Joint Reductions in Algebra and Integrable Systems

• Joint reductions are asymptotic reduction tools that select elements from families of ideals or modules to stabilize mixed products for large indices.
• They enable the conversion of mixed multiplicities into standard Hilbert–Samuel multiplicities and are key in proving generalized Rees theorems through multigraded filtrations.
• Construction methods using general elements, superficial sequences, and hyperplane restrictions extend joint reductions to settings in commutative algebra and integrable lattice equations.

Joint reductions are asymptotic reduction data attached to a family of ideals, filtrations, or modules: one chooses elements or submodules from each component so that sufficiently large mixed products, filtration terms, or symmetric powers satisfy a stable reduction equation. In commutative algebra, the notion is used in the study of Rees algebras, multigraded fiber cones, mixed multiplicities, complete and integrally closed ideals, and Buchsbaum–Rim theory. In a distinct literature on integrable lattice equations, the same expression appears in a different sense, namely dimension reduction by passing to joint invariants of commuting symmetries (Goel et al., 2018, Thanh et al., 2019, Katz et al., 10 Aug 2025, Kamp et al., 2010).

1. Basic definitions for ideals, modules, and types

Let $(R,\m)$ be a Noetherian local ring, let $I_1,\dots,I_s$ be ideals, and for $\underline n=(n_1,\dots,n_s)\in \mathbb N^s$ write

$$\underline I^{\underline n}=I_1^{n_1}I_2^{n_2}\cdots I_s^{n_s}.$$

A sequence $(a_1,\dots,a_s)$ with $a_i\in I_i$ is called a joint reduction of $(I_1,\dots,I_s)$ if there exists

$\underline r=(r_1,\dots,r_s)\in \mathbb N^s$

such that for all $\underline n\ge \underline r$ coordinate-wise,

$$\sum_{i=1}^s a_i\,\underline I^{\underline n-e_i}=\underline I^{\underline n}.$$

The multi-index $I_1,\dots,I_s$0 is then called a joint reduction vector corresponding to the chosen joint reduction (Goel et al., 2018).

A second, equivalent style of notation records how many elements are chosen from each ideal. A collection

$I_1,\dots,I_s$1

is called a joint reduction of type $I_1,\dots,I_s$2 if, for each large multi-index $I_1,\dots,I_s$3, the product with all exponents shifted by $I_1,\dots,I_s$4 is generated by the chosen elements times the corresponding one-step smaller products. In the fiber-cone formulation, this is expressed by equality of the relevant multigraded components of

$I_1,\dots,I_s$5

and the smallest $I_1,\dots,I_s$6 beyond which this equality holds is the joint reduction number of type $I_1,\dots,I_s$7. When $I_1,\dots,I_s$8, this recovers the classical notion of a reduction of a single ideal (Caviglia, 2021).

For mixed multiplicity theory one often adjoins an $I_1,\dots,I_s$9-primary ideal $\underline n=(n_1,\dots,n_s)\in \mathbb N^s$0. With $\underline n=(n_1,\dots,n_s)\in \mathbb N^s$1 a Noetherian local ring, $\underline n=(n_1,\dots,n_s)\in \mathbb N^s$2 a finitely generated $\underline n=(n_1,\dots,n_s)\in \mathbb N^s$3-module, $\underline n=(n_1,\dots,n_s)\in \mathbb N^s$4, and ideals $\underline n=(n_1,\dots,n_s)\in \mathbb N^s$5, a sequence

$\underline n=(n_1,\dots,n_s)\in \mathbb N^s$6

with each entry in the corresponding ideal is a joint reduction of $\underline n=(n_1,\dots,n_s)\in \mathbb N^s$7 with respect to $\underline n=(n_1,\dots,n_s)\in \mathbb N^s$8 of type $\underline n=(n_1,\dots,n_s)\in \mathbb N^s$9 if for all large $$\underline I^{\underline n}=I_1^{n_1}I_2^{n_2}\cdots I_s^{n_s}.$$0,

$$\underline I^{\underline n}=I_1^{n_1}I_2^{n_2}\cdots I_s^{n_s}.$$1

When $$\underline I^{\underline n}=I_1^{n_1}I_2^{n_2}\cdots I_s^{n_s}.$$2, this recovers Northcott–Rees reductions (Thanh et al., 2019).

