Parallel (1+ϵ)-Approximate Mincost Flow in Almost Optimal Depth and Work
September 3, 2025 (GHC 8102)

We present a parallel algorithm for computing $(1+\epsilon)$-approximate

mincost flow on an undirected graph with $m$ edges, where capacities and

costs are assigned to both edges and vertices. Our algorithm achieves

$m^{1+o(1)}$ work and $n^{o(1)}$ depth when $\epsilon > 1/\text{polylog}(m)$,

making both the work and depth almost optimal, up to a subpolynomial

factor.

Previous algorithms with $m^{1+o(1)}$ work required $\Omega(m)$ depth,

even for special cases of mincost flow with only edge capacities or max

flow with vertex capacities.

Our key technical contribution is the first construction of

length-constrained flow shortcuts with $(1+\epsilon)$ length slack,

$n^{o(1)}$ congestion slack, and $n^{o(1)}$ step bound. This provides a

strict generalization of the influential concept of

$(n^{o(1)},\epsilon)$-hopsets [Cohen94], allowing for additional control

over congestion. Previous length-constrained flow shortcuts [HHLRS24]

incur a large constant in the length slack, which would lead to a large

approximation factor.