Mirror descent iterates converge to a Karush-Kuhn-Tucker point for a broad class of nonconvex problems with entropy and power kernels, via a metric-flattening reparameterization.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
math.OC 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Establishing Boundary KKT Convergence of Mirror Descent through Reparameterization
Mirror descent iterates converge to a Karush-Kuhn-Tucker point for a broad class of nonconvex problems with entropy and power kernels, via a metric-flattening reparameterization.