2. Multigraded filtrations and joint reduction vectors

Joint reductions extend naturally from products of ideals to multigraded filtrations. If $$\underline I^{\underline n}=I_1^{n_1}I_2^{n_2}\cdots I_s^{n_s}.$$3 are ideals in a Noetherian local ring $$\underline I^{\underline n}=I_1^{n_1}I_2^{n_2}\cdots I_s^{n_s}.$$4, an $$\underline I^{\underline n}=I_1^{n_1}I_2^{n_2}\cdots I_s^{n_s}.$$5-graded filtration $$\underline I^{\underline n}=I_1^{n_1}I_2^{n_2}\cdots I_s^{n_s}.$$6 is a family of ideals satisfying

$$\underline I^{\underline n}=I_1^{n_1}I_2^{n_2}\cdots I_s^{n_s}.$$7

and

$$\underline I^{\underline n}=I_1^{n_1}I_2^{n_2}\cdots I_s^{n_s}.$$8

It is an $$\underline I^{\underline n}=I_1^{n_1}I_2^{n_2}\cdots I_s^{n_s}.$$9-good filtration if, in addition, $(a_1,\dots,a_s)$0 for $(a_1,\dots,a_s)$1 and the Rees algebra

$(a_1,\dots,a_s)$2

is module-finite over

$(a_1,\dots,a_s)$3

This setting allows joint reductions to be stated for general filtrations rather than only for ordinary powers (Goel et al., 2018).

Goel–Roy–Verma prove an Eakin–Sathaye type theorem for $(a_1,\dots,a_s)$4-graded good filtrations. If $(a_1,\dots,a_s)$5 has infinite residue field, $(a_1,\dots,a_s)$6 satisfies

$(a_1,\dots,a_s)$7

and $(a_1,\dots,a_s)$8, where $(a_1,\dots,a_s)$9 is the component-wise maximum of the multi-degrees of a fixed generating set of the fiber-cone module

$a_i\in I_i$0

over

$a_i\in I_i$1

then for each $a_i\in I_i$2 there exist general elements $a_i\in I_i$3 such that

$a_i\in I_i$4

Thus the sequences $a_i\in I_i$5 form a joint reduction with joint reduction vector $a_i\in I_i$6. In the case $a_i\in I_i$7, this recovers the bound on the reduction number of an $a_i\in I_i$8-graded good filtration (Goel et al., 2018).

The proof strategy is multigraded. It works in the fiber-cone module, applies Nakayama’s lemma, and proceeds by double induction on $a_i\in I_i$9 and $(I_1,\dots,I_s)$0. The hyperplane-section argument, annihilator filtrations, and the multi-binomial coefficient comparison are central, while the invariant $(I_1,\dots,I_s)$1 plays the role of a “Castelnuovo–Mumford–regularity” for the fiber-cone module (Goel et al., 2018).

3. Mixed multiplicities and Hilbert–Samuel multiplicity

A major function of joint reductions is to convert mixed multiplicities into ordinary Hilbert–Samuel multiplicities. Let $(I_1,\dots,I_s)$2 be a Noetherian local ring with infinite residue field, $(I_1,\dots,I_s)$3 a finitely generated $(I_1,\dots,I_s)$4-module of dimension $(I_1,\dots,I_s)$5, $(I_1,\dots,I_s)$6 an $(I_1,\dots,I_s)$7-primary ideal, and $(I_1,\dots,I_s)$8 arbitrary ideals. For large $(I_1,\dots,I_s)$9, the length

$\underline r=(r_1,\dots,r_s)\in \mathbb N^s$0

is a polynomial of total degree $\underline r=(r_1,\dots,r_s)\in \mathbb N^s$1, and the coefficients of its highest-degree part define the mixed multiplicities

$\underline r=(r_1,\dots,r_s)\in \mathbb N^s$2

Equivalently, they are obtained by finite differences of the Hilbert polynomial. When only $\underline r=(r_1,\dots,r_s)\in \mathbb N^s$3 appears, the mixed multiplicity reduces to the usual Hilbert–Samuel multiplicity $\underline r=(r_1,\dots,r_s)\in \mathbb N^s$4 (Thanh et al., 2019).

Thanh–Viet prove that if $\underline r=(r_1,\dots,r_s)\in \mathbb N^s$5 and

$\underline r=(r_1,\dots,r_s)\in \mathbb N^s$6

then any joint reduction of $\underline r=(r_1,\dots,r_s)\in \mathbb N^s$7 with respect to $\underline r=(r_1,\dots,r_s)\in \mathbb N^s$8 of type $\underline r=(r_1,\dots,r_s)\in \mathbb N^s$9 is automatically a system of parameters for $\underline n\ge \underline r$0, and

$\underline n\ge \underline r$1

The paper emphasizes two removals of hypotheses: it no longer requires $\underline n\ge \underline r$2, and it does not assume a priori that the chosen joint reduction is already a system of parameters (Thanh et al., 2019).

Trung–Verma establish a related generalized Rees theorem under the height condition

$\underline n\ge \underline r$3

If $\underline n\ge \underline r$4 is a joint reduction of $\underline n\ge \underline r$5 of type $\underline n\ge \underline r$6 and a system of parameters for $\underline n\ge \underline r$7, then

$\underline n\ge \underline r$8

The same paper shows that a Rees-superficial sequence of the corresponding type and length $\underline n\ge \underline r$9, once it is a system of parameters, is automatically a joint reduction of the same type (Viet et al., 2011).

Duong Quoc Viet gives a recursion formula for mixed multiplicities of maximal degrees with respect to joint reductions. If

$$\sum_{i=1}^s a_i\,\underline I^{\underline n-e_i}=\underline I^{\underline n}.$$0

is a joint reduction of type $$\sum_{i=1}^s a_i\,\underline I^{\underline n-e_i}=\underline I^{\underline n}.$$1 and $$\sum_{i=1}^s a_i\,\underline I^{\underline n-e_i}=\underline I^{\underline n}.$$2, then

$$\sum_{i=1}^s a_i\,\underline I^{\underline n-e_i}=\underline I^{\underline n}.$$3

Under $$\sum_{i=1}^s a_i\,\underline I^{\underline n-e_i}=\underline I^{\underline n}.$$4-regularity or $$\sum_{i=1}^s a_i\,\underline I^{\underline n-e_i}=\underline I^{\underline n}.$$5-filter-regularity, the second term vanishes. Corollaries further identify mixed multiplicities with Hilbert–Samuel multiplicities of subsystems of the joint reduction under dimension or height hypotheses (Viet, 2021).

4. Construction methods: general elements, superficial sequences, and hyperplane restriction

Several construction techniques recur across the literature. In multigraded filtrations, the Eakin–Sathaye type theorem furnishes “general” elements $$\sum_{i=1}^s a_i\,\underline I^{\underline n-e_i}=\underline I^{\underline n}.$$6 once the multibinomial bound on $$\sum_{i=1}^s a_i\,\underline I^{\underline n-e_i}=\underline I^{\underline n}.$$7 is satisfied; the resulting joint reduction vector is read directly from the integers $$\sum_{i=1}^s a_i\,\underline I^{\underline n-e_i}=\underline I^{\underline n}.$$8 (Goel et al., 2018).

In mixed multiplicity theory, weak-$$\sum_{i=1}^s a_i\,\underline I^{\underline n-e_i}=\underline I^{\underline n}.$$9 elements and weak-$I_1,\dots,I_s$00 sequences play a similar role. Thanh–Viet use them to pass inductively to quotients by one chosen element, and Corollary 3.6 states that any weak-$I_1,\dots,I_s$01-sequence of type $I_1,\dots,I_s$02 satisfying the same dimension bound is automatically a joint reduction, so its ordinary multiplicity equals the corresponding mixed multiplicity (Thanh et al., 2019).

Rees-superficial sequences provide another practical construction. Trung–Verma define a Rees-superficial sequence by the asymptotic intersection property

$I_1,\dots,I_s$03

for all large exponents. Corollary 2.7 in that paper shows that a Rees-superficial sequence of the appropriate type, once it is a system of parameters, automatically satisfies the joint reduction property (Viet et al., 2011).

Caviglia gives a different construction paradigm by strengthening Green’s general hyperplane restriction theorem. In the standard graded fiber cone of a product ideal, the “bad” linear forms that fail the expected Hilbert-function estimate lie in a finite union of proper linear subspaces. Choosing successive linear forms outside these forbidden subspaces forces vanishing in a prescribed degree, hence

$I_1,\dots,I_s$04

and after lifting from the fiber cone to the local ring, Nakayama’s lemma yields the desired reduction. When the ideal is a product

$I_1,\dots,I_s$05

the lifted generators may be chosen diagonally, so that each reduction generator factors as $I_1,\dots,I_s$06 with $I_1,\dots,I_s$07. The paper presents this as a method for recovering and extending results of O’Carroll on complete and joint reductions (Caviglia, 2021).

5. Representative examples and special classes

The examples in the literature show that joint reductions are often controlled by explicit generator counts. For contracted ideals in a $I_1,\dots,I_s$08-dimensional regular local ring, if $I_1,\dots,I_s$09 and $I_1,\dots,I_s$10 are contracted with orders $I_1,\dots,I_s$11 and $I_1,\dots,I_s$12, then

$I_1,\dots,I_s$13

The Eakin–Sathaye bound becomes

$I_1,\dots,I_s$14

and the choice

$I_1,\dots,I_s$15

gives a valid joint reduction vector for $I_1,\dots,I_s$16. Hence there exist $I_1,\dots,I_s$17 and $I_1,\dots,I_s$18 with

$I_1,\dots,I_s$19

For lex-segment ideals $I_1,\dots,I_s$20 with $I_1,\dots,I_s$21 and $I_1,\dots,I_s$22, one has

$I_1,\dots,I_s$23

and if $I_1,\dots,I_s$24, then $I_1,\dots,I_s$25 is the smallest solution of the joint-reduction inequality, giving a joint reduction vector $I_1,\dots,I_s$26 (Goel et al., 2018).

For closure filtrations, the same paper treats two hypersurface rings. In

$I_1,\dots,I_s$27

the integral-closure and tight-closure filtrations are $I_1,\dots,I_s$28-good, both the associated graded ring and the fiber cone are Cohen–Macaulay, and

$I_1,\dots,I_s$29

The one-variable Eakin–Sathaye theorem then gives $I_1,\dots,I_s$30 exactly. In

$I_1,\dots,I_s$31

one has

$I_1,\dots,I_s$32

so $I_1,\dots,I_s$33 for all $I_1,\dots,I_s$34, the generator-degree bound is $I_1,\dots,I_s$35, and for $I_1,\dots,I_s$36,

$I_1,\dots,I_s$37

in agreement with the direct computation that the reduction number is $I_1,\dots,I_s$38 (Goel et al., 2018).

In dimension $I_1,\dots,I_s$39, joint reductions of the integral-closure filtration $I_1,\dots,I_s$40 are studied through local cohomology. A triple $I_1,\dots,I_s$41 is a joint reduction if there exists $I_1,\dots,I_s$42 such that for all $I_1,\dots,I_s$43 with $I_1,\dots,I_s$44,

$I_1,\dots,I_s$45

Under depth hypotheses on associated graded rings, or if the extended Rees algebra is Cohen–Macaulay, one can choose a good complete reduction matrix, hence a good joint reduction. The paper then computes

$I_1,\dots,I_s$46

and states equivalences between the vanishing of this local cohomology, the vanishing of that linear combination of normal Hilbert coefficients, and the condition that the joint reduction number is $I_1,\dots,I_s$47. For $I_1,\dots,I_s$48-primary monomial ideals in $I_1,\dots,I_s$49, the Rees algebra $I_1,\dots,I_s$50 is Cohen–Macaulay, hence every joint reduction is good and has joint-reduction-number $I_1,\dots,I_s$51 (Masuti et al., 2014).

6. Joint reductions of modules and mixed Buchsbaum–Rim multiplicity

The module-theoretic extension replaces ideals by finite-colength submodules of free modules. Let $I_1,\dots,I_s$52 be a Noetherian local ring of dimension $I_1,\dots,I_s$53, and for $I_1,\dots,I_s$54, let $I_1,\dots,I_s$55 be free of rank $I_1,\dots,I_s$56 and $I_1,\dots,I_s$57 an $I_1,\dots,I_s$58-submodule of finite colength. An ordered collection $I_1,\dots,I_s$59, with each $I_1,\dots,I_s$60 an $I_1,\dots,I_s$61-generated submodule, is a joint reduction of $I_1,\dots,I_s$62 if for some $I_1,\dots,I_s$63,

$I_1,\dots,I_s$64

in the symmetric algebra $I_1,\dots,I_s$65. Equivalently, the ideal generated by all of $I_1,\dots,I_s$66 contains a power of the irrelevant ideal, and the smallest such $I_1,\dots,I_s$67 is the joint-reduction number (Katz et al., 10 Aug 2025).

Katz–Kodiyalam–Verma prove several structural facts. If $I_1,\dots,I_s$68, then at least one joint reduction exists. Reduction modulo minimal primes preserves the notion. Most notably, for $I_1,\dots,I_s$69 with each $I_1,\dots,I_s$70 of rank $I_1,\dots,I_s$71, the equational condition is equivalent to a valuative condition over DVRs and to a determinantal condition saying that the maximal-minor determinants $I_1,\dots,I_s$72 form a joint reduction of the corresponding ideal-modules $I_1,\dots,I_s$73 in the sense of Rees. If $I_1,\dots,I_s$74 and $I_1,\dots,I_s$75, each $I_1,\dots,I_s$76 may be chosen free of rank $I_1,\dots,I_s$77; when $I_1,\dots,I_s$78, every $I_1,\dots,I_s$79 is part of a minimal generating set of $I_1,\dots,I_s$80, and a minimal generating set of each $I_1,\dots,I_s$81 extends to one of $I_1,\dots,I_s$82 (Katz et al., 10 Aug 2025).

The same paper defines the mixed Buchsbaum–Rim multiplicity $I_1,\dots,I_s$83 as the normalized leading coefficient of the joint Buchsbaum–Rim polynomial

$I_1,\dots,I_s$84

and proves that if $I_1,\dots,I_s$85 is a joint reduction, then

$I_1,\dots,I_s$86

where the $I_1,\dots,I_s$87 are endomorphisms with image $I_1,\dots,I_s$88. A further K-theoretic argument identifies this Euler–Poincaré characteristic with the corresponding mixed multiplicity of the determinant ideals. When each $I_1,\dots,I_s$89, so that $I_1,\dots,I_s$90 is an $I_1,\dots,I_s$91-primary ideal, this recovers Rees’s mixed multiplicity $I_1,\dots,I_s$92 (Katz et al., 10 Aug 2025).

A central two-dimensional result is the joint-reduction-number-zero theorem. If $I_1,\dots,I_s$93 is a two-dimensional regular local ring and $I_1,\dots,I_s$94, $I_1,\dots,I_s$95 are integrally closed modules of finite colength, then for any joint reduction $I_1,\dots,I_s$96,

$I_1,\dots,I_s$97

in the symmetric algebra $I_1,\dots,I_s$98; equivalently, the joint-reduction number is zero. The paper supplies two proofs, one using order valuations and contracted modules, and another using Tor vanishing, the Hoskin–Deligne length formula, and bilinearity of mixed Buchsbaum–Rim multiplicity (Katz et al., 10 Aug 2025).

7. A distinct usage in integrable lattice equations

In the theory of periodic reductions of integrable lattice equations, “joint reductions” refers to a different construction. Starting from an $I_1,\dots,I_s$99-dimensional mapping $\underline n=(n_1,\dots,n_s)\in \mathbb N^s$00 obtained by an $\underline n=(n_1,\dots,n_s)\in \mathbb N^s$01-periodic reduction and a monodromy matrix $\underline n=(n_1,\dots,n_s)\in \mathbb N^s$02, the staircase method yields $\underline n=(n_1,\dots,n_s)\in \mathbb N^s$03 functionally independent integrals from the characteristic polynomial or, equivalently, from the traces $\underline n=(n_1,\dots,n_s)\in \mathbb N^s$04. If $\underline n=(n_1,\dots,n_s)\in \mathbb N^s$05, the paper shows that one can introduce $\underline n=(n_1,\dots,n_s)\in \mathbb N^s$06 variables that reduce the dimension of the mapping from $\underline n=(n_1,\dots,n_s)\in \mathbb N^s$07 to $\underline n=(n_1,\dots,n_s)\in \mathbb N^s$08; these variables are obtained as joint invariants of $\underline n=(n_1,\dots,n_s)\in \mathbb N^s$09-symmetries of the mapping (Kamp et al., 2010).

More precisely, if the mapping admits commuting symmetry generators

$\underline n=(n_1,\dots,n_s)\in \mathbb N^s$10

one seeks functions $\underline n=(n_1,\dots,n_s)\in \mathbb N^s$11 satisfying

$\underline n=(n_1,\dots,n_s)\in \mathbb N^s$12

These joint invariants define reduced coordinates, and because the mapping commutes with the symmetries, it induces a closed $\underline n=(n_1,\dots,n_s)\in \mathbb N^s$13-dimensional map on the $\underline n=(n_1,\dots,n_s)\in \mathbb N^s$14-variables alone. The original integrals descend to this reduced phase space. In the Boussinesq $\underline n=(n_1,\dots,n_s)\in \mathbb N^s$15-reduction, the staircase method yields

$\underline n=(n_1,\dots,n_s)\in \mathbb N^s$16

independent invariants; a translation symmetry gives a $\underline n=(n_1,\dots,n_s)\in \mathbb N^s$17-dimensional reduction, and for even $\underline n=(n_1,\dots,n_s)\in \mathbb N^s$18, two additional parity-dependent $\underline n=(n_1,\dots,n_s)\in \mathbb N^s$19-symmetries give a final reduction to dimension $\underline n=(n_1,\dots,n_s)\in \mathbb N^s$20 (Kamp et al., 2010).

This usage is terminologically related only at the level of simultaneous reduction with respect to several directions or symmetries. It is not the same notion as a joint reduction of ideals, filtrations, or modules in commutative algebra.