Local complementation: Difference between revisions

From HandWiki
OrgMain (talk | contribs)
add
 
TextAI2 (talk | contribs)
correction
 
Line 1: Line 1:
{{Short description|Operation in graph theory}}
{{Short description|Operation in graph theory}}
In [[Graph theory|graph theory]], the '''local complementation''' (also known as a '''vertex inversion''') of a simple undirected graph <math>G</math> at a vertex <math>v</math> is an operation that produces a new graph, denoted by <math>G \star v</math>. This operation is defined by toggling the adjacencies between all pairs of neighbors of <math>v</math> in <math>G</math>. Formally, two distinct vertices <math>x</math> and <math>y</math> are adjacent in <math>G \star v</math> if and only if exactly one of the following holds:
In [[Graph theory|graph theory]], '''local complementation''' (also known as '''vertex inversion''') is an operation on a graph that toggles adjacencies among the neighbours of a chosen vertex, while all other adjacencies remain unchanged. Despite its simple definition, it preserves interesting properties and generates a complex equivalence relation. The operation was introduced by Anton Kotzig and later studied in depth by André Bouchet and [[Biography:Dima Von-Der-Flaass|Von-Der-Flaass]].


# <math>x</math> and <math>y</math> are adjacent in G; or
Formally, the local complementation of a simple undirected graph <math>G</math> at a vertex <math>v</math> is an operation that produces a new graph, denoted by <math>G \star v</math>. This operation is defined by replacing the subgraph of <math>G</math> induced by <math>N_G(v)</math> with its complementary subgraph. In other words, two distinct vertices <math>x</math> and <math>y</math> are adjacent in the graph <math>G \star v</math> when exactly one of the following holds:
# both <math>x</math> and <math>y</math> are neighbours of <math>v</math> in <math>G</math>.


Equivalently, the local complementation at <math>v</math> replaces the subgraph of <math>G</math> induced by <math>N_G(v)</math> with its complementary subgraph.
# vertices <math>x</math> and <math>y</math> are adjacent in <math>G</math>; or
# both vertices <math>x</math> and <math>y</math> are neighbours of <math>v</math> in <math>G</math>.
Two graphs are said to be '''locally equivalent''' if one can be obtained from the other through a sequence of local complementations. This defines an equivalence relation on graphs, whose equivalence classes are known as '''local equivalence classes'''. For example, the star graph and complete graph on <math>n</math> vertices are locally equivalent, and they form a local equivalence class. The local equivalence classes on graphs with up to 12 vertices has been computed.<ref name=":1">{{Cite journal |last1=Cabello |first1=Adan |title=Optimal preparation of graph states |date=2011-04-16 |last2=Danielsen |first2=Lars Eirik |last3=Lopez-Tarrida |first3=Antonio J. |last4=Portillo |first4=Jose R. |journal=Physical Review A |volume=83 |issue=4 |article-number=042314 |doi=10.1103/PhysRevA.83.042314 |arxiv=1011.5464 |bibcode=2011PhRvA..83d2314C }}</ref> The size of a local equivalence class is at most <math>3^n</math>, and this collection of graphs can be enumerated efficiently.


The concept was introduced by Anton Kotzig and later studied in depth by André Bouchet and [[Biography:Dima Von-Der-Flaass|Von-Der-Flaass]].
== Applications ==


== Local equivalence ==
=== Structural graph theory ===
Two graphs are said to be '''locally equivalent''' if one can be obtained from the other through a sequence of local complementations. This defines an equivalence relation on graphs, whose equivalence classes are known as '''local equivalence classes'''. For a positive integer <math>n</math>, the star graph and complete graph on <math>n</math> vertices forms a trivial local equivalence class. The local equivalence classes on graphs up to 12 vertices is known....<ref name=":1">{{Citation |last1=Cabello |first1=Adan |title=Optimal preparation of graph states |date=2011-04-16 |last2=Danielsen |first2=Lars Eirik |last3=Lopez-Tarrida |first3=Antonio J. |last4=Portillo |first4=Jose R. |journal=Physical Review A |volume=83 |issue=4 |article-number=042314 |doi=10.1103/PhysRevA.83.042314 |arxiv=1011.5464 |bibcode=2011PhRvA..83d2314C }}</ref>
The [[Robertson–Seymour theorem]] states that the graph minor relation is a [[Well-quasi-ordering|well-quasi ordering]]. It was proved in a series of twenty papers spanning over 500 pages from 1983 to 2004. The algorithmic consequences are vast - together with an efficient algorithm for graph minor testing, the result provides efficient algorithms for solving a range of computational problems where the optimal value is monotonic in the graph minor relation.  


Using ideas from isotropic subsystems, Bouchet<ref name=":0">{{Cite journal |last=Bouchet |first=André |date=1991-12-01 |title=An efficient algorithm to recognize locally equivalent graphs |url=https://doi.org/10.1007/BF01275668 |journal=Combinatorica |language=en |volume=11 |issue=4 |pages=315–329 |doi=10.1007/BF01275668 |issn=1439-6912|url-access=subscription }}</ref> introduced a <math>\mathcal O(n^3)</math> algorithm to determine whether two graphs are locally equivalent. Additionally, Fon-Der-Flass shows that all locally equivalent graphs can be reached after a sequence of at most <math>\max(n+1, 10n/9)</math> local complementations (see page 55 of "Local complementations of simple and directed graphs").
Local complementations are central to the vertex-minor relation, which shares many similarities with the graph minor relation. Better understanding of the local complementation operation could extend the Robertson–Seymour theorem to prove that the vertex-minor relation is also a well-quasi ordering.


Bouchet and Fon-Der-Flass independently proved Mulder's conjecture that locally equivalent trees are isomorphic. Furthermore, Bouchet gives a simple closed form expression for number of such trees (only leaves and their neighbours can be swapped), and Fon-Der-Flass proves a stronger statement that if a graph <math>G</math> is locally equivalent to a tree <math>T</math>, then <math>G</math> has a subgraph isomorphic to <math>T</math><ref>{{Cite journal |last=Bouchet |first=André |date=1988 |title=Transforming trees by successive local complementations |url=https://onlinelibrary.wiley.com/doi/abs/10.1002/jgt.3190120210 |journal=Journal of Graph Theory |language=en |volume=12 |issue=2 |pages=195–207 |doi=10.1002/jgt.3190120210 |issn=1097-0118|url-access=subscription }}</ref>
=== Measurement based quantum computing ===
For a given the [[Graph state|graph state]] <math>|G\rang</math>, the action of the [[Clifford gate|local Clifford operation]] is equivalent<ref>{{Cite journal |last1=Ji |first1=Z.-F. |last2=Chen |first2=J.-X. |last3=Wei |first3=Z.-H. |last4=Ying |first4=M.-S. |date=January 2010 |title=The LU-LC conjecture is false |url=http://www.rintonpress.com/journals/doi/QIC10.1-2-8.html |journal=Quantum Information and Computation |volume=10 |issue=1&2 |pages=97–108 |doi=10.26421/QIC10.1-2-8|url-access=subscription }}</ref> to the local complementation transformation on the graph <math>G</math>. The study of graph states that are locally equivalent is relevant to building quantum circuits in measurement based quantum computing (MBQC)<ref name=":1" />


Fon-Der-Flass proved that graphs locally equivalent to the cycle graph contain a Hamiltonian cycle.
Similarly, local complementation is also related to state preparation in [[Linear optical quantum computing|photonic quantum computing (PQC)]].
 
