Purified Determinant: Braid Invariants & Applications

• Purified Determinant is a concept in braid theory defined as the determinant of a crossing matrix obtained from a pure power of a braid, ensuring it is invariant under conjugation.
• It is computed by summing permutation translates of the crossing matrix, leading to invariants such as the P-pair that can distinguish braids beyond common linking metrics.
• The concept extends to determinant expansions and Hodge theory, where purification compresses the Leibniz expansion and isolates codimension-one data in singular varieties.

“Purified determinant” is a context-dependent expression rather than a single uniform term of art. In the most explicit usage, it denotes a braid invariant $P(b)$ obtained by first replacing a braid by a pure power and then taking the determinant of its crossing matrix (Shimizu, 10 Sep 2025). In other determinant-centered settings, “purification” refers not to a formal name but to a recognizable operation: compressing the Leibniz expansion into Bell-number-many partition terms, resolving the determinant hypersurface into pure Hodge-theoretic pieces, or replacing a top exterior power on a singular variety by its reflexive hull so that only codimension-one data remains (Houston et al., 2023, Perlman et al., 2020, Dan et al., 2016).

1. Braid-theoretic purified determinant

For an $n$-braid diagram $B$, let $\rho$ be its braid permutation, so that the $i$-th strand on the top ends in position $\rho(i)$ on the bottom. A braid is pure precisely when $\rho=\mathrm{id}$. If $|\rho|$ denotes the order of the permutation, then $B^{|\rho|}$ has trivial permutation and is therefore a pure braid diagram. The purified determinant is defined by

$P(B)=\det\!\big(C(B^{|\rho|})\big),$

where $n$0 is the crossing matrix of $n$1: $n$2 is the $n$3 matrix with $n$4, and for $n$5, $n$6 is the number of positive crossings minus negative crossings between the $n$7-th and $n$8-th strands in which the $n$9-th strand passes over the $B$0-th strand (Shimizu, 10 Sep 2025).

The same paper records the permutation action on matrices,

$B$1

and the product rule

$B$2

where $B$3 is the braid permutation of $B$4. Repeated application of this identity yields the direct computational formula

$B$5

Accordingly, purification is the operation of removing the permutation obstruction by passing to a power whose braid permutation is trivial before taking the determinant.

The construction is a braid invariant because the crossing matrix itself is invariant under braid relations. It is therefore well defined on a braid $B$6, not merely on a chosen diagram, and the notation $B$7 is unambiguous in this setting.

2. Strong conjugation, similarity, and invariance

The principal theorem states that

$B$8

for any pair of $B$9-braids $\rho$0. The paper calls the move $\rho$1 strong conjugation, and the theorem implies in particular that $\rho$2 is unchanged under ordinary conjugacy as well (Shimizu, 10 Sep 2025).

The proof proceeds by similarity of purified crossing matrices. In the pure case, if $\rho$3 is pure with $\rho$4, then

$\rho$5

satisfy

$\rho$6

where $\rho$7 is the transposition swapping $\rho$8 and $\rho$9. Thus the two crossing matrices are similar. By repeatedly sliding generators, the same conclusion extends to arbitrary pure products: if $i$0 is pure, then $i$1 and $i$2 are similar. In the non-pure case, if $i$3 is the order of the permutation of $i$4, then both $i$5 and $i$6 are pure, and their crossing matrices are similar as well. Determinants of similar matrices agree, giving the theorem.

This mechanism is stronger than determinant equality alone. Because the purified matrices are similar, the rank, eigenvalues, eigenvectors, and characteristic polynomial of $i$7 are also braid invariants unchanged under conjugation. The determinant is therefore one member of a larger family of similarity invariants attached to the purified crossing matrix.

3. Computation, discrimination, and limitations

The paper gives an explicit $i$8-braid example with $i$9 and

$\rho(i)$0

Summing its permutation translates gives

$\rho(i)$1

and therefore

$\rho(i)$2

The same paper also exhibits two $\rho(i)$3-braid diagrams $\rho(i)$4 and $\rho(i)$5 with

$\rho(i)$6

showing that the invariant can distinguish braids that have the same order of permutation and the same linking numbers of closures but are not strongly conjugate (Shimizu, 10 Sep 2025).

A refinement called the $\rho(i)$7-pair is defined by

$\rho(i)$8

This records the order of the braid permutation together with the purified determinant. It is also a braid invariant and is unchanged under conjugation. The example

$\rho(i)$9

shows that $\rho=\mathrm{id}$0 can separate a braid from its purified power even when $\rho=\mathrm{id}$1.

The principal limitation is that $\rho=\mathrm{id}$2 is not invariant under stabilization. The paper states explicitly that the purified determinant can change by stabilization, so it is not a link invariant of braid closures. This motivates the later introduction of

$\rho=\mathrm{id}$3

the $\rho=\mathrm{id}$4-determinant, which is invariant under stabilization but not under conjugation. The two constructions therefore capture different aspects of braid equivalence.

4. Structural purification of determinant formulas

A distinct use of “purification” appears in work on explicit determinant expansions. For an $\rho=\mathrm{id}$5 matrix $\rho=\mathrm{id}$6, the determinant is written as a sum over partial partitions $\rho=\mathrm{id}$7,

$\rho=\mathrm{id}$8

with

$\rho=\mathrm{id}$9

If $|\rho|$0 contains a singleton part, the corresponding factor is empty and the term vanishes, so the effective indexing set is the collection of partial partitions with no singleton parts, in bijection with ordinary set partitions of $|\rho|$1. The number of potentially nonzero terms is therefore exactly the Bell number $|\rho|$2, rather than the $|\rho|$3 terms of the Leibniz formula. The paper also compares the formula to the previous best explicit bound $|\rho|$4 and proves

$|\rho|$5

so the compression is superexponential (Houston et al., 2023).

Two proofs are given. The combinatorial proof expands coefficients of monomials $|\rho|$6, shows cancellation by a sign-reversing involution when $|\rho|$7 is not a permutation, and recovers $|\rho|$8 when $|\rho|$9 is a permutation via Stirling-number cancellation. The geometric proof defines the cuboid

$B^{|\rho|}$0

the lattice

$B^{|\rho|}$1

and an axis-aligned polytope $B^{|\rho|}$2 obtained by removing overlaps of translated cuboids from $B^{|\rho|}$3. The translates $B^{|\rho|}$4 tile $B^{|\rho|}$5, and inclusion–exclusion computes $B^{|\rho|}$6 as the right-hand side of the formula, giving $B^{|\rho|}$7.

This compressed formula has immediate rank-theoretic consequences. It yields

$B^{|\rho|}$8

over any field $B^{|\rho|}$9, characteristic-$P(B)=\det\!\big(C(B^{|\rho|})\big),$0 refinements

$P(B)=\det\!\big(C(B^{|\rho|})\big),$1

the exact value

$P(B)=\det\!\big(C(B^{|\rho|})\big),$2

and, when $P(B)=\det\!\big(C(B^{|\rho|})\big),$3 or $P(B)=\det\!\big(C(B^{|\rho|})\big),$4,

$P(B)=\det\!\big(C(B^{|\rho|})\big),$5

The paper states that this Waring-rank bound improves the previous $P(B)=\det\!\big(C(B^{|\rho|})\big),$6 bound for all $P(B)=\det\!\big(C(B^{|\rho|})\big),$7. In this sense, the determinant is “purified” by replacing permutation-by-permutation bookkeeping with Bell-number-many partition structures.

5. Hodge-theoretic and singular-variety purification

For the determinant hypersurface, purification can mean passage from coarse pole-order data to explicit Hodge-theoretic structure. Let

$P(B)=\det\!\big(C(B^{|\rho|})\big),$8

and let $P(B)=\det\!\big(C(B^{|\rho|})\big),$9 be the ideal generated by the $n$00 minors. The localization $n$01 carries Saito’s Hodge filtration $n$02, with Hodge ideals defined by

$n$03

The paper computes these ideals explicitly: $n$04 and identifies the weight-graded pieces by

$n$05

with $n$06 for $n$07 or $n$08. The Hodge filtration on $n$09 has generation level $n$10. Here purification means that the mixed object $n$11 is decomposed into pure Hodge modules supported on the rank strata (Perlman et al., 2020).

A different purification appears for determinants of coherent sheaves on singular projective varieties. If $n$12 is projective and $n$13 is coherent of rank $n$14, the determinant is defined by

$n$15

The reflexive hull suppresses codimension-$n$16 defects and retains codimension-one information. In flat families with normal integral fibers, the Hilbert polynomial of the determinant is not constant, but the paper proves an upper semi-continuous behavior and constructs a determinant morphism after stratifying the base by determinant Hilbert polynomial. In the curve-on-surface application, if $n$17 is the ideal sheaf of a curve $n$18 on a normal surface, there is an exact sequence

$n$19

relating the determinant to the Cohen–Macaulayfication of the curve (Dan et al., 2016).

Taken together, these two works suggest two distinct purification paradigms for determinants: decomposition into pure Hodge-module constituents, and reflexive removal of codimension-$n$20 singular behavior.

6. Related determinant analogues and terminological boundaries

Several nearby notions should be distinguished from the braid-theoretic purified determinant. For a Hermitian matrix $n$21, the pseudo determinant is

$n$22

and the paper proves the canonical derivative formula

$n$23

where $n$24 is the Moore–Penrose pseudoinverse. This is a determinant-like construction that removes zero spectral directions, but the term used there is “pseudo determinant,” not “purified determinant” (Holbrook, 2018).

In operator-algebraic language, the determinant map on pure C$n$25-algebras is again different. For a separable, simple, pure C$n$26-algebra of stable rank one in which every bounded $n$27-quasitrace is a trace, the de la Harpe–Skandalis determinant has kernel equal to the commutator subgroup on $n$28, and determinant-zero elements are finite products of commutators. Here “pure” modifies the algebra, not the determinant (Ng et al., 2014).

Commutative-algebraic uses of purity also differ. The hypersurface defined by

$n$29

is proved to be $n$30-pure for all matrix sizes and all positive prime characteristics. The result concerns Frobenius purity of the quotient ring $n$31, not a determinant object called purified determinant (Kadyrsizova, 2019).

Finally, “purified gravity” is a nonmetricity-based reformulation of General Relativity called Coincident General Relativity. It uses the metric density $n$32 in the action

$n$33

but it does not introduce a determinant construction named purified determinant (Jiménez et al., 2019).

Accordingly, the term has one precise braid-theoretic meaning and several looser determinant-related senses in which purification denotes removal of permutation data, null spectral directions, mixed Hodge contributions, or codimension-$n$34 singular defects. Context is therefore decisive.