arXiv++ Combinatorics

A metric space (X,d) is declared to be natural if (X,d) determines an up to isomorphism unique group structure (X,+) on the set X such that all the group translations and group inversion are isometries. A group is called natural if it emerges like this from a natural metric. A simple graph X is declared to be natural if (X,d) with geodesic metric d is natural. We look here at some examples and some general statements like that the graphical regular representations of a finite group is always a natural graphs or that the direct product on groups or the Shannon product of finite graphs preserves the property of being natural. The semi-direct product of finite natural groups is natural too as they are represented by Zig-Zag products of suitable Cayley graphs. It follows that wreath products preserve natural groups. The Rubik cube for example is natural. Also free products of finitely generated natural groups are natural. A major theme is that non-natural groups often can be upgraded to become natural by extending them to become Coxeter groups. Examples of non-natural groups are cyclic groups whose order is divisible by 4, the quaternion group, the integers, the lamplighter group, the free groups or the group of p-adic integers. The prototype feature is to extend the integers and get the infinite dihedral group, replacing the single generator by two free reflections. We conclude with a short discussion of the hypothesis of using the dihedral group as a physical time in dynamical system theory.

The Babylonian graph B has the positive integers as vertices and connects two if they define a Pythagorean triple. Triangular subgraphs correspond to Euler bricks. What are the properties of this graph? Are there tetrahedral subgraphs corresponding to Euler tesseracts? Is there only one infinite connected component? Are there two Euler bricks in the graph that are disconnected? Do the number of edges or triangles in the subgraph generated by the first n vertices grow like of the order n W(n), where n is the product log? We prove here some first results like the threshold where B(n) becomes non-planar. In an appendix, we include handout from a talk on Euler cuboids given in the year 2009.

We prove that each eigenvalue l(k) of the Kirchhoff Laplacian K of a graph or quiver is bounded above by d(k)+d(k-1) for all k in {1,...,n}. Here l(1),...,l(n) is a non-decreasing list of the eigenvalues of K and d(1),..,d(n) is a non-decreasing list of vertex degrees with the additional assumption d(0)=0. We also prove that in general the weak Brouwer-Haemers lower bound d(k) + (n-k) holds for all eigenvalues l(k) of the Kirchhoff matrix of a quiver.

The number of rooted spanning forests divided by the number of spanning rooted trees in a graph G with Kirchhoff matrix K is the spectral quantity tau(G)= det(1+K)/det(K) of G by the matrix tree and matrix forest theorems. We prove that that under Barycentric refinements, the tree index T(G)=log(det(K))/n and forest index F(G)=log(det(1+K))/n and so the tree-forest index i=F-G=log(tau(G))/n converge to numbers that only depend on the size of the maximal clique in the graph. In the 1-dimensional case, all numbers are known: T(G)=0, F(G)=i(G) =2 log(phi), where phi is the golden ratio. The convergent proof uses the Barycentral limit theorem assuring the Kirchhoff spectrum converges weakly to a measure dk on the positive real axis that only depends on dimension of G. Trees and forests indices are potential values i = U(-1)-U(0) for the subharmonic function U(z)=int_R log|x-z| dk(x) defined by the Riesz measure dk=Delta U which only depends on the dimension of G. The potential U(z) is defined for all z away from the support of dk and finite at z=0. Convergence follows from the tail estimate k[x,infty] < C exp(-a x) where the decay rate a only depends on the maximal dimension. With the normalized zeta function zeta(s) = (1/n) sum_k lambda_k^-s, we have for all finite graphs of maximal dimension larger than 1 the identity i(G) = sum_t (-1)^(s+1) zeta(s)/s. The limiting zeta function zeta(s) = int_R x^(-s) dk(x) is analytic in s for s<0. The Hurwitz spectral zeta function zeta_z(s)=U_s(z) = int_R (x-z)^(-s) dk(x) complements U(z) = int_R log(x-z) dk(x) and is analytic for z in C - R^+ and for fixed z in C-R^+ is an entire function in s in C.

