Abstract | ||
---|---|---|
Read-k oblivious algebraic branching programs are a natural generalization of the well-studied model of read-once oblivious algebraic branching program (ABP). In this work, we give an exponential lower bound of exp (n/kO(k)) on the width of any read-k oblivious ABP computing some explicit multilinear polynomial f that is computed by a polynomial-size depth-3 circuit. We also study the polynomial identity testing (PIT) problem for this model and obtain a white-box subexponential-time PIT algorithm. The algorithm runs in time 2O(n1−1/2k−1) and needs white box access only to know the order in which the variables appear in the ABP. |
Year | DOI | Venue |
---|---|---|
2018 | 10.1145/3170709 | Electronic Colloquium on Computational Complexity |
Keywords | DocType | Volume |
Algebraic complexity theory, algebraic circuits, lower bounds, polynomial identity testing | Journal | 10 |
Issue | ISSN | Citations |
1 | 1942-3454 | 6 |
PageRank | References | Authors |
0.42 | 47 | 5 |
Name | Order | Citations | PageRank |
---|---|---|---|
Matthew Anderson | 1 | 33 | 2.67 |
Michael A. Forbes | 2 | 12 | 0.85 |
Ramprasad Saptharishi | 3 | 184 | 13.72 |
Amir Shpilka | 4 | 1095 | 64.27 |
Ben Lee Volk | 5 | 12 | 5.62 |