Abstract:
Dual fitting is a general technique for analyzing heuristics. The idea is
to construct a non-feasible dual solution with enough cost to pay for the
primal solution; then this dual solution is scaled down by a r factor to
make it feasible. The best (i.e., smallest) possible value for r is found
via a factor revealing program.
I'll show how this ideas can be applied to the analysis of a greedy
algorithm for the cycle cover problem.
Reference: "Cycle Cover with Short Cycles" by Immorlica, Mahdian, and
Mirrokni. (STACS 05)