ScoreQuant

General Differential and Lagrangian Theory for Optimal Experimental Design

Pukelsheim, Titterington1983Optimal experimental design

doi:10.1214/aos/1176346321

Pukelsheim & Titterington (1983) — subgradient and Lagrangian duality (2 Sep 2026)

Use

the comparison template for DS19's tilt certificate — a dual multiplier certifies optimality.

Does not transfer

strong duality relies on the convex feasible set; on hard partitions the minimax interchange fails by exact N=3N=3 witnesses (CE-DS-TILT-DUAL-GAP-002).

General Differential and Lagrangian Theory for Optimal Experimental Design, Ann. Statist. 11(4):1060–1068 (primary record checked by the DS19 audit). Duality by subgradients and by Lagrangian theory for information functionals of the design moment matrix, including singular designs and subset-parameter (K′θK'\theta, hence DsD_s) criteria, over the convex set of design measures on a compact design space.

Ann. Statist. 11(4):1060-1068 (primary record checked 2 Sep 2026). Subgradient and Lagrangian duality for information functionals, including subset-parameter (D_s) criteria, over the convex set of design measures. Comparison template for DS19's certificate; its strong duality needs the convex feasible set that hard partitions lack.

Cited by