First Fit Decreasing Bin Packing (FFDBP)
Input
A list of $n$ items $v_1, \ldots, v_n$, where each $v_i$ is rational number between $0$ and $1$, and two distinguished indices $i$ and $b$.
Problem
Is the $i^\text{th}$ item packed into the $b^\text{th}$ bin by the first fit decreasing bin packing heuristic?
Status
$\P$-complete (Anderson, Mayr, and Warmuth [1]).
Remarks
The reduction to show hardness is from AM2CVP. The problem remains $\P$-complete even if unary representations are used for the numbers involved. This is one of the first such problem where large numbers do not appear to be required for $\P$-completeness (in contrast see MaxFlow).
The problem of determining if $I$ is the packing produced by the best fit decreasing algorithm is also $\P$-complete [1]. In [1] there is an $\NC$ algorithm that produces a packing within $\frac{11}{9}$ of optimal. This is the same performance as for first fit decreasing. Mayr also describes $\NC$ approximation algorithms for bin packing [2].
References
- R. J. Anderson, E. W. Mayr, and M. K. Warmuth (1989). Parallel approximation algorithms for bin packing.
- E. W. Mayr (1988). Parallel approximation algorithms.
- Greenlaw, R., Hoover, H. J., and Ruzzo, W. L. (1995). Limits to Parallel Computation: P-Completeness Theory. Oxford University Press. Problem A.4.7.