Symmetry Barcodes: Invariants and Applications

• Symmetry barcodes are barcode-like invariants that record the birth and death of symmetries in parameterized configurations, encoding structural invariance through algebraic and geometric data.
• They organize persistence barcode spaces via permutation actions, Coxeter complexes, and combinatorial lattices, revealing discrete ordering and intrinsic symmetry dynamics.
• Applied in theoretical analysis and visual barcode engineering, these invariants stabilize metrics and inform algorithmic dualities in topological data analysis.

Symmetry barcodes are barcode-like invariants in which symmetry is elevated from a background invariance principle to a primary encoded object. In the most direct recent sense, a symmetry barcode records the birth and death of individual symmetries of a parameterized finite configuration in a metric space, while the associated polybarcode records disappearance and reappearance of the same symmetry as a closed subset of the parameter line (Liu et al., 11 Aug 2025). In adjacent work on persistent-homology barcodes themselves, symmetry refers to permutation invariance of interval multisets, Coxeter-complex stratifications of barcode space, permutation-type invariants, and symmetry-aware coordinate systems and metrics (Brück et al., 2021). The literature therefore uses the expression in several related but non-identical senses, all organized by the principle that relabelling, ordering, or geometric repetition should be encoded through explicit algebraic or geometric structure rather than treated informally.

1. Scope of the term

Current usage separates into three main strands. In one strand, symmetry barcodes are invariants of parameterized configurations whose symmetry groups evolve with the parameter. In a second strand, the space of ordinary persistence barcodes is itself analyzed through symmetric-group actions, Coxeter complexes, weak Bruhat order, and related combinatorics. In a third, distinct engineering strand, symmetry is used as the calibration signal for a two-dimensional visual barcode (Liu et al., 11 Aug 2025, Brück et al., 2021, Ma et al., 2023).

Usage	Core object	Symmetry mechanism
Persistent symmetries of data	Configuration $X_t \subset M$	Isometry groups, spans of groups, symmetry bars
Symmetry in barcode spaces	Barcode with $n$ bars	$\mathrm{Sym}_n$, Coxeter complexes, Bruhat-type order
Visual barcode engineering	OAcode	Central symmetry of the data area

A recurring source of ambiguity is that these strands do not encode the same mathematical object. In the first, the barcode is built from symmetries. In the second, symmetry organizes the space of persistence barcodes. In the third, symmetry is an operational cue for localization and decoding. The shared theme is structural invariance under relabelling, conjugation, or repeated patterning.

2. Symmetric-group structure on the space of persistence barcodes

For barcodes with exactly $n$ bars, a basic construction identifies the barcode space with a quotient by the symmetric group. A barcode is a finite multiset

$$B=\{(b_i,d_i)\}_{i\in J},\qquad (b_i,d_i)\in\mathbb{R}^2,\ b_i<d_i,$$

and for fixed $n$,

$$\mathcal{B}_n=\{\text{barcodes with }n\text{ bars}\}.$$

Choosing an indexing $J=\{1,\dots,n\}$, one represents a barcode as $(b_i,d_i)_{i=1}^n$, but reordering indices does not change the barcode itself. This leads to the diagonal action of $\mathrm{Sym}_n$ on $n$0 and the quotient

$n$1

Imposing $n$2 defines

$n$3

and the natural map $n$4 is a bijection, so $n$5 inherits topology and geometry from this quotient model (Brück et al., 2021).

The geometry of this quotient is controlled by the Coxeter complex of type $n$6. Writing $n$7, where $n$8 with $n$9 and

$\mathrm{Sym}_n$0

the hyperplanes $\mathrm{Sym}_n$1 restrict to reflecting hyperplanes on $\mathrm{Sym}_n$2. Their intersections with a sphere $\mathrm{Sym}_n$3 induce a triangulation equivariantly isomorphic to the Coxeter complex $\mathrm{Sym}_n$4. Chambers correspond to strict orderings

$\mathrm{Sym}_n$5

for unique $\mathrm{Sym}_n$6, while lower-dimensional faces encode coordinate equalities and hence parabolic subgroups (Brück et al., 2021).

This yields a stratification of barcode space. The top-dimensional strata are indexed by permutations associated to barcodes as defined by Kanari, Garin and Hess. More generally, strata correspond to marked double cosets of parabolic subgroups of $\mathrm{Sym}_n$7. For a barcode $\mathrm{Sym}_n$8, the minimal stratum is indexed by

$\mathrm{Sym}_n$9

where $n$0 preserves equalities among birth times, $n$1 preserves equalities among death times, and $n$2 is the associated double coset. In the strict case, $n$3, and one recovers the permutation type

$n$4

The same framework shows that $n$5 decomposes into regions consisting of barcodes with the same averages and standard deviations of birth and death times and the same permutation type (Brück et al., 2021).