== Properties ==
# The local complement operation is self inverse; that is, <math>(G\star v)\star v=G</math>.
# Local complementations commute only when vertices are non-adjacent; that is, <math>(G\star v)\star w = (G\star w)\star v</math> unless <math>vw \in E(G)</math>.
# Connected components are preserved by the local complementation operation, so it is common analyse each component of the graph independently. So without loss of generality, all graphs will be assumed to be connected.
# All locally equivalent graphs can be reached after a sequence of at most <math>\max(n+1, 10n/9)</math> local complementations<ref name=":3">{{Cite book |last=Koršunov |first=Aleksej D. |title=Discrete Analysis and Operations Research |date=1996 |publisher=Springer Netherlands |isbn=978-94-010-7217-5 |series=Mathematics and Its Applications |location=Dordrecht}}</ref>.
# Locally equivalent trees are isomorphic (Mulder's conjecture), and there exists a simple expression to count such trees.<ref name=":5">{{Cite journal |last=Bouchet |first=André |date=1988 |title=Transforming trees by successive local complementations |url=https://onlinelibrary.wiley.com/doi/abs/10.1002/jgt.3190120210 |journal=Journal of Graph Theory |language=en |volume=12 |issue=2 |pages=195–207 |doi=10.1002/jgt.3190120210 |issn=1097-0118|url-access=subscription }}</ref>
# If a graph is locally equivalent to a tree, it has a subgraph isomorphic to that tree.<ref name=":6">{{Cite web |last=Fon-Der-Flaas |first=D. G. |date=1988 |title=On local complementations of graphs |url=https://zbmath.org/0728.05048 |access-date=2026-02-26 |website=zbmath.org}}</ref>
# If a graph is locally equivalent to the cycle graph, it contains a Hamiltonian cycle.<ref name=":6" />
# If a graph is locally equivalent to its complement, it is [[Self-complementary graph|self-complementary]].<ref name=":6" />
# The number of essentially different ways of transforming one graph into another via local complementation is <math>0, 3, 6</math> or <math>2^k</math> for some <math>k \ge 0</math>.<ref name=":6" />
# Any class of locally equivalent distance-hereditary graphs is equal to the class of the circle graphs of the Euler tours of some [[Quartic graph|4-regular graph]].
# Locally equivalent graphs have the same [[Rank-width|rank-width]].
# Counting locally equivalent graphs is [[♯P-complete|#P-complete]].<ref>{{Cite journal |last1=Dahlberg |first1=Axel |title=Counting single-qubit Clifford equivalent graph states is #P-complete |date=2019-07-18 |arxiv=1907.08024 |last2=Helsen |first2=Jonas |last3=Wehner |first3=Stephanie |journal=Journal of Mathematical Physics |volume=61 |issue=2 |article-number=022202 |doi=10.1063/1.5120591 }}</ref>
# Determining the minimum degree that can be reached by means of local complementation is [[NP-completeness|NP-complete]] and [[APX|APX-hard]], and can be computed by an [[Circle graph|<math>\mathcal O^*(1.938^n)</math>]] algorithm.<ref>{{cite book |last1=Javelle |first1=Jérôme |title=Graph-Theoretic Concepts in Computer Science |date=2012-04-20 |last2=Mhalla |first2=Mehdi |last3=Perdrix |first3=Simon |chapter=On the Minimum Degree up to Local Complementation: Bounds and Complexity |series=Lecture Notes in Computer Science |volume=7551 |pages=138–147 |doi=10.1007/978-3-642-34611-8_16 |arxiv=1204.4564 |isbn=978-3-642-34610-1 }}</ref><ref>{{cite book |last1=Cattanéo |first1=David |title=Algorithms and Computation |date=2016-08-17 |arxiv=1503.04702 |last2=Perdrix |first2=Simon |chapter=Minimum Degree up to Local Complementation: Bounds, Parameterized Complexity, and Exact Algorithms |series=Lecture Notes in Computer Science |volume=9472 |pages=259–270 |doi=10.1007/978-3-662-48971-0_23 |isbn=978-3-662-48970-3 }}</ref>
# Determining the minimum total edge weight by means of local complementation is NP-complete.<ref>{{Cite journal |last1=Sharma |first1=Hemant |title=Minimising the number of edges in LC-equivalent graph states |date=2026-01-27 |arxiv=2506.00292 |last2=Goodenough |first2=Kenneth |last3=Borregaard |first3=Johannes |last4=Rozpędek |first4=Filip |last5=Helsen |first5=Jonas |journal=Quantum |volume=10 |article-number=2001 |doi=10.22331/q-2026-02-09-2001 }}</ref>
# Determining the minimum number of edges by means of local complementation is likely NP-complete.


== Relation to quantum computing ==
== Pivot operation ==
If <math>|G\rang</math> represents a graph state in quantum computing, the action of the [[Clifford gate|local Clifford operation]] on the [[Graph state|graph state]] is roughly equivalent.<ref>{{Cite journal |last1=Ji |first1=Z.-F. |last2=Chen |first2=J.-X. |last3=Wei |first3=Z.-H. |last4=Ying |first4=M.-S. |date=January 2010 |title=The LU-LC conjecture is false |url=http://www.rintonpress.com/journals/doi/QIC10.1-2-8.html |journal=Quantum Information and Computation |volume=10 |issue=1&2 |pages=97–108 |doi=10.26421/QIC10.1-2-8|url-access=subscription }}</ref> to the local complementation transformation on the graph. The study of graph states that are locally Clifford equivalent graphs is relevant to building quantum circuits in measurement based quantum computing (MBQC)<ref name=":1" />
Local complementations do not commute for adjacent vertices, motivating the following operation that performs complementations at adjacent vertices.


Similarly, local complementation is also related to state preparation in [[Linear optical quantum computing|photonic quantum computing (PQC)]].  
For adjacent vertices <math>x</math> and <math>y</math>, the '''pivot operation''' (also historically known as an edge complementation) is defined as<math display="block">G\wedge xy = G\star x\star y\star x.</math>It can be shown that <math>G\star x\star y\star x = G\star y\star x\star y</math>, and hence the pivot operation is well defined. Alternatively, the graph <math>G\wedge xy</math> can be obtained from <math>G</math> by toggling adjacencies between every pair of vertices in two different sets among <math>N_G(x) - (N_G(y) \cup {y})</math>, <math>N_G(x) \cap N_G(y)</math> and <math>N_G(y) - (N_G(x) \cup {x})</math> then switching the labels <math>x</math> and <math>y</math>. The pivot operation satisfies the identities <math>G \wedge xy \wedge xy = G</math> and <math>G \wedge xy \wedge yz = G \wedge xz</math><ref>{{Cite journal |last=Oum |first=Sang-il |date=2005-09-01 |title=Rank-width and vertex-minors |url=https://www.sciencedirect.com/science/article/pii/S0095895605000389 |journal=Journal of Combinatorial Theory, Series B |volume=95 |issue=1 |pages=79–100 |doi=10.1016/j.jctb.2005.03.003 |issn=0095-8956|url-access=subscription }}</ref>.


== Properties ==
Two graphs are said to be '''pivot equivalent''' if one can be obtained from the other through a sequence of pivot operations. Since pivot operations consist of local complementation operations, pivot equivalent graphs are locally equivalent. The converse is true for bipartite graphs<ref name=":3" />, but is not generally true. If <math>G</math> and <math>H</math> are pivot equivalent graphs, there are pairwise disjoint edges <math>e_1, e_2, \dots, e_k</math> such that <math>G \wedge e_1 \wedge e_2 \dots\wedge e_k = H</math>. Fon-Der-Flass proved that if graphs <math>G</math> and <math>H</math> are locally equivalent, they are pivot equivalent to some <math>G'</math> and <math>H'</math> respectively such that <math>G' * v_1 v_2 \dots v_k = H'</math> and <math>\{v_1, v_2, \dots, v_k\}</math> is an independent set in <math>G'</math> and <math>H'</math><ref>{{Cite journal |last=D. G. |first=Fon-Der-Flaass |date=1989 |title=Distance between locally equivalent graphs |journal=Metody Diskret. Analiz |volume=48 |pages=85–94 |via=Novosibirsk: Institute of Mathematics of the USSR Academy of Sciences}}</ref>.
# The local complement operation is self inverse <math>(G\star v)\star v=G</math>.


# Local complementations do not, in general, commute; that is, <math>(G\star v)\star w \ne (G\star w)\star v</math> in general.
The size of a pivot equivalence class is at most <math>2^n</math>, and this collection of graphs can be enumerated efficiently.
# Connected components are preserved by the local complementation operation, so it is common analyse each component of <math>G</math> independently. Without loss of generality, we assume that <math>G</math> is connected.
# Any class of locally equivalent distance-hereditary graphs is equal to the class of the circle graphs of the Euler tours of some [[Quartic graph|<math>4</math>-regular graph]].
# Dahlberg, Helsen, and Wehner<ref>{{Citation |last1=Dahlberg |first1=Axel |title=Counting single-qubit Clifford equivalent graph states is #P-complete |date=2019-07-18 |arxiv=1907.08024 |last2=Helsen |first2=Jonas |last3=Wehner |first3=Stephanie |journal=Journal of Mathematical Physics |volume=61 |issue=2 |article-number=022202 |doi=10.1063/1.5120591 }}</ref> proved that counting locally equivalent graphs is [[♯P-complete|#P-complete]] by reducing this problem on circle graphs to the problem of counting Eulerian circuits in a 4-regular graph.


