Tracy–Widom limit conjecture for the centered statistic T1(Y)T_1(Y)

About 1 year old · traced to

Let T1(Y)T_1(Y) be the statistic in the separability test, and let γ0\gamma_0 and E+E_+ be the quantities defined in the paper's equation (sec4.3.eq3). Let TW1TW_1 denote the Tracy–Widom distribution of type 1. Tracy–Widom limit conjecture. Under the same conditions as in Theorem sec4.3.thm1,

γ0n2/3(T1(Y)−E+)⇒TW1.\gamma_0n^{2/3}(T_1(Y)-E_+)\Rightarrow TW_1.

The conjecture proposes replacing the random centering E^+\hat E_+ by the deterministic quantity E+E_+ in the asymptotic null distribution. It is motivated by convergence of E^+\hat E_+ to E+E_+ in probability and empirical evidence, but remains unproved in the cited context.

References

Primary source

Bongjung Sung and Peter D. Hoff, “Testing Separability of High-Dimensional Covariance Matrices”, arXiv:2506.17463 (2026).

Progress summary

Never refreshed

Nothing recorded yet. Refresh searches the literature and the public web for attempts on this problem, and writes the first summary here.

Solutions 0

No solutions have been posted yet.