A closely related viewpoint appears in the merge-tree inverse problem. For strict degree-$n$6 barcodes with one essential bar and $n$7 finite bars, death order is a permutation of birth order, so combinatorial barcode type is exactly an element of $n$8. Two strict barcodes are combinatorially equivalent precisely when they have the same associated permutation $n$9, and $$B=\{(b_i,d_i)\}_{i\in J},\qquad (b_i,d_i)\in\mathbb{R}^2,\ b_i<d_i,$$0 in this strict setting (Curry et al., 2021).

3. Combinatorial barcode lattices and inverse problems

A purely combinatorial version of symmetry barcodes arises when geometric positions and lengths are forgotten and only ordered multiplicity data are retained. Fix a multiplicity vector

$$B=\{(b_i,d_i)\}_{i\in J},\qquad (b_i,d_i)\in\mathbb{R}^2,\ b_i<d_i,$$1

and consider multipermutations of the multiset $$B=\{(b_i,d_i)\}_{i\in J},\qquad (b_i,d_i)\in\mathbb{R}^2,\ b_i<d_i,$$2 subject to the normalization that, for each $$B=\{(b_i,d_i)\}_{i\in J},\qquad (b_i,d_i)\in\mathbb{R}^2,\ b_i<d_i,$$3, the first occurrence of $$B=\{(b_i,d_i)\}_{i\in J},\qquad (b_i,d_i)\in\mathbb{R}^2,\ b_i<d_i,$$4 appears before the first occurrence of $$B=\{(b_i,d_i)\}_{i\in J},\qquad (b_i,d_i)\in\mathbb{R}^2,\ b_i<d_i,$$5. Equivalently, with refined labels $$B=\{(b_i,d_i)\}_{i\in J},\qquad (b_i,d_i)\in\mathbb{R}^2,\ b_i<d_i,$$6, the subsequence $$B=\{(b_i,d_i)\}_{i\in J},\qquad (b_i,d_i)\in\mathbb{R}^2,\ b_i<d_i,$$7 appears in that order and, within each label, $$B=\{(b_i,d_i)\}_{i\in J},\qquad (b_i,d_i)\in\mathbb{R}^2,\ b_i<d_i,$$8 appears in that order. The resulting set is denoted $$B=\{(b_i,d_i)\}_{i\in J},\qquad (b_i,d_i)\in\mathbb{R}^2,\ b_i<d_i,$$9 (Bouquet et al., 2023).

On $n$0 one imposes the weak Bruhat-type cover relation: $n$1 covers $n$2 if $n$3 is obtained from $n$4 by a single transposition of two distinct adjacent entries that are in increasing order in $n$5 and decreasing order in $n$6. The resulting graded poset is the combinatorial barcode lattice. Its structural symmetry is explicit: if $n$7, then $n$8 is isomorphic to the principal order ideal

$n$9

in weak Bruhat order, for a specific permutation $$\mathcal{B}_n=\{\text{barcodes with }n\text{ bars}\}.$$0 built from the block decomposition of $$\mathcal{B}_n=\{\text{barcodes with }n\text{ bars}\}.$$1. Consequently, the Möbius function of $$\mathcal{B}_n=\{\text{barcodes with }n\text{ bars}\}.$$2 is the restriction of the Möbius function of $$\mathcal{B}_n=\{\text{barcodes with }n\text{ bars}\}.$$3, and $$\mathcal{B}_n=\{\text{barcodes with }n\text{ bars}\}.$$4 is distributive if and only if

$$\mathcal{B}_n=\{\text{barcodes with }n\text{ bars}\}.$$5

(Bouquet et al., 2023).

This combinatorial lattice has several explicit invariants. Join-irreducible elements admit a closed counting formula, the rank-generating function is

$$\mathcal{B}_n=\{\text{barcodes with }n\text{ bars}\}.$$6

and maximal chains are counted by a hook-length formula associated with the Ferrers diagram $$\mathcal{B}_n=\{\text{barcodes with }n\text{ bars}\}.$$7. In the special case $$\mathcal{B}_n=\{\text{barcodes with }n\text{ bars}\}.$$8, the lattice identifies with perfect matchings on $$\mathcal{B}_n=\{\text{barcodes with }n\text{ bars}\}.$$9; intervals in the lattice recover permutational matchings and non-nesting matchings, with sizes $J=\{1,\dots,n\}$0 and $J=\{1,\dots,n\}$1, respectively (Bouquet et al., 2023).

