range of the signature?
Let X^(n)_1 be the number of distinct adjacent pairs of identical letters in a geometrically distributed word of length n, and let ξ^(n) be the sum of independent random variables ξ_i^(n) with Poisson(nP_i^2) distributions.…
conjecture. Let A \subset N be a finite subset. Denote by DA the set of denominators d of reduced rational numbers b 1 = d 1 a1 + 1 a2 + .. 1 .+ ak whose coefficients a1 , a2 , . . . , ak in its continued fraction expansion are all in A.…
Let ρrenewcommandarraystretch1.2∈P(n) be a partition of n elements with r blocks of sizes n_1≤ n_2≤ …≤ n_r. Let K_ρ,u denote the quantity associated with the partition ρ and the uniform density, and let g^(k)(0) denote the kth derivative…
Fix n and k, and let d be an exceptional point, meaning a parameter value where several basis elements leave simultaneously during a transition between consecutive realizable bases. Reflection conjecture. If d is an exceptional point for…
Let α ∈ (0,1) be the particle density imposed by the reservoir at site 1, and let η_t^step denote the half-line open TASEP with step initial condition. For α>frac12, let μ_1/2^α be the probability measure defined by the matrix product…
For stable commutator length on random elements or random walks, what is the constant and error term in the conjectured asymptotic law; can a finitely generated or finitely presented group have intermediate growth; and is the outer…
For a continuous TASEP on a ring with n particles, let c_i,j(n)= P(w_a=i,w_a+1=j) be the stationary two-point correlation for an adjacent pair, where a is arbitrary. The upper-right correlation conjecture. For every i+1<j,…
\begin{enumerate}[label=\alph*)] \item Relate tr(T(m,n)) to rings of quasi-invariants and to A. Wilson's conjecture on ∇_p_1^n. \item Describe H^*(Hilb(x^nd=y^n)) as a module over rational Double Affine Hecke algebra (DAHA) and relate it…
Characterizing the optimization landscape for RBMs How do we characterize local optima on the optimization landscape for RBMs as a function of the architecture and number of empirical samples?
Let A_R(λ_j) denote the normalized quantities attached to the cyclically ordered lattice points, and let A_∞(r,θ) be their limiting random model. For k≥1, consider the random vectors (A_R(λ_j),A_R(λ_j+k))_1≤ j≤ K. Strengthened vanishing…
For the limiting maximum-flow density v(p) in oriented bond percolation on the square lattice, prove 1-v(p)~√2(1-p) as puparrow1.
Consider Bernoulli distributed edge weights with parameter p. Let g(p) denote the time constant in the coordinate direction for this weight distribution. Is g(p) differentiable as a function of p?
Part (3) of the problem: Let s≤ d≤ m and z_1,…,z_m∈ R^d i.i.d. random vectors with i.i.d. entries ( z_k)_j = rcc -frac1√s with probability s/2m 0 with probability 1-s/m frac1√s with probability s/2m . Note that E z_kz_k^T = frac1m I_d× d.…
Is using dropout during RBM training equivalent to adding a regularizing penalty term on to the training objective function? Is using dropout during RBM training equivalent to adding a regularizing penalty term on to the training objective…