[Paper Review] Notes on the multiplicity conjecture

Juergen Herzog, Xinxian Zheng|ArXiv.org|May 9, 2005

Commutative Algebra and Its Applications11 references21 citations

TL;DR

This paper advances the multiplicity conjecture in commutative algebra by proving that the multiplicity bounds—both upper and lower—remain valid when passing to quotients by regular sequences and in the limit under ideal powers. It establishes that equality in these bounds occurs if and only if the resolution is pure, extending known results to broader classes of ideals, including those with almost pure resolutions and componentwise linear ideals.

ABSTRACT

New cases of the multiplicity conjecture are considered.

Motivation & Objective

• To extend the multiplicity conjecture to homogeneous algebras that are quotients by regular sequences, even when not Cohen-Macaulay.
• To investigate whether the multiplicity bounds in the conjecture are preserved under taking powers of ideals.
• To determine whether reaching the multiplicity bounds implies purity of the resolution, generalizing known results for codimension 2 and Gorenstein algebras.
• To establish the improved multiplicity conjecture for rings with almost pure resolutions and componentwise linear ideals.
• To verify that the upper or lower bound is achieved if and only if the resolution is pure, in the case of ideals of maximal minors.

Proposed method

• Uses induction on the number of regular sequence elements to prove that multiplicity bounds are preserved under quotienting by regular sequences.
• Applies the tensor product of minimal free resolutions to compute the Betti numbers of the quotient ring $ R/(f_1,\ldots,f_m) $, tracking shifts in the resolution.
• Analyzes the shift patterns $ M_i(I,f) = \max\{M_i, M_{i-1} + \deg f\} $ to relate the new Betti numbers to the original ones.
• Employs combinatorial inequalities involving products of shift minima and maxima to bound the multiplicity $ e(R/(f)) $ in terms of $ \prod M_i / s! $.
• Applies the Eagon-Northcott complex to analyze resolutions of ideals of maximal minors, linking shift uniformity to purity of resolution.
• Uses duality and reflection principles to prove the lower bound inequality by reducing it to the upper bound case via variable substitution.

Experimental results

Research questions

• RQ1 Does the multiplicity conjecture's upper and lower bounds hold for $ R/(f_1,\ldots,f_m) $ when $ R = S/I $ and $ f_1,\ldots,f_m $ is a regular sequence?
• RQ2 Is the multiplicity conjecture preserved in the limit as $ k \to \infty $ for powers of an ideal $ I^k $, specifically $ \lim_{k\to\infty} \frac{e(S/I^k)}{\frac{1}{s!}\prod M_i(I^k)} \leq 1 $?
• RQ3 If the multiplicity of $ R $ reaches the upper or lower bound in the conjecture, does this imply that $ R $ has a pure resolution?
• RQ4 Does the improved multiplicity conjecture hold for rings with almost pure resolutions or componentwise linear ideals?
• RQ5 For ideals of maximal minors, is the multiplicity bound achieved if and only if the resolution is pure?

Key findings

• The multiplicity bounds $ \frac{\prod m_i}{s!} \leq e(R) \leq \frac{\prod M_i}{s!} $ are preserved when passing to $ R/(f_1,\ldots,f_m) $ for any homogeneous regular sequence $ f_1,\ldots,f_m $, even if $ R $ is not Cohen-Macaulay.
• The limit of the normalized multiplicity of $ S/I^k $ satisfies $ \lim_{k\to\infty} \frac{e(S/I^k)}{\frac{1}{s!}\prod M_i(I^k)} \leq 1 $, showing the upper bound holds asymptotically.
• Equality in the multiplicity bounds for $ R $ occurs if and only if the resolution is pure, extending prior results to rings with almost pure resolutions and componentwise linear ideals.
• For ideals of maximal minors of maximal grade, the multiplicity bounds hold, and equality in the upper or lower bound occurs precisely when the resolution is pure.
• The multiplicity of $ S/I_m(H) $ reaches the upper bound if and only if all shifts in the Eagon-Northcott complex are equal, i.e., the resolution is pure.
• Equality in the upper bound for $ e(R/({\bf f})) $ occurs if and only if $ M_i = i d $ and $ \deg f_i = d $ for some fixed $ d $, indicating a uniform shift pattern.