The inverse problem for merge trees gives a second major combinatorial use of symmetric-group data. For a strict barcode $J=\{1,\dots,n\}$2, the tree realization number $J=\{1,\dots,n\}$3 is the number of merge trees whose $J=\{1,\dots,n\}$4-dimensional persistence barcode is exactly $J=\{1,\dots,n\}$5. If

$J=\{1,\dots,n\}$6

then

$J=\{1,\dots,n\}$7

When $J=\{1,\dots,n\}$8 has permutation type $J=\{1,\dots,n\}$9, one defines the left inversion vector

$(b_i,d_i)_{i=1}^n$0

and obtains

$(b_i,d_i)_{i=1}^n$1

Thus the tree-realization number depends only on permutation type. Summing over $(b_i,d_i)_{i=1}^n$2 yields

$(b_i,d_i)_{i=1}^n$3

the number of combinatorial merge trees with $(b_i,d_i)_{i=1}^n$4 leaves, which is also the number of maximal chains in the lattice of partitions of $(b_i,d_i)_{i=1}^n$5. This differs from the number of BHV phylogenetic-tree strata, $(b_i,d_i)_{i=1}^n$6, and formalizes a combinatorial distinction between merge-tree space and BHV tree space (Curry et al., 2021).

4. Coordinate systems and symmetry-aware invariants

One important development is the construction of explicit coordinates on barcode space that preserve permutation symmetry. For $(b_i,d_i)_{i=1}^n$7, define

$(b_i,d_i)_{i=1}^n$8

Away from the diagonal line $(b_i,d_i)_{i=1}^n$9, one has a bijection

$\mathrm{Sym}_n$0

given by $\mathrm{Sym}_n$1, equivariantly with respect to $\mathrm{Sym}_n$2. Applying this simultaneously to birth and death vectors, a barcode $\mathrm{Sym}_n$3 with not all births equal and not all deaths equal is uniquely determined by

$\mathrm{Sym}_n$4

These coordinates extend the invariant of Kanari–Garin–Hess and refine permutation type by adding mean and dispersion data (Brück et al., 2021).

A coarser but still symmetry-sensitive decomposition fixes the real parameters and the parabolic/double-coset data. The region containing $\mathrm{Sym}_n$5 consists of all barcodes $\mathrm{Sym}_n$6 with the same mean birth, mean death, birth spread, death spread, and the same

$\mathrm{Sym}_n$7

For strict barcodes this reduces to equality of $\mathrm{Sym}_n$8 and $\mathrm{Sym}_n$9. This suggests a geometric picture in which symmetry classes are determined by both continuous statistics and discrete order-type data (Brück et al., 2021).

A different coordinate program is given by tropical geometry. Barcode space

$n$00

is quotiented further by identifying zero-length intervals, and tropical functions are required to be invariant under permutation of interval pairs. A tropical rational function in $n$01 variables is 2-symmetric if it is invariant under permutations of the $n$02 pairs $n$03. Elementary 2-symmetric max-plus polynomials separate orbits in $n$04, and this supplies symmetry-respecting coordinate functions before quotienting by zero-length bars (Verovsek, 2016).

On the actual barcode space, the max-plus semiring $n$05 consists of those functions that are independent of $n$06 whenever $n$07. Its symmetric part is generated by

$n$08

where $n$09 is the sum of the lengths of the $n$10 longest bars. This yields stable length-only coordinates but does not separate barcodes that differ only in birth coordinates. To recover separating power, one introduces

$n$11

and the generators

$n$12

The resulting filtered tropical semiring $n$13 separates barcodes in $n$14 (Verovsek, 2016).

A fundamental limitation is that no finite subset of tropical functions on $n$15 respecting barcode equivalence separates all non-equivalent barcodes. The symmetry-aware coordinate algebras are therefore intrinsically infinite, and finite feature sets are necessarily application-dependent truncations (Verovsek, 2016).

5. Metrics, stability, and algorithmic dualities

Symmetry-aware parametrizations of barcode space lead naturally to quotient metrics. On

$n$16

the $n$17-norm induces the quotient metric

$n$18

where the minimum runs over representatives of the two orbits. On barcodes with exactly $n$19 bars this becomes the modified bottleneck distance

$n$20

Similarly, the $n$21-norm gives the modified Wasserstein distance

$n$22

These metrics differ from the usual bottleneck and Wasserstein distances in that both barcodes must have exactly $n$23 bars and matching to the diagonal is not allowed (Brück et al., 2021).