== Relation to circle graphs ==
Pivot equivalence has been studied using even binary [[Delta-matroid|delta-matroids]]<ref>{{Cite journal |last1=Bouchet |first1=A. |last2=Duchamp |first2=A. |date=1991-02-15 |title=Representability of △-matroids over GF(2) |url=https://dx.doi.org/10.1016/0024-3795%2891%2990020-W |journal=Linear Algebra and Its Applications |volume=146 |pages=67–78 |doi=10.1016/0024-3795(91)90020-W |issn=0024-3795|url-access=subscription }}</ref>.
If <math>G</math> is a '''circle graph''', then performing a local complementation at <math>v</math> corresponds to flipping one side of the circle divided by the chord representing <math>v</math> on the chord diagram<ref name=":0" /><ref>{{Cite journal |last=Bouchet |first=André |date=1993-04-28 |title=Recognizing locally equivalent graphs |url=https://dx.doi.org/10.1016/0012-365X%2893%2990357-Y |journal=Discrete Mathematics |volume=114 |issue=1 |pages=75–86 |doi=10.1016/0012-365X(93)90357-Y |issn=0012-365X|url-access=subscription }}</ref>. The class of circle graphs are hence closed under local complementation, and they are also closed under taking vertex-minors.


== Invariants ==
== Invariants ==


=== Cut-rank function ===
=== Cut-rank function ===
For a graph <math>G</math> with vertex set <math>V</math>, the cut-rank function (also historically known as the connectivity function) is denoted <math>\rho_G: 2^V \to \mathbb Z_{\ge 0}</math>. It is defined over the vertex subsets <math>X \subseteq V</math> such that <math>\rho_G(X)</math>
As a graph undergoes local complementation, its adjacency matrix changes in a well defined way. In particular, the entries that change are exactly some sub-matrix (excluding the main diagonal). Hence, it may be natural to study the local complementation operation using linear algebra.
is the rank of the [[Adjacency matrix#Of a bipartite graph|bi-adjacency matrix]] of the partition <math>(X, V-X)</math>, defined over the [[Finite field|finite field]] [[GF(2)]]. That is, the rank of the <math>X\times(V-X)</math> binary matrix <math>M</math> where <math>M_{uv} = 1</math> if and only if <math>uv</math> is an edge in <math>G</math>.


Since the rank of a matrix is preserved by elementary row and column operations, a straightforward argument shows that locally equivalent graphs have identical cut-rank functions. However, the converse is not true - a counterexample can be constructed using labelled [[Petersen graph|Petersen graphs]]. It is known that [[Bipartite graph|bipartite graphs]] with identical cut-rank functions are pivot-equivalent<ref>{{Cite journal |last=Seymour |first=P. D |date=1988-08-01 |title=On the connectivity function of a matroid |url=https://dx.doi.org/10.1016/0095-8956%2888%2990052-4 |journal=Journal of Combinatorial Theory, Series B |volume=45 |issue=1 |pages=25–30 |doi=10.1016/0095-8956(88)90052-4 |issn=0095-8956|url-access=subscription }}</ref>
For a graph <math>G</math> with vertex set <math>V</math>, the cut-rank function (also historically known as the connectivity function) is denoted <math display="inline">\rho_G: \mathcal P(V) \to \mathbb N</math>. It is defined over the vertex subsets <math>X \subseteq V</math> such that <math>\rho_G(X)</math> is the rank of the [[Adjacency matrix#Of a bipartite graph|bi-adjacency matrix]] of the partition <math>(X, V-X)</math>, defined over the [[Finite field|finite field]] [[GF(2)]]. That is, the rank of the <math>X\times(V-X)</math> binary matrix <math>M</math> where <math>M_{uv} = 1</math> when <math>uv</math> is an edge in <math>G</math>. Intuitively, <math>\rho_G(X)</math> is a measure of how complex the connectivity is between <math>X</math> and the remaining vertices.


=== Totally isotropic subsystems ===
Since the rank of a matrix is preserved by elementary row and column operations, a straightforward argument shows that locally equivalent graphs have identical cut-rank functions. However, the converse is not true - a counterexample can be constructed using labelled [[Petersen graph|Petersen graphs]]. It is known that [[Bipartite graph|bipartite graphs]] with identical cut-rank functions are pivot-equivalent<ref>{{Cite journal |last=Seymour |first=P. D |date=1988-08-01 |title=On the connectivity function of a matroid |url=https://dx.doi.org/10.1016/0095-8956%2888%2990052-4 |journal=Journal of Combinatorial Theory, Series B |volume=45 |issue=1 |pages=25–30 |doi=10.1016/0095-8956(88)90052-4 |issn=0095-8956|url-access=subscription }}</ref>, and so locally equivalent bipartite graphs are also pivot-equivalent.
Bouchet develops a structure using the Klein four-group to represent a class of locally equivalent graphs. Each locally equivalent graph corresponds to some graphical representations of the same totally isotropic system. This leads to results about determining local equivalence and enumerating locally equivalent graphs.


=== Global interlace polynomial ===
The cut-rank function is [[Submodular set function|submodular]] since it can be shown that <math>\rho_G(X)+\rho_G(Y) \ge \rho_G(X\cup Y)+\rho_G(X\cap Y)</math> for any vertex set <math>X</math> and <math>Y</math>. However, it is not [[Monotonic function|monotone]].
The '''global interlace polynomial''' <math>Q(G, x)</math>, also known as the Tutte-Martin polynomial, is a polynomial that corresponds to a simple graph <math>G</math>. It is defined recursively as <math>Q(G, x) = \begin{cases}
 
    Q(G-u, x) + Q((G*u)-u, x) + Q((G*uvu)-u, x) & uv \in E(G) \\
=== Local sets ===
    x^n & G \text{ has no edges}
The cut-rank can be considered 'most interesting' when it is either rank 1 or full rank. Since the cut-rank is invariant over local complementation, so are the sets with a particular cut-rank. The sets of rank 1 are studied using canonical split decompositions in a later section. Here, we define a local set as a vertex set which does not have full cut-rank.
\end{cases}</math>
 
Define <math display="block">\mathrm{Odd}_G(X) = \{v \in V(G) \mid |N(v)\cap X| = 1 \bmod 2 \}.</math>A set of vertices <math>S</math> is '''local''' in <math>G</math> if <math>S = X \cup \mathrm{Odd}_G(X)</math> for some subset <math>X \subseteq V(G)</math>. Now if <math>S</math> is local in <math>G</math>, it is also local in any graph locally equivalent to <math>G</math>. Local sets can equivalently be defined as the sets of vertices which do not have full cut-rank (i.e. the sets <math>S</math> where <math>\rho_G(S) < |S|</math>). Intuitively, a local set is some linear combination of the closed neighbourhoods of <math>G</math>. Local sets are used to study the degrees of vertices in <math>G</math>.<ref>{{Cite journal |last=Høyer |first=Peter |last2=Mhalla |first2=Mehdi |last3=Perdrix |first3=Simon |date=2006 |editor-last=Asano |editor-first=Tetsuo |title=Resources Required for Preparing Graph States |url=https://link.springer.com/chapter/10.1007/11940128_64?error=cookies_not_supported&code=bcd69664-6723-4241-80fa-838f1fd751cd |journal=Algorithms and Computation |language=en |location=Berlin, Heidelberg |publisher=Springer |pages=638–649 |doi=10.1007/11940128_64 |isbn=978-3-540-49696-0}}</ref>
 
This invariant may be a helpful tool to quickly show that graphs are not locally equivalent. However, there are graphs with the same local sets which are not locally equivalent, since there are graphs with identical cut-rank functions which are not locally equivalent.<ref name=":4">{{Cite journal |last=Bouchet |first=André |date=1993-04-28 |title=Recognizing locally equivalent graphs |url=https://dx.doi.org/10.1016/0012-365X%2893%2990357-Y |journal=Discrete Mathematics |volume=114 |issue=1 |pages=75–86 |doi=10.1016/0012-365X(93)90357-Y |issn=0012-365X|url-access=subscription }}</ref>
 
=== Totally isotropic subspaces ===
The richest description of a class of locally equivalent graphs is an extension of the local sets idea, involving a linear algebra structure over the [[Klein four-group]]. Each locally equivalent graph, equipped with two specific vectors, corresponds to some graphic presentations of the same totally isotropic subspace.
 
Let <math>K_4</math> be the Klein four-group. A vector <math>\mathbf a \in K_4^n</math> is said to be '''complete''' if <math>\mathbf a_i</math> is nonzero for every <math>1 \le i \le n</math>. Two vectors <math>\mathbf a, \mathbf b \in K_4^n</math> are said to be '''supplementary''' if <math>\mathbf a_i</math> and <math>\mathbf b_i</math> are nonzero and distinct for every <math>1 \le i \le n</math>. For a set <math>S</math>, define <math>\mathbf a[S]</math> as the vector where <math>\mathbf a[S]_i = a_i</math> if <math>i \in S</math>, and <math>\mathbf a[S]_i = 0</math> otherwise.


Now if <math>G</math> and <math>H</math> are locally equivalent, <math>Q(G, x) = Q(H, x)</math> for any <math>x</math>. Additionally, <math>Q(G, 2)</math> is related to the number of graphs locally equivalent to <math>G</math>.
A subspace <math>L</math> of <math>K_4^n</math> is called a '''totally isotropic subspace''' if <math>\dim(L) = \dim(K_4^n)/2 = n</math>, and every two complete vectors in <math>L</math> are not supplementary.


Let <math>A</math> be the adjacency matrix of a graph <math>G</math> over the binary field. For a subset <math>S</math> of <math>V(G)</math>, we write <math>A[S]</math> to denote the <math>S\times S</math> submatrix of <math>A</math>. Let <math>I_X</math> be the <math>V(G)\times V(G)</math> diagonal matrix over the binary field such that the <math>(v,v)</math>-entry is 1 if and only if <math>v \in X</math>. The global interlace polynomial can equivalently be defined as 
A '''graphic presentation''' of a totally isotropic subspace <math>L</math> is a triple <math>(G, \mathbf a, \mathbf b)</math> where <math>G</math> is a simple graph on the vertex set <math>V=\{1, \dots, n\}</math> and <math>\mathbf a</math> and <math>\mathbf b</math> are supplementary vectors of <math>K_4^n</math>, such that <math>L</math> is spanned by the set <math display="block">\{\mathbf a[N(v)]+ \mathbf b[\{v\}] \mid v \in V\}.</math>For a fixed <math>L</math>, there is a one-to-one correspondence between every graphic presentation <math>(G, \mathbf a, \mathbf b)</math> and every complete vector <math>\mathbf a</math> where <math>\mathbf a \not\in L</math>, the vectors <math>\mathbf a</math> here are known are '''Eulerian vectors'''. Furthermore, if <math>(G, \mathbf a, \mathbf b)</math> is a graphic presentation of <math>L</math>, so is <math display="block">(G\star v, \mathbf a+\mathbf b[\{v\}], \mathbf b+\mathbf a[N(v)])</math> and <math display="block">(G\wedge vw, \mathbf a[V\setminus\{v,w\}]+\mathbf b[\{v,w\}], \mathbf a[\{v,w\}]+\mathbf b[V\setminus\{v,w\}]).</math>The '''fundamental graphs''' of <math>L</math> form a local equivalence class. These facts leads to important results about determining local equivalence and locally equivalent class sizes, which is discussed in the next 2 sections.


<math>Q(G, x) = \sum_{R \subseteq S \subseteq V(G)} (x-2)^{\mathrm{nullity}((A+I_R)[S])}</math>
This structure can be considered as an extension of the local sets structure, since <math>L = \{\mathbf a[\mathrm{Odd}_G(X)]+\mathbf b[X] \mid X \subseteq V\}</math>, and when <math>\mathbf a=\mathbf b</math>, <math>L = \{\mathbf a[S] \mid S \text{ is local in } G\}</math>.<ref>{{Cite journal |last=Bouchet |first=André |date=1988-08-01 |title=Graphic presentations of isotropic systems |url=https://www.sciencedirect.com/science/article/pii/009589568890055X |journal=Journal of Combinatorial Theory, Series B |volume=45 |issue=1 |pages=58–76 |doi=10.1016/0095-8956(88)90055-X |issn=0095-8956|url-access=subscription }}</ref>


This polynomial has a nice formula in certain cases, such as when G is locally equivalent to a cycle or a tree.
=== Characterisation of local equivalence ===
The following characterisations of locally equivalent graphs are straightforward results using totally isotropic subspaces. This leads to a efficient algorithm to recognise locally equivalent graphs.


It is interesting to note a couple of similarities to the canonical [[Tutte polynomial]]. In particular, the recurrence looks similar to the [[Deletion–contraction formula|deletion-contraction formula]], and both polynomials can be formulated as a sum of terms raised to the power of a matrix rank.
Let <math>G_1</math> and <math>G_2</math> be simple graphs and fix supplementary vectors <math>\mathbf a_1</math> and <math>\mathbf b_1</math>. he graphs <math>G_1</math> and <math>G_2</math> are locally equivalent if and only if there are supplementary vectors <math>\mathbf a_2</math> and <math>\mathbf b_2</math> such that <math>(G_1, \mathbf a_1, \mathbf b_1)</math> and <math>(G_2, \mathbf a_2, \mathbf b_2)</math> are graphic presentations of the same totally isotropic subspace. The existence of such vectors <math>\mathbf a_2</math> and <math>\mathbf b_2</math> can be determined in <math>\mathcal O(n^3)</math> by solving a system of linear equations<ref name=":0">{{Cite journal |last=Bouchet |first=André |date=1991-12-01 |title=An efficient algorithm to recognize locally equivalent graphs |url=https://doi.org/10.1007/BF01275668 |journal=Combinatorica |language=en |volume=11 |issue=4 |pages=315–329 |doi=10.1007/BF01275668 |issn=1439-6912|url-access=subscription }}</ref>. A modification to this algorithm can, with the same time complexity, recover a sequence of at most <math>3n/2</math> local complementations transforming <math>G_1</math> into <math>G_2</math><ref>{{Citation |last=Fon-Der-Flaass |first=D. G. |title=Local Complementations of Simple and Directed Graphs |date=1996 |work=Discrete Analysis and Operations Research |pages=15–34 |editor-last=Korshunov |editor-first=Alekseǐ D. |url=https://doi.org/10.1007/978-94-009-1606-7_3 |access-date=2026-02-26 |place=Dordrecht |publisher=Springer Netherlands |language=en |doi=10.1007/978-94-009-1606-7_3 |isbn=978-94-009-1606-7|url-access=subscription }}</ref>.


=== Local sets ===
An equivalent characterisation can be formulated using vertex sets. Let <math>N_1</math> and <math>N_2</math> denote the neighbourhood functions of <math>G_1</math> and <math>G_2</math> respectively. Then graphs <math>G_1</math> and <math>G_2</math> are locally equivalent if and only if there are vertex subsets <math>X, Y, Z, T</math> such that for every pair of vertices <math>v, w</math>, <math display="block">|X\cap N_1(v)\cap N_2(w)|+|Y\cap N_1(v)\cap \{w\}|+|Z\cap N_2(w)\cap\{v\}|+|T\cap\{v,w\}| \equiv 0 \pmod 2,</math>and the [[Symmetric difference|symmetric difference]] of <math>X\cap T</math> with <math>Y\cap Z</math> is the entire vertex set.<ref name=":0" />
Define <math>Odd_G(D) = \{v \in V(G) \mid |N(v)\cap D| = 1 \bmod 2 \}</math>. We say that a set of vertices <math>L</math> is '''local''' if <math>L = X \cup Odd_G(X)</math> for some subset <math>X \subseteq L</math>. Now if <math>L</math> is local in <math>G</math>, it is local in any graph locally equivalent to <math>G</math>. Local sets can equivalently be defined as the sets which do not have full cut-rank.


This invariant may be a helpful tool to quickly show that graphs are not locally equivalent. However, there are graphs with the same local sets which are not locally equivalent.
=== Global interlace polynomial ===
The '''global interlace polynomial''' (also known as the Tutte-Martin polynomial), is a polynomial that corresponds to a simple graph <math>G</math>. It is defined recursively as<math display="block">Q(G;x) = \begin{cases}
    Q(G-u;x) + Q((G*u)-u; x) + Q((G \wedge uv)-u; x) & uv \in E(G) \\
    x^n & G \text{ has no edges and } n \text{ vertices}
\end{cases}</math>Now if <math>G</math> and <math>H</math> are locally equivalent, <math>Q(G, x) = Q(H, x)</math> for any <math>x</math><ref name=":4" />. Additionally,


=== Canonical split decomposition ===
* <math>Q(G, 2)</math> is a multiple of the number of graphs locally equivalent to <math>G</math>. In particular, if <math>G</math> is a fundamental graph of a totally isotropic subspace <math>L</math>, this gives the number of Eulerian vectors in <math>L</math>.
A '''split''' of a graph <math>G</math> is a partition <math>(X, Y)</math> of the vertex set of <math>G</math> such that <math>|X|, |Y| \ge 2</math> and <math>\rho_G(X) = 1</math>.
* <math>Q(G; 4)/2^n</math> is the number of [[Induced subgraph|induced]] Eulerian subgraphs in <math>G</math>. (An Eulerian graph contains only vertices of even degree).


If a graph admits a split, it can be built by the '''1-join''' of two graphs. The 1-join of two graphs <math>G_1</math> and <math>G_2</math> with marker vertices <math>v_1 \in V(G_1)</math> and <math>v_2 \in V(G_2)</math> is defined to be the graph obtained from the disjoint union of <math>G_1-v_1</math> and <math>G_2-v_2</math> by adding an edge between every neighbour of <math>v_1</math> in <math>G_1</math> and every neighbour of <math>v_2</math> in <math>G_2</math>.
The polynomial has closed-form formulas in certain cases:


The '''canonical split decomposition''' is constructed as follows: Start from a set <math>\{G\}</math> consisting of a single graph. Repeatedly pick a graph <math>H</math> that admits a split. We then replace <math>H</math> with <math>2</math> smaller graphs whose 1-join reconstructs <math>H</math>. This process is applied only when <math>H</math> is neither a complete graph nor a star. We can associate a tree by having one node for each graph in the resulting set and adding edges between corresponding marker vertices.
* The [[Path graph|path graphs]] <math>P_n</math> have the closed form <math>Q(P_n; x) = \lambda_{1}\left(1+\sqrt{1+x}\right)^{n-1}+\lambda_{2}\left(1-\sqrt{1+x}\right)^{n-1}</math> where <math>\lambda_1=x/2+x/\sqrt{1+x}</math> and <math>\lambda_2=x/2-x/\sqrt{1+x}</math>.
* More generally, a tree <math>T</math> has the global interlace polynomial <math display="inline">Q(T; x) = \sum_{k\ge0} \Phi_k(T) 2^{n-2k}x^k</math>, where <math>\Phi_k(T)</math> is the number of [[Matching (graph theory)|matchings]] of size <math>k</math> in <math>T</math>.
* The [[Cycle graph|cycle graphs]] <math>C_n</math> have the closed form <math>Q(C_n; 2) = (1+\sqrt3)^n+(1-\sqrt3)^n-4(2^{n-1}+(-1)^n)/3</math><ref name=":4" />.


The canonical split decomposition is unique and has nice properties regarding local complementation.
An equivalent definition of the global interlace polynomial involves a summation over subsets of vertex sets. Let the graph <math>G</math> have the adjacency matrix <math>A</math> over the binary field. For a vertex set <math>S \in V(G)</math>, write <math>A[S]</math> to denote the <math>S\times S</math> sub-matrix of <math>A</math>. Let <math>I_X</math> be the <math>V(G)\times V(G)</math> diagonal matrix over the binary field such that the <math>(v,v)</math>-entry is 1 when <math>v \in X</math>, and 0 otherwise. The global interlace polynomial can equivalently be defined as<math display="block">Q(G; x) = \sum_{R \subseteq S \subseteq V(G)} (x-2)^{\mathrm{nullity}((A+I_R)[S])}.</math>There are some interesting similarities to the canonical [[Tutte polynomial]]. In particular, the recurrence looks similar to the [[Deletion–contraction formula|deletion-contraction formula]], and both polynomials can be formulated as a sum of terms raised to the power of a matrix rank. Isomorphic graphs have the same Tutte polynomial, while locally equivalent or isomorphic graphs have the same global interlace polynomial.


== Vertex-minor relation ==
=== Canonical split decomposition ===
A graph <math>H</math> is a '''vertex-minor''' of a graph <math>G</math> if <math>H</math> is an [[Induced subgraph|induced subgraph]] of a graph locally equivalent to <math>G</math>. The name ‘vertex-minor’ was coined by Oum but it appeared previously under various names such as l-reduction and i-minor<ref name=":2">{{Cite journal |last=Bouchet |first=André |date=1987-08-01 |title=Unimodularity and circle graphs |url=https://dx.doi.org/10.1016/0012-365X%2887%2990132-4 |journal=Discrete Mathematics |volume=66 |issue=1 |pages=203–208 |doi=10.1016/0012-365X(87)90132-4 |issn=0012-365X|url-access=subscription }}</ref>
A '''split''' of a graph <math>G</math> is a partition of its vertex set <math>(X, Y)</math> such that <math>|X|, |Y| \ge 2</math> and <math>\rho_G(X) = 1</math>. This happens exactly when the connectivity between <math>X</math> and <math>Y</math> is a complete bipartite graph, and complete bipartite graphs have a simple known structure over local complementation.


Deciding whether H is a vertex-minor of G for two input graphs G and H with is [[NP-completeness|NP-complete]], even if H is a [[Complete graph|complete graph]] and G is a [[Circle graph|circle graph]].<ref>{{Citation |last1=Dahlberg |first1=Axel |title=The complexity of the vertex-minor problem |date=2019-06-12 |arxiv=1906.05689 |last2=Helsen |first2=Jonas |last3=Wehner |first3=Stephanie}}</ref>. However, for each fixed circle graph <math>H</math>, there is an <math>\mathcal O(n^3)</math>-time algorithm to decide whether an input <math>n</math>-vertex graph <math>G</math> contains a vertex-minor [[Graph isomorphism|isomorphic]] to <math>H</math><ref>{{Cite journal |last1=Courcelle |first1=Bruno |last2=Oum |first2=Sang-il |date=2007-01-01 |title=Vertex-minors, monadic second-order logic, and a conjecture by Seese |url=https://www.sciencedirect.com/science/article/pii/S0095895606000463 |journal=Journal of Combinatorial Theory, Series B |volume=97 |issue=1 |pages=91–126 |doi=10.1016/j.jctb.2006.04.003 |issn=0095-8956|url-access=subscription }}</ref>. Every prime graph with at least 6 vertices has a prime vertex-minor with one less vertex <ref name=":2" />
If a graph admits a split, it can be built by the '''1-join''' of two graphs. The 1-join of two graphs <math>G_1</math> and <math>G_2</math> with marker vertices <math>v_1 \in V(G_1)</math> and <math>v_2 \in V(G_2)</math> is defined to be the graph obtained from the disjoint union of <math>G_1-v_1</math> and <math>G_2-v_2</math> by adding an edge between every neighbour of <math>v_1</math> in <math>G_1</math> and every neighbour of <math>v_2</math> in <math>G_2</math>.


[[Distance-hereditary graph|Distance-hereditary graphs]] are exactly the graphs with no vertex-minor isomorphic to <math>C_5</math><ref>{{Citation |last1=Kwon |first1=O.-joung |title=Unavoidable vertex-minors in large prime graphs |date=2014-03-24 |last2=Oum |first2=Sang-il |journal=European Journal of Combinatorics |volume=41 |pages=100–127 |doi=10.1016/j.ejc.2014.03.013 |arxiv=1306.3066 }}</ref>, and exactly the graphs which are the vertex-minor of a tree<ref>{{Cite journal |last=Bouchet |first=André |date=1987-09-01 |title=Reducing prime graphs and recognizing circle graphs |url=https://doi.org/10.1007/BF02579301 |journal=Combinatorica |language=en |volume=7 |issue=3 |pages=243–254 |doi=10.1007/BF02579301 |issn=1439-6912|url-access=subscription }}</ref>
The '''canonical split decomposition''' can be constructed using the following procedure: Start from a set <math>\{G\}</math> consisting of a single graph. Repeatedly pick a graph <math>H</math> from this set that admits a split. Then replace <math>H</math> with <math>2</math> smaller graphs whose 1-join reconstructs <math>H</math>. This process is applied only when <math>H</math> is neither a complete graph nor a star. The set now contains all induced prime subgraphs of <math>G</math> along with some star graphs and cliques. Lastly, associate a tree by having one node for each graph in the resulting set and adding edges between corresponding marker vertices.<ref name=":5" />


== Pivot operation ==
The canonical split decomposition is unique and has preserves important structural properties over local complementation.
For two adjacent vertices x and y, the '''pivot operation''' is defined as


<math>G\wedge xy = G\star x\star y\star x</math>
# The rank-width of <math>G</math> is the maximum rank-width of all prime induced subgraphs.


It can be shown that <math>G\star x\star y\star x = G\star y\star x\star y</math>, and hence the pivot operation is well defined. The graph <math>G\wedge xy</math> can be obtained from <math>G</math> by toggling adjacencies between every pair of vertices in two different sets among <math>N_G(x) - (N_G(y) \cup {y})</math>, <math>N_G(x) \cap N_G(y)</math>, and <math>N_G(y) - (N_G(x) \cup {x})</math> and then switching labels of <math>x</math> and <math>y</math>.
== Vertex-minor relation ==
Local complementation gives rise to several derived operations that play a central role in graph minor theory. A graph <math>H</math> is a '''vertex-minor''' (also historically known as l-reduction or i-minor) of a graph <math>G</math> if <math>H</math> is an [[Induced subgraph|induced subgraph]] of a graph locally equivalent to <math>G</math>.<ref name=":2">{{Cite journal |last=Bouchet |first=André |date=1987-08-01 |title=Unimodularity and circle graphs |url=https://dx.doi.org/10.1016/0012-365X%2887%2990132-4 |journal=Discrete Mathematics |volume=66 |issue=1 |pages=203–208 |doi=10.1016/0012-365X(87)90132-4 |issn=0012-365X|url-access=subscription }}</ref>


Two graphs are said to be '''pivot equivalent''' if one can be obtained from the other through a sequence of pivot operations. Pivot equivalent graphs are locally equivalent, but the converse is not true. However, Fon-Der-Flass proved that if graphs <math>G</math> and <math>H</math> are locally equivalent, they are pivot equivalent to some <math>G'</math> and <math>H'</math> respectively such that <math>G' * v_1 v_2 \dots v_k = H'</math> and <math>\{v_1, v_2, \dots, v_k\}</math> is an independent set in <math>G'</math>.
Deciding whether <math>H</math> is a vertex-minor of <math>G</math> for two input graphs <math>G</math> and <math>H</math> is [[NP-completeness|NP-complete]], even if <math>H</math> is a [[Complete graph|complete graph]] and <math>G</math> is a [[Circle graph|circle graph]].<ref>{{Cite arXiv|last1=Dahlberg |first1=Axel |title=The complexity of the vertex-minor problem |date=2019-06-12 |eprint=1906.05689 |last2=Helsen |first2=Jonas |last3=Wehner |first3=Stephanie |class=math.CO }}</ref>. However, for each fixed circle graph <math>H</math>, there is an <math>\mathcal O(n^3)</math>-time algorithm to decide whether an input <math>n</math>-vertex graph <math>G</math> contains a vertex-minor [[Graph isomorphism|isomorphic]] to <math>H</math><ref>{{Cite journal |last1=Courcelle |first1=Bruno |last2=Oum |first2=Sang-il |date=2007-01-01 |title=Vertex-minors, monadic second-order logic, and a conjecture by Seese |url=https://www.sciencedirect.com/science/article/pii/S0095895606000463 |journal=Journal of Combinatorial Theory, Series B |volume=97 |issue=1 |pages=91–126 |doi=10.1016/j.jctb.2006.04.003 |issn=0095-8956|url-access=subscription }}</ref>. Every prime graph with at least 6 vertices has a prime vertex-minor with one less vertex.<ref name=":2" />


Pivot equivalence has been studied using even binary delta-matroids.
[[Distance-hereditary graph|Distance-hereditary graphs]] are exactly the graphs with no vertex-minor isomorphic to <math>C_5</math><ref>{{cite journal |last1=Kwon |first1=O.-Joung |title=Unavoidable vertex-minors in large prime graphs |date=2014-03-24 |last2=Oum |first2=Sang-il |journal=European Journal of Combinatorics |volume=41 |pages=100–127 |doi=10.1016/j.ejc.2014.03.013 |arxiv=1306.3066 }}</ref>, and exactly the graphs which are the vertex-minor of a tree<ref>{{Cite journal |last=Bouchet |first=André |date=1987-09-01 |title=Reducing prime graphs and recognizing circle graphs |url=https://doi.org/10.1007/BF02579301 |journal=Combinatorica |language=en |volume=7 |issue=3 |pages=243–254 |doi=10.1007/BF02579301 |issn=1439-6912|url-access=subscription }}</ref>.


== Pivot-minor relation ==
== Pivot-minor relation ==
A graph H is a pivot-minor of a graph G if H is an induced subgraph of a graph pivot-equivalent to G. Note that bipartite graphs with the pivot-minor relation are essentially equivalent to [[Binary matroid|binary matroids]] with the [[Matroid minor|matroid minor relation]].
A graph <math>H</math> is a pivot-minor of a graph <math>G</math> if <math>H</math> is an induced subgraph of a graph pivot-equivalent to <math>G</math>. Bipartite graphs with the pivot-minor relation are essentially equivalent to [[Binary matroid|binary matroids]] with the [[Matroid minor|matroid minor relation]].
 
== Relation to circle graphs ==
For a [[Circle graph|circle graph]] [[Circle graph|<math>G</math>]], performing a local complementation at <math>v</math> corresponds to an graphical transformation of its chord diagram. In particular, the chord diagram of [[Circle graph|<math>G\star v</math>]] is obtains from the chord diagram of [[Circle graph|<math>G</math>]] by cutting the circumference by chord representing <math>v</math>, then reversing one arc<ref name=":0" /><ref name=":4" />. The class of circle graphs are hence closed under local complementation, and they are also closed under taking vertex-minors.


== References ==
== References ==

Latest revision as of 02:55, 14 April 2026

Short description: Operation in graph theory

In graph theory, local complementation (also known as vertex inversion) is an operation on a graph that toggles adjacencies among the neighbours of a chosen vertex, while all other adjacencies remain unchanged. Despite its simple definition, it preserves interesting properties and generates a complex equivalence relation. The operation was introduced by Anton Kotzig and later studied in depth by André Bouchet and Von-Der-Flaass.

Formally, the local complementation of a simple undirected graph G at a vertex v is an operation that produces a new graph, denoted by Gv. This operation is defined by replacing the subgraph of G induced by NG(v) with its complementary subgraph. In other words, two distinct vertices x and y are adjacent in the graph Gv when exactly one of the following holds:

  1. vertices x and y are adjacent in G; or
  2. both vertices x and y are neighbours of v in G.

Two graphs are said to be locally equivalent if one can be obtained from the other through a sequence of local complementations. This defines an equivalence relation on graphs, whose equivalence classes are known as local equivalence classes. For example, the star graph and complete graph on n vertices are locally equivalent, and they form a local equivalence class. The local equivalence classes on graphs with up to 12 vertices has been computed.[1] The size of a local equivalence class is at most 3n, and this collection of graphs can be enumerated efficiently.

Applications

Structural graph theory

The Robertson–Seymour theorem states that the graph minor relation is a well-quasi ordering. It was proved in a series of twenty papers spanning over 500 pages from 1983 to 2004. The algorithmic consequences are vast - together with an efficient algorithm for graph minor testing, the result provides efficient algorithms for solving a range of computational problems where the optimal value is monotonic in the graph minor relation.

Local complementations are central to the vertex-minor relation, which shares many similarities with the graph minor relation. Better understanding of the local complementation operation could extend the Robertson–Seymour theorem to prove that the vertex-minor relation is also a well-quasi ordering.

Measurement based quantum computing

For a given the graph state |G, the action of the local Clifford operation is equivalent[2] to the local complementation transformation on the graph G. The study of graph states that are locally equivalent is relevant to building quantum circuits in measurement based quantum computing (MBQC)[1]

Similarly, local complementation is also related to state preparation in photonic quantum computing (PQC).

Properties

  1. The local complement operation is self inverse; that is, (Gv)v=G.
  2. Local complementations commute only when vertices are non-adjacent; that is, (Gv)w=(Gw)v unless vwE(G).
  3. Connected components are preserved by the local complementation operation, so it is common analyse each component of the graph independently. So without loss of generality, all graphs will be assumed to be connected.
  4. All locally equivalent graphs can be reached after a sequence of at most max(n+1,10n/9) local complementations[3].
  5. Locally equivalent trees are isomorphic (Mulder's conjecture), and there exists a simple expression to count such trees.[4]
  6. If a graph is locally equivalent to a tree, it has a subgraph isomorphic to that tree.[5]
  7. If a graph is locally equivalent to the cycle graph, it contains a Hamiltonian cycle.[5]
  8. If a graph is locally equivalent to its complement, it is self-complementary.[5]
  9. The number of essentially different ways of transforming one graph into another via local complementation is 0,3,6 or 2k for some k0.[5]
  10. Any class of locally equivalent distance-hereditary graphs is equal to the class of the circle graphs of the Euler tours of some 4-regular graph.
  11. Locally equivalent graphs have the same rank-width.
  12. Counting locally equivalent graphs is #P-complete.[6]
  13. Determining the minimum degree that can be reached by means of local complementation is NP-complete and APX-hard, and can be computed by an 𝒪*(1.938n) algorithm.[7][8]
  14. Determining the minimum total edge weight by means of local complementation is NP-complete.[9]
  15. Determining the minimum number of edges by means of local complementation is likely NP-complete.

Pivot operation

Local complementations do not commute for adjacent vertices, motivating the following operation that performs complementations at adjacent vertices.

For adjacent vertices x and y, the pivot operation (also historically known as an edge complementation) is defined asGxy=Gxyx.It can be shown that Gxyx=Gyxy, and hence the pivot operation is well defined. Alternatively, the graph Gxy can be obtained from G by toggling adjacencies between every pair of vertices in two different sets among NG(x)(NG(y)y), NG(x)NG(y) and NG(y)(NG(x)x) then switching the labels x and y. The pivot operation satisfies the identities Gxyxy=G and Gxyyz=Gxz[10].

Two graphs are said to be pivot equivalent if one can be obtained from the other through a sequence of pivot operations. Since pivot operations consist of local complementation operations, pivot equivalent graphs are locally equivalent. The converse is true for bipartite graphs[3], but is not generally true. If G and H are pivot equivalent graphs, there are pairwise disjoint edges e1,e2,,ek such that Ge1e2ek=H. Fon-Der-Flass proved that if graphs G and H are locally equivalent, they are pivot equivalent to some G and H respectively such that G*v1v2vk=H and {v1,v2,,vk} is an independent set in G and H[11].

The size of a pivot equivalence class is at most 2n, and this collection of graphs can be enumerated efficiently.

Pivot equivalence has been studied using even binary delta-matroids[12].

Invariants

Cut-rank function

As a graph undergoes local complementation, its adjacency matrix changes in a well defined way. In particular, the entries that change are exactly some sub-matrix (excluding the main diagonal). Hence, it may be natural to study the local complementation operation using linear algebra.

For a graph G with vertex set V, the cut-rank function (also historically known as the connectivity function) is denoted ρG:𝒫(V). It is defined over the vertex subsets XV such that ρG(X) is the rank of the bi-adjacency matrix of the partition (X,VX), defined over the finite field GF(2). That is, the rank of the X×(VX) binary matrix M where Muv=1 when uv is an edge in G. Intuitively, ρG(X) is a measure of how complex the connectivity is between X and the remaining vertices.

Since the rank of a matrix is preserved by elementary row and column operations, a straightforward argument shows that locally equivalent graphs have identical cut-rank functions. However, the converse is not true - a counterexample can be constructed using labelled Petersen graphs. It is known that bipartite graphs with identical cut-rank functions are pivot-equivalent[13], and so locally equivalent bipartite graphs are also pivot-equivalent.

The cut-rank function is submodular since it can be shown that ρG(X)+ρG(Y)ρG(XY)+ρG(XY) for any vertex set X and Y. However, it is not monotone.

Local sets

The cut-rank can be considered 'most interesting' when it is either rank 1 or full rank. Since the cut-rank is invariant over local complementation, so are the sets with a particular cut-rank. The sets of rank 1 are studied using canonical split decompositions in a later section. Here, we define a local set as a vertex set which does not have full cut-rank.

Define OddG(X)={vV(G)|N(v)X|=1mod2}.A set of vertices S is local in G if S=XOddG(X) for some subset XV(G). Now if S is local in G, it is also local in any graph locally equivalent to G. Local sets can equivalently be defined as the sets of vertices which do not have full cut-rank (i.e. the sets S where ρG(S)<|S|). Intuitively, a local set is some linear combination of the closed neighbourhoods of G. Local sets are used to study the degrees of vertices in G.[14]

This invariant may be a helpful tool to quickly show that graphs are not locally equivalent. However, there are graphs with the same local sets which are not locally equivalent, since there are graphs with identical cut-rank functions which are not locally equivalent.[15]

Totally isotropic subspaces

The richest description of a class of locally equivalent graphs is an extension of the local sets idea, involving a linear algebra structure over the Klein four-group. Each locally equivalent graph, equipped with two specific vectors, corresponds to some graphic presentations of the same totally isotropic subspace.

Let K4 be the Klein four-group. A vector 𝐚K4n is said to be complete if 𝐚i is nonzero for every 1in. Two vectors 𝐚,𝐛K4n are said to be supplementary if 𝐚i and 𝐛i are nonzero and distinct for every 1in. For a set S, define 𝐚[S] as the vector where 𝐚[S]i=ai if iS, and 𝐚[S]i=0 otherwise.

A subspace L of K4n is called a totally isotropic subspace if dim(L)=dim(K4n)/2=n, and every two complete vectors in L are not supplementary.

A graphic presentation of a totally isotropic subspace L is a triple (G,𝐚,𝐛) where G is a simple graph on the vertex set V={1,,n} and 𝐚 and 𝐛 are supplementary vectors of K4n, such that L is spanned by the set {𝐚[N(v)]+𝐛[{v}]vV}.For a fixed L, there is a one-to-one correspondence between every graphic presentation (G,𝐚,𝐛) and every complete vector 𝐚 where 𝐚∉L, the vectors 𝐚 here are known are Eulerian vectors. Furthermore, if (G,𝐚,𝐛) is a graphic presentation of L, so is (Gv,𝐚+𝐛[{v}],𝐛+𝐚[N(v)]) and (Gvw,𝐚[V{v,w}]+𝐛[{v,w}],𝐚[{v,w}]+𝐛[V{v,w}]).The fundamental graphs of L form a local equivalence class. These facts leads to important results about determining local equivalence and locally equivalent class sizes, which is discussed in the next 2 sections.

This structure can be considered as an extension of the local sets structure, since L={𝐚[OddG(X)]+𝐛[X]XV}, and when 𝐚=𝐛, L={𝐚[S]S is local in G}.[16]

Characterisation of local equivalence

The following characterisations of locally equivalent graphs are straightforward results using totally isotropic subspaces. This leads to a efficient algorithm to recognise locally equivalent graphs.

Let G1 and G2 be simple graphs and fix supplementary vectors 𝐚1 and 𝐛1. he graphs G1 and G2 are locally equivalent if and only if there are supplementary vectors 𝐚2 and 𝐛2 such that (G1,𝐚1,𝐛1) and (G2,𝐚2,𝐛2) are graphic presentations of the same totally isotropic subspace. The existence of such vectors 𝐚2 and 𝐛2 can be determined in 𝒪(n3) by solving a system of linear equations[17]. A modification to this algorithm can, with the same time complexity, recover a sequence of at most 3n/2 local complementations transforming G1 into G2[18].

An equivalent characterisation can be formulated using vertex sets. Let N1 and N2 denote the neighbourhood functions of G1 and G2 respectively. Then graphs G1 and G2 are locally equivalent if and only if there are vertex subsets X,Y,Z,T such that for every pair of vertices v,w, |XN1(v)N2(w)|+|YN1(v){w}|+|ZN2(w){v}|+|T{v,w}|0(mod2),and the symmetric difference of XT with YZ is the entire vertex set.[17]

Global interlace polynomial

The global interlace polynomial (also known as the Tutte-Martin polynomial), is a polynomial that corresponds to a simple graph G. It is defined recursively asQ(G;x)={Q(Gu;x)+Q((G*u)u;x)+Q((Guv)u;x)uvE(G)xnG has no edges and n verticesNow if G and H are locally equivalent, Q(G,x)=Q(H,x) for any x[15]. Additionally,

  • Q(G,2) is a multiple of the number of graphs locally equivalent to G. In particular, if G is a fundamental graph of a totally isotropic subspace L, this gives the number of Eulerian vectors in L.
  • Q(G;4)/2n is the number of induced Eulerian subgraphs in G. (An Eulerian graph contains only vertices of even degree).

The polynomial has closed-form formulas in certain cases:

  • The path graphs Pn have the closed form Q(Pn;x)=λ1(1+1+x)n1+λ2(11+x)n1 where λ1=x/2+x/1+x and λ2=x/2x/1+x.
  • More generally, a tree T has the global interlace polynomial Q(T;x)=k0Φk(T)2n2kxk, where Φk(T) is the number of matchings of size k in T.
  • The cycle graphs Cn have the closed form Q(Cn;2)=(1+3)n+(13)n4(2n1+(1)n)/3[15].

An equivalent definition of the global interlace polynomial involves a summation over subsets of vertex sets. Let the graph G have the adjacency matrix A over the binary field. For a vertex set SV(G), write A[S] to denote the S×S sub-matrix of A. Let IX be the V(G)×V(G) diagonal matrix over the binary field such that the (v,v)-entry is 1 when vX, and 0 otherwise. The global interlace polynomial can equivalently be defined asQ(G;x)=RSV(G)(x2)nullity((A+IR)[S]).There are some interesting similarities to the canonical Tutte polynomial. In particular, the recurrence looks similar to the deletion-contraction formula, and both polynomials can be formulated as a sum of terms raised to the power of a matrix rank. Isomorphic graphs have the same Tutte polynomial, while locally equivalent or isomorphic graphs have the same global interlace polynomial.

Canonical split decomposition

A split of a graph G is a partition of its vertex set (X,Y) such that |X|,|Y|2 and ρG(X)=1. This happens exactly when the connectivity between X and Y is a complete bipartite graph, and complete bipartite graphs have a simple known structure over local complementation.

If a graph admits a split, it can be built by the 1-join of two graphs. The 1-join of two graphs G1 and G2 with marker vertices v1V(G1) and v2V(G2) is defined to be the graph obtained from the disjoint union of G1v1 and G2v2 by adding an edge between every neighbour of v1 in G1 and every neighbour of v2 in G2.

