Grothendieck constant is norm of Strassen matrix multiplication tensor

by   Jinjie Zhang, et al.

We show that two important quantities from two disparate areas of complexity theory --- Strassen's exponent of matrix multiplication ω and Grothendieck's constant K_G --- are intimately related. They are different measures of size for the same underlying object --- the matrix multiplication tensor, i.e., the 3-tensor or bilinear operator μ_l,m,n : F^l × m×F^m × n→F^l × n, (A,B) AB defined by matrix-matrix product over F = R or C. It is well-known that Strassen's exponent of matrix multiplication is the greatest lower bound on (the log of) a tensor rank of μ_l,m,n. We will show that Grothendieck's constant is the least upper bound on a tensor norm of μ_l,m,n, taken over all l, m, n ∈N. Aside from relating the two celebrated quantities, this insight allows us to rewrite Grothendieck's inequality as a norm inequality μ_l,m,n_1,2,∞ =_X,Y,M≠0|tr(XMY)|/ X_1,2 Y_2,∞ M_∞,1≤ K_G, and thereby allows a natural generalization to arbitrary p,q, r, 1< p,q,r<∞. We show that the following generalization is locally sharp: μ_l,m,n_p,q,r=_X,Y,M≠0|tr(XMY)|/X_p,qY_q,rM_r,p≤ K_G · l^|1/q-1/2|· m^1-1/p· n^1/r, and conjecture that Grothendieck's inequality is unique: (p,q,r )=(1,2,∞) is the only choice for which μ_l,m,n_p,q,r is uniformly bounded by a constant. We establish two-thirds of this conjecture: uniform boundedness of μ_l,m,n_p,q,r over all l,m,n necessarily implies that {p,q,r}=1 and {p,q,r}=∞.


page 1

page 2

page 3

page 4


Further limitations of the known approaches for matrix multiplication

We consider the techniques behind the current best algorithms for matrix...

Faster Rectangular Matrix Multiplication by Combination Loss Analysis

Duan, Wu and Zhou (FOCS 2023) recently obtained the improved upper bound...

Laderman matrix multiplication algorithm can be constructed using Strassen algorithm and related tensor's isotropies

In 1969, V. Strassen improves the classical 2x2 matrix multiplication al...

Convolutional number-theoretic method to optimise integer matrix multiplication

There have been several algorithms designed to optimise matrix multiplic...

A Note on Random Sampling for Matrix Multiplication

This paper extends the framework of randomised matrix multiplication to ...

Fast Feasible and Unfeasible Matrix Multiplication

Fast matrix-by-matrix multiplication (hereafter MM) is a highly recogniz...

Bad and good news for Strassen's laser method: Border rank of the 3x3 permanent and strict submultiplicativity

We determine the border ranks of tensors that could potentially advance ...

Please sign up or login with your details

Forgot password? Click here to reset