Universal bound conjecture for Gaussian entropic optimal transport
Universal bound conjecture for Gaussian entropic optimal transport
Let be positive-definite covariance operators on , and let and denote the entropic and unregularized optimal transport couplings, respectively. Universal bound conjecture. There exists a dimension-dependent constant such that
for all . The asymptotic scaling is already guaranteed in the non-degenerate finite-dimensional setting. The identity case gives , while numerical experiments suggest rates near for random full-rank covariances; it remains open whether the identity case is the worst case or whether the smaller observed rates reflect concentration in random matrix ensembles.
Progress summary
The conjecture remains open: existing work proves the expected small-regularization scaling but not a bound uniform over all covariance pairs.
Ho Yun formulated the universal-bound conjecture in 2025 for Gaussian entropic optimal transport. It asks whether the scaled coupling error is bounded by a constant depending only on the dimension.
Known results
- In the nondegenerate finite-dimensional setting, the error has the required scaling, but a universal constant has not been established (Ho Yun, 2025).
- The identity case gives the lower bound (Ho Yun, 2025).
- Numerical experiments with random full-rank covariances report rates near ; whether this reflects the worst case or random-matrix concentration is unknown (Ho Yun, 2025).
Current status (as of August 2026): The conjecture, including whether the identity case is extremal, remains open; no proof, counterexample, or claimed settlement was found.
Sources
Sources & referencesView supporting material
Primary source
Ho Yun, “Spectral Shrinkage of Gaussian Entropic Optimal Transport”, arXiv:2512.19457 (2026).
Solutions 0
Sign in to submit a solution.
No solutions have been posted yet.