Kronecker scaling of tensors with applications to arithmetic circuits and algorithms
By: Andreas Björklund , Petteri Kaski , Tomohiro Koana and more
Potential Business Impact:
Makes computers solve hard math problems much faster.
We show that sufficiently low tensor rank for the balanced tripartitioning tensor $P_d(x,y,z)=\sum_{A,B,C\in\binom{[3d]}{d}:A\cup B\cup C=[3d]}x_Ay_Bz_C$ for a large enough constant $d$ implies uniform arithmetic circuits for the matrix permanent that are exponentially smaller than circuits obtainable from Ryser's formula. We show that the same low-rank assumption implies exponential time improvements over the state of the art for a wide variety of other related counting and decision problems. As our main methodological contribution, we show that the tensors $P_n$ have a desirable Kronecker scaling property: They can be decomposed efficiently into a small sum of restrictions of Kronecker powers of $P_d$ for constant $d$. We prove this with a new technique relying on Steinitz's lemma, which we hence call Steinitz balancing. As a consequence of our methods, we show that the mentioned low rank assumption (and hence the improved algorithms) is implied by Strassen's asymptotic rank conjecture [Progr. Math. 120 (1994)], a bold conjecture that has recently seen intriguing progress.
Similar Papers
Rank Bounds and PIT for $Σ^3 ΠΣΠ^d$ circuits via a non-linear Edelstein-Kelly theorem
Computational Complexity
Makes computers understand math formulas faster.
Kronecker Powers, Orthogonal Vectors, and the Asymptotic Spectrum
Data Structures and Algorithms
Makes computers solve harder math problems faster.
New results in canonical polyadic decomposition over finite fields
Computational Complexity
Proves if computers can multiply numbers faster.