Analytic torsion is a functional on graphs which only needs linear algebra to be defined. In the continuum it corresponds to the Ray-Singer analytic torsion. We have formulas for analytic torsion if the graph is contractible or if it is a discrete sphere. A key insight is that analytic torsion is the super determinant of the Dirac operator of the graph.

We show that the curvature K_(G*H)(x,y) at a point (x,y) in the strong product G*H of two arbitrary finite simple graphs is equal to the product K_G(x) K_H(y) of the curvatures.

Discrete d-manifolds are classes of finite simple graphs which can triangulate classical manifolds but which are defined entirely within graph theory. We show that the chromatic number X(G) of a discrete d-manifold G is sandwiched between d+1 and 2(d+1). From the general identity X(A+B)=X(A)+X(B) for the join A+B of two finite simple graphs, it follows that there are (2k)-spheres with chromatic number X=(3k+1) and (2k-1)-spheres with chromatic number X=3k. Examples of 2-manifolds with X(G)=5 have been known since the pioneering work of Fisk. Current data support the that the ceiling function of 3(d+1)/2 could be an upper bound for all d-manifolds G, generalizing a conjecture of Albertson-Stromquist, stating X(G) is bounded above by 5 for all 2-manifolds. For a d-manifold, Fisk has introduced the (d-2)-variety O(G). This graph O(G) has maximal simplices of dimension (d-2) and correspond to complete complete subgraphs K_{d-1} of G for which the dual circle has odd cardinality. In general, O(G) is a union of (d-2)-manifolds. We note that if O(S(x)) is either empty or a (d-3)-sphere for all x then O(G) is a (d-2)-manifold or empty. The knot O(G) is already interesting for 3-manifolds G because Fisk has demonstrated that every possible knot can appear as O(G) for some 3-manifold. For 4-manifolds G especially, the Fisk variety O(G) is a 2-manifold in G as long as all O(S(x)) are either empty or a knot in every unit 3-sphere S(x).

Graph complements G(n) of cyclic graphs are circulant, vertex-transitive, claw-free, strongly regular, Hamiltonian graphs with a Z(n) symmetry, Shannon capacity 2 and known Wiener and Harary index. There is an explicit spectral zeta function and tree or forest data. The forest-tree ratio converges to e. The graphs G(n) are Cayley graphs and so Platonic with isomorphic unit spheres G(n-3)^+, complements of path graphs. G(3d+3) are homotop to wedge sums of two d-spheres and G(3d+2),G(3d+4) are homotop to d-spheres, G(3d+1)^+ are contractible, G(3d+2)^+,G(3d+3)^+ are d-spheres. Since disjoint unions are dual to Zykov joins, graph complements of 1-dimensional discrete manifolds G are homotop to a point, a sphere or a wedge sums of spheres. If the length of every connected component of a 1-manifold is not divisible by 3, the graph complement of G is a sphere. In general, the graph complement of a forest is either contractible or a sphere. All induced strict subgraphs of G(n) are either contractible or homotop to spheres. The f-vectors G(n) or G(n)^+ satisfy a hyper Pascal triangle relation, the total number of simplices are hyper Fibonacci numbers. The simplex generating functions are Jacobsthal polynomials, generating functions of k-king configurations on a circular chess board. While the Euler curvature of circle complements G(n) is constant by symmetry, the discrete Gauss-Bonnet curvature of path complements G(n)^+ can be expressed explicitly from the generating functions. There is now a non-trivial 6 periodic Gauss-Bonnet curvature universality in the complement of Barycentric limits. The Brouwer-Lefschetz fixed point theorem produces a 12-periodicity of the Lefschetz numbers of all graph automorphisms of G(n). There is also a 12-periodicity of Wu characteristic. This is a 4 periodicity in dimension.These are manifestations of stable homotopy features, but combinatorial.

