Intersection multiplicity over noncommutative algebras (Q1849102)

For a commutative ring \(R\) and two finitely generated \(R\)-modules \(M\) and \(N\) the intersection multiplicity \(\chi(M,N)\) is defined to be the alternating sum over \(i\) of the length of \(\text{Tor}_R^i(M,N)\). It is a classical result that under suitable hypotheses on \(M\), \(N\) and \(R\) one gets that the sum of the Krull dimensions of \(M\) and \(N\) is less than the Krull dimension of \(R\), and if the sum is strictly inferior then \(\chi(M, N)= 0\). It is conjectured that if equality in the inequality holds, then \(\chi(M,N)>0\). \textit{C.-Y. J. Chan} proved [Proc. Am. Math. Soc. 130, No. 2, 327-336 (2002; Zbl 1057.13011)] that under suitable hypotheses on \(R\) it is possible to replace \(\text{Tor}^i\) by \(\text{Ext}^i\). For noncommutative \(k\)-algebras \(R\) one replaces the length by the \(k\)-dimension and obtains similar conjectures. The author studies graded algebras with one-dimensional degree \(0\) term, and studies the above questions for Artin-Schelter regular, Artin-Schelter Cohen Macaulay and Artin-Schelter Gorenstein rings, together with the corresponding cohomology groups, defined as limits over the first \(n\) homogeneous degrees. Krull dimension is then most naturally replaced by Gel'fand-Kirillov dimension and under some relatively mild hypotheses the author gets a positive answer for all three questions. The precise statements are too technical to be stated here.