Tropical coordinates were designed to be stable with respect to the classical metrics that do allow diagonal matching. Every $n$24 satisfies

$n$25

and similarly

$n$26

The same style of Lipschitz estimate holds for every $n$27. The significance is methodological: symmetry-compatible coordinates are not only permutation invariant but also stable in the natural barcode metrics (Verovsek, 2016).

At the algorithmic level, symmetry enters barcode computation through a row–column duality in matrix reduction. The standard persistence algorithm reduces columns of the boundary matrix and reads barcode intervals from column pivot pairings $n$28. The row-based dual reduces rows and reads the same intervals from row pivot pairings $n$29. Both are detected by the same rank-increment quantity $n$30, so pivot pairings are intrinsic to the boundary operator rather than to the direction of elimination. This duality clarifies the symmetry between the clear and compress optimizations: clear is effective for column reduction on coboundary matrices, while compress becomes its true dual under row reduction on boundary matrices (Giunti, 2021).

In the persistent-symmetry setting, stability is formulated in interleaving terms. For persistence configurations $n$31 and $n$32, the associated symmetry modules satisfy

$n$33

and the barcode functor then yields

$n$34

For finite-type polybarcodes,

$n$35

and for persistence configurations,

$n$36

Thus both the ordinary-barcode and persistent-symmetry strands admit stability theorems, but the ambient categories and distances are different (Liu et al., 11 Aug 2025).

6. Persistent symmetry groups, polybarcodes, and other applications

In the direct sense of the term, symmetry barcodes arise from parameterized finite configurations in a metric space. Let $n$37 be a metric space and $n$38 the configuration category whose objects are $n$39-point configurations $n$40 with distinct points and whose morphisms are restrictions of global homeomorphisms of $n$41. The symmetry group of a configuration is

$n$42

A persistence $n$43-configuration is a functor

$n$44

For a morphism $n$45, conjugation does not in general restrict to a homomorphism $n$46, so the theory uses the subgroup

$n$47

and the span

$n$48

This produces a pseudofunctor from the configuration category to $n$49 and thereby a non-abelian persistence formalism (Liu et al., 11 Aug 2025).

For $n$50, the $n$51-persistent symmetry group is obtained from the span associated with $n$52. Linearization yields a persistence module, and in the discrete-time case the associated $n$53-module decomposes into free and torsion summands exactly as in ordinary persistence. A symmetry bar is the maximal interval $n$54 on which a symmetry born at $n$55 persists as a nontrivial symmetry. The symmetry barcode is the multiset of all such bars (Liu et al., 11 Aug 2025).

Polybarcodes refine this by tracking exact isometries rather than only interval-module generators. For a fixed isometry $n$56,

$n$57

is a closed subset of $n$58, called the polybar of $n$59. The polybarcode is the collection $n$60. This framework captures persistence, disappearance, and reappearance of the same symmetry transformation, which ordinary interval decompositions do not record directly (Liu et al., 11 Aug 2025).

The same work also formalizes quantitative asymmetry. The degree of symmetry is

$n$61

with $n$62 exactly when only the identity symmetry is present. Symmetry entropy records the distribution of element orders, and the symmetry degree polynomial packages the same information in generating-function form. Asymmetry is measured by the symmetry defect

$n$63

where $n$64 minimizes a Wasserstein-type matching cost to $n$65 while excluding the canonical matching $n$66. In Euclidean space, approximate symmetry sets $n$67 form compact finite approximate subgroups of $n$68, linking the theory to approximate group methods (Liu et al., 11 Aug 2025).

A separate, application-oriented use of the expression appears in two-dimensional visual coding. OAcode is an overall aesthetic 2D barcode in which the position detection pattern is canceled and detection is based on the pre-designed symmetrical data area. Its symmetry map is computed by an auto-convolution formula, its peaks are used with Hough transform and a homography model to recover perspective distortion, and an enhanced demodulation method addresses lens distortion. The reported experimental results state that when a $n$69 cm OAcode is captured with resolution $n$70 pixels, at screen-camera distance $n$71 cm and angle less or equal to $n$72, OAcode has $n$73 detection rate and $n$74 demodulation accuracy; for $n$75 cm OAcode, extraction at $n$76 cm with around $n$77 accuracy is reported (Ma et al., 2023).

The resulting picture is not a single unified theory but a family of mathematically adjacent constructions. Symmetry barcodes may denote barcodes of persistent symmetries of parameterized configurations, symmetry-controlled decompositions of barcode space itself, or engineered barcodes whose decoders exploit designed symmetry. What unifies these strands is the replacement of ad hoc invariance arguments by explicit algebraic, combinatorial, geometric, or categorical structure.