A finite abstract simplicial complex G defines the Barycentric refinement graph phi(G) = (G,{ (a,b), a subset b or b subset a }) and the connection graph psi(G) = (G,{ (a,b), a intersected with b not empty }). We note here that both functors phi and psi from complexes to graphs are invertible on the image (Theorem 1) and that G,phi(G),psi(G) all have the same automorphism group and that the Cartesian product of G corresponding to the Stanley-Reisner product of phi(G) and the strong Shannon product of psi(G), have the product automorphism groups. Second, we see that if G is a Barycentric refinement, then phi(G) and psi(G) are graph homotopic (Theorem 2). Third, if gamma is the geometric realization functor, assigning to a complex or to a graph the geometric realization of its clique complex, then gamma(G) and gamma(phi(G)) and gamma(psi(G)) are all classically homotopic for a Barycentric refined simplicial complex G (Theorem 3). The Barycentric assumption is necessary in Theorem 2 and 3. There is compatibility with Cartesian products of complexes which manifests in the strong graph product of connection graphs: if two graphs A,A' are homotopic and B,B' are homotopic, then A . B is homotopic to A' . B' (Theorem 4) leading to a commutative ring of homotopy classes of graphs. Finally, we note (Theorem 5) that for all simplicial complexes G as well as product G=G_1 x G_2 ... x G_k, the Shannon capacity Theta(psi(G)) of psi(G) is equal to the number m of zero-dimensional sets in G. An explicit Lowasz umbrella in R^m leads to the Lowasz number theta(G) leq m and so Theta(psi(G))=theta(psi(G))=m making Theta compatible with disjoint union addition and strong multiplication.

If h is a ring-valued function on a simplicial complex G we can define two matrices L and g, where the matrix entries are the h energy of homoclinic intersections. We know that the sum over all h values on G is equal to the sum of the Green matrix entries g(x,y). We also have already seen that that the determinants of L or g are both the product of the h(x). In the case where h(x) is the parity of dimension, the sum of the energy values was the standard Euler characteristic and the determinant was a unit. If h(x) was the unit in the ring then L,g are integral quadratic forms which are isospectral and inverse matrices of each other. We prove here that the quadratic energy expression summing over all pairs h(x)^* h(y) of intersecting sets is a signed sum of squares of Green function entries. The quadratic energy expression is Wu characteristic in the case when h is dimension parity. For general h, the quadratic energy expression resembles an Ising Heisenberg type interaction. The conjugate of g is the inverse of L if h takes unit values in a normed ring or in the group of unitary operators in an operator algebra.

We look at connection Laplacians L,g defined by a field h:G to K, where G is a finite set of sets and K is a normed division ring which does not need to be commutative, nor associative but has a conjugation leading to the norm as the square root of h^* h. The target space K can be a normed real division algebra like the quaternions or an algebraic number field like a quadratic field. For parts of the results we can even assume K to be a Banach algebra like an operator algebra on a Hilbert space. The K-valued function h on G then defines connection matrices L,g in which the entries are in K. We show that the Dieudonne determinants of L and g are both equal to the abelianization of the product of all the field values on G. If G is a simplicial complex and h takes values in the units U of K, then g^* is the inverse of L and the sum of the energy values is equal to the sum of the Green function entries g(x,y). If K is the field C of complex numbers, we can study the spectrum of L(G,h) in dependence of the field h. The set of matrices with simple spectrum defines a |G|-dimensional non-compact Kaehler manifold that is disconnected in general and for which we can compute the fundamental group of each connected component.

An integral geometric curvature is defined as the index expectation K(x) = E[i(x)] if a probability measure m is given on vector fields on a Riemannian manifold or on a finite simple graph. Such curvatures are local, satisfy Gauss-Bonnet and are independent of any embedding in an ambient space. While realizing constant Gauss-Bonnet-Chern curvature is not possible in general already for 4-manifolds, we prove that for compact connected manifolds, constant curvature K_m can always be realized with m supported on Morse gradient fields. We give examples of finite simple graphs which do not allow for any constant m-curvature and prove that for one-dimensional connected graphs, there is a convex set of constant curvature configurations with dimension of the first Betti number of the graph. In particular, there is always a unique constant curvature solution for trees.

We illustrate connections between differential geometry on finite simple graphs G=(V,E) and Riemannian manifolds (M,g). The link is that curvature can be defined integral geometrically as an expectation in a probability space of Poincare-Hopf indices of coloring or Morse functions. Regge calculus with an isometric Nash embedding links then the Gauss-Bonnet-Chern integrand of a Riemannian manifold with the graph curvature. There is also a direct nonstandard approach: if V is a finite set containing all standard points of M and E contains pairs which are infinitesimally close in the sense of internal set theory, one gets a finite simple graph (V,E) which gets a curvature which as a measure corresponds to the standard curvature. The probabilistic approach is an umbrella framework which covers discrete spaces, piecewise linear spaces, manifolds or varieties.

We generalize the Poincare-Hopf theorem sum_v i(v) = X(G) to vector fields on a finite simple graph (V,E) with Whitney complex G. To do so, we define a directed simplicial complex as a finite abstract simplicial complex equipped with a bundle map F: G to V telling which vertex T(x) in x dominates the simplex x. The index i(v) of a vertex v is defined as X(F^-1(v)), where X is the Euler characteristic. We get a flow by adding a section map F: V to G. The resulting map T on G is a discrete model for a differential equation x'=F(x) on a compact manifold. Examples of directed complexes are defined by Whitney complexes defined by digraphs with no cyclic triangles or gradient fields on finite simple graphs defined by a locally injective function. The result extends to simplicial complexes equipped with an energy function H:G to Z that implements a divisor. The index sum is then the total energy.

A finite simple graph G is declared to have positive curvature if every in G embedded wheel graph has five or six vertices. A d-graph is a finite simple graph G for which every unit sphere is a (d-1)-sphere. A d-sphere is a d-graph G for which there exists a vertex x such that G-x is contractible. A graph G is contractible if there is a vertex x such that S(x) and G-x are contractible. The empty graph 0 is the (-1)-sphere. The 1-point graph 1 is contractible. The theorem is that for d bigger than 1, every connected positive curvature d-graph is a d-sphere. A discrete Synge result follows: a positive curvature graph is simply connected and orientable. For every d larger than 1, there are only finitely many positive curvature graphs. There are six for d=2 and all have diameter less or equal to 3. To prove the theorem, we use a "geomag lemma" which shows that every geodesic in G can be extended to an immersed 2-graph S of positive curvature and must so be a 2-sphere with positive curvature. As none of these has diameter larger than 3, also G has a diameter 3 or less. This can be used to show that G-x is contractible and so must be a sphere.

For a simplicial complex with n sets, let W^-(x) be the set of sets in G contained in x and W^+(x) the set of sets in G containing x. An integer-valued function h on G defines for every A subset G an energy E[A]=sum_x in A h(x). The function energizes the geometry similarly as divisors do in the continuum, where the Riemann-Roch quantity chi(G)+deg(D) plays the role of the energy. Define the n times n matrices L=L^--(x,y)=E[W^-(x) cap W^-(y)] and L^++(x,y) = E[W^+(x) cap W^+(y)]. With the notation S(x,y)=1_n omega(x) =delta(x,y) (-1)dim(x) and str(A)=tr(SA) define g=S L^++ S. The results are: det(L)=det(g) = prod_x in G h(x) and E[G] = sum_x,y g(x,y) and E[G]=str(g). The number of positive eigenvalues of g is equal to the number of positive energy values of h. In special cases, more is true: A) If h(x) in -1, 1}, the matrices L=L^--,L^++ are unimodular and L^-1 = g, even if G is a set of sets. B) In the constant energy h(x)=1 case, L and g are isospectral, positive definite matrices in SL(n,Z). For any set of sets G we get so isospectral multi-graphs defined by adjacency matrices L^++ or L^-- which have identical spectral or Ihara zeta function. The positive definiteness holds for positive divisors in general. C) In the topological case h(x)=omega(x), the energy E[G]=str(L) = str(g) = sum_x,y g(x,y)=chi(G) is the Euler characteristic of G and phi(G)=prod_x omega(x), a product identity which holds for arbