Separation between Estimation and Approximation


August 18, 2015


Uriel Feige


ILDC - Weizmann Institute


We show (under standard assumptions) that there are NP optimization problems for which estimation is easier than approximation. Namely, one can estimate the value of the optimal solution within a ratio of r, but it is difficult to find a solution whose value is within r of optimal. As an important special case, we show that there are linear programming relaxations for which no polynomial time rounding technique matches the integrality gap of the linear program.

Joint work with Shlomo Jozeph.


Uriel Feige

Uriel Feige received his PhD at the Weizmann Institute, and except for a few years in Princeton University, IBM Research, Compaq Research and Microsoft Research, stayed there ever since. His research explores the border between P and NP, whether for worst case or average case input instances, and involves the design and analysis of algorithms, and the proof of computational hardness results.