Grothendieck constant is norm of Strassen matrix multiplication tensor
Abstract
We show that two important quantities from two disparate areas of complexity theory --- Strassen's exponent of matrix multiplication ω and Grothendieck's constant KG --- 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 μl,m,n : Fl × m × Fm × n Fl × 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 μl,m,n. We will show that Grothendieck's constant is the least upper bound on a tensor norm of μl,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 \[ μl,m,n1,2,∞ =X,Y,M≠0|tr(XMY)| X1,2 Y2,∞ M∞,1 KG. \] We prove that Grothendieck's inequality is unique: If we generalize the (1,2,∞)-norm to arbitrary p,q, r ∈ [1, ∞], \[ μl,m,np,q,r=X,Y,M≠0|tr(XMY)|\|X\|p,q\|Y\|q,r\|M\|r,p, \] then (p,q,r )=(1,2,∞) is, up to cyclic permutations, the only choice for which μl,m,np,q,r is uniformly bounded by a constant independent of l,m,n.
Turn this paper into a lesson
ArcXiv compiles a structured reading guide from this paper's metadata: plain-English importance, contributions, prerequisite concepts, which sections to read first, flashcards, and a quiz. Grounded in the abstract, never invented.