A worst-case optimal algorithm to compute the Minkowski sum of convex polytopes
Sandip Das,Subhadeep Ranjan Dev,Sarvottamananda
DOI: https://doi.org/10.1016/j.dam.2024.02.004
IF: 1.254
2024-03-06
Discrete Applied Mathematics
Abstract:We propose algorithms to compute the Minkowski sum of two or more convex polytopes represented by their face lattices in Rd . The time and space complexities of the pair-wise algorithm are O(dωnm) and O(n+m+M) , respectively, where n , m and M are the face lattice sizes of the two summands and the sum, respectively, and ω≈2.37 , currently, is the matrix multiplication exponent. We also show that this algorithm is worst-case optimal for fixed d . Next, we generalize the pair-wise algorithm to r>2 convex polytopes in Rd . The time and space complexities of the r -summands generalized algorithm are O(dωmin{MN,r+∏i=1rNi}) and O(M+N) , respectively, where Ni , 1≤i≤r , is the size of i th summand, N=∑i=1rNi is the total input size and M is the output size of the Minkowski sum. We show that the r -summands generalized algorithm is worst-case optimal for fixed d≥3 and r .
mathematics, applied