The canonical split decomposition can be constructed using the following procedure: Start from a set {G} consisting of a single graph. Repeatedly pick a graph H from this set that admits a split. Then replace H with 2 smaller graphs whose 1-join reconstructs H. This process is applied only when H is neither a complete graph nor a star. The set now contains all induced prime subgraphs of G along with some star graphs and cliques. Lastly, associate a tree by having one node for each graph in the resulting set and adding edges between corresponding marker vertices.[4]

The canonical split decomposition is unique and has preserves important structural properties over local complementation.

  1. The rank-width of G is the maximum rank-width of all prime induced subgraphs.

Vertex-minor relation

Local complementation gives rise to several derived operations that play a central role in graph minor theory. A graph H is a vertex-minor (also historically known as l-reduction or i-minor) of a graph G if H is an induced subgraph of a graph locally equivalent to G.[19]

Deciding whether H is a vertex-minor of G for two input graphs G and H is NP-complete, even if H is a complete graph and G is a circle graph.[20]. However, for each fixed circle graph H, there is an 𝒪(n3)-time algorithm to decide whether an input n-vertex graph G contains a vertex-minor isomorphic to H[21]. Every prime graph with at least 6 vertices has a prime vertex-minor with one less vertex.[19]

Distance-hereditary graphs are exactly the graphs with no vertex-minor isomorphic to C5[22], and exactly the graphs which are the vertex-minor of a tree[23].

Pivot-minor relation

A graph H is a pivot-minor of a graph G if H is an induced subgraph of a graph pivot-equivalent to G. Bipartite graphs with the pivot-minor relation are essentially equivalent to binary matroids with the matroid minor relation.

Relation to circle graphs

For a circle graph G, performing a local complementation at v corresponds to an graphical transformation of its chord diagram. In particular, the chord diagram of Gv is obtains from the chord diagram of G by cutting the circumference by chord representing v, then reversing one arc[17][15]. The class of circle graphs are hence closed under local complementation, and they are also closed under taking vertex-minors.

References

  1. 1.0 1.1 Cabello, Adan; Danielsen, Lars Eirik; Lopez-Tarrida, Antonio J.; Portillo, Jose R. (2011-04-16). "Optimal preparation of graph states". Physical Review A 83 (4). doi:10.1103/PhysRevA.83.042314. Bibcode2011PhRvA..83d2314C. 
  2. Ji, Z.-F.; Chen, J.-X.; Wei, Z.-H.; Ying, M.-S. (January 2010). "The LU-LC conjecture is false". Quantum Information and Computation 10 (1&2): 97–108. doi:10.26421/QIC10.1-2-8. http://www.rintonpress.com/journals/doi/QIC10.1-2-8.html. 
  3. 3.0 3.1 Koršunov, Aleksej D. (1996). Discrete Analysis and Operations Research. Mathematics and Its Applications. Dordrecht: Springer Netherlands. ISBN 978-94-010-7217-5. 
  4. 4.0 4.1 Bouchet, André (1988). "Transforming trees by successive local complementations" (in en). Journal of Graph Theory 12 (2): 195–207. doi:10.1002/jgt.3190120210. ISSN 1097-0118. https://onlinelibrary.wiley.com/doi/abs/10.1002/jgt.3190120210. 
  5. 5.0 5.1 5.2 5.3 Fon-Der-Flaas, D. G. (1988). "On local complementations of graphs". https://zbmath.org/0728.05048. 
  6. Dahlberg, Axel; Helsen, Jonas; Wehner, Stephanie (2019-07-18). "Counting single-qubit Clifford equivalent graph states is #P-complete". Journal of Mathematical Physics 61 (2). doi:10.1063/1.5120591. 
  7. Javelle, Jérôme; Mhalla, Mehdi; Perdrix, Simon (2012-04-20). "On the Minimum Degree up to Local Complementation: Bounds and Complexity". Graph-Theoretic Concepts in Computer Science. Lecture Notes in Computer Science. 7551. pp. 138–147. doi:10.1007/978-3-642-34611-8_16. ISBN 978-3-642-34610-1. 
  8. Cattanéo, David; Perdrix, Simon (2016-08-17). "Minimum Degree up to Local Complementation: Bounds, Parameterized Complexity, and Exact Algorithms". Algorithms and Computation. Lecture Notes in Computer Science. 9472. pp. 259–270. doi:10.1007/978-3-662-48971-0_23. ISBN 978-3-662-48970-3. 
  9. Sharma, Hemant; Goodenough, Kenneth; Borregaard, Johannes; Rozpędek, Filip; Helsen, Jonas (2026-01-27). "Minimising the number of edges in LC-equivalent graph states". Quantum 10. doi:10.22331/q-2026-02-09-2001. 
  10. Oum, Sang-il (2005-09-01). "Rank-width and vertex-minors". Journal of Combinatorial Theory, Series B 95 (1): 79–100. doi:10.1016/j.jctb.2005.03.003. ISSN 0095-8956. https://www.sciencedirect.com/science/article/pii/S0095895605000389. 
  11. D. G., Fon-Der-Flaass (1989). "Distance between locally equivalent graphs". Metody Diskret. Analiz 48: 85–94. 
  12. Bouchet, A.; Duchamp, A. (1991-02-15). "Representability of △-matroids over GF(2)". Linear Algebra and Its Applications 146: 67–78. doi:10.1016/0024-3795(91)90020-W. ISSN 0024-3795. https://dx.doi.org/10.1016/0024-3795%2891%2990020-W. 
  13. Seymour, P. D (1988-08-01). "On the connectivity function of a matroid". Journal of Combinatorial Theory, Series B 45 (1): 25–30. doi:10.1016/0095-8956(88)90052-4. ISSN 0095-8956. https://dx.doi.org/10.1016/0095-8956%2888%2990052-4. 
  14. Høyer, Peter; Mhalla, Mehdi; Perdrix, Simon (2006). Asano, Tetsuo. ed. "Resources Required for Preparing Graph States" (in en). Algorithms and Computation (Berlin, Heidelberg: Springer): 638–649. doi:10.1007/11940128_64. ISBN 978-3-540-49696-0. https://link.springer.com/chapter/10.1007/11940128_64?error=cookies_not_supported&code=bcd69664-6723-4241-80fa-838f1fd751cd. 
  15. 15.0 15.1 15.2 15.3 Bouchet, André (1993-04-28). "Recognizing locally equivalent graphs". Discrete Mathematics 114 (1): 75–86. doi:10.1016/0012-365X(93)90357-Y. ISSN 0012-365X. https://dx.doi.org/10.1016/0012-365X%2893%2990357-Y. 
  16. Bouchet, André (1988-08-01). "Graphic presentations of isotropic systems". Journal of Combinatorial Theory, Series B 45 (1): 58–76. doi:10.1016/0095-8956(88)90055-X. ISSN 0095-8956. https://www.sciencedirect.com/science/article/pii/009589568890055X. 
  17. 17.0 17.1 17.2 Bouchet, André (1991-12-01). "An efficient algorithm to recognize locally equivalent graphs" (in en). Combinatorica 11 (4): 315–329. doi:10.1007/BF01275668. ISSN 1439-6912. https://doi.org/10.1007/BF01275668. 
  18. Fon-Der-Flaass, D. G. (1996), Korshunov, Alekseǐ D., ed., "Local Complementations of Simple and Directed Graphs" (in en), Discrete Analysis and Operations Research (Dordrecht: Springer Netherlands): pp. 15–34, doi:10.1007/978-94-009-1606-7_3, ISBN 978-94-009-1606-7, https://doi.org/10.1007/978-94-009-1606-7_3, retrieved 2026-02-26 
  19. 19.0 19.1 Bouchet, André (1987-08-01). "Unimodularity and circle graphs". Discrete Mathematics 66 (1): 203–208. doi:10.1016/0012-365X(87)90132-4. ISSN 0012-365X. https://dx.doi.org/10.1016/0012-365X%2887%2990132-4. 
  20. Dahlberg, Axel; Helsen, Jonas; Wehner, Stephanie (2019-06-12). "The complexity of the vertex-minor problem". arXiv:1906.05689 [math.CO].
  21. Courcelle, Bruno; Oum, Sang-il (2007-01-01). "Vertex-minors, monadic second-order logic, and a conjecture by Seese". Journal of Combinatorial Theory, Series B 97 (1): 91–126. doi:10.1016/j.jctb.2006.04.003. ISSN 0095-8956. https://www.sciencedirect.com/science/article/pii/S0095895606000463. 
  22. Kwon, O.-Joung; Oum, Sang-il (2014-03-24). "Unavoidable vertex-minors in large prime graphs". European Journal of Combinatorics 41: 100–127. doi:10.1016/j.ejc.2014.03.013. 
  23. Bouchet, André (1987-09-01). "Reducing prime graphs and recognizing circle graphs" (in en). Combinatorica 7 (3): 243–254. doi:10.1007/BF02579301. ISSN 1439-6912. https://doi.org/10.1007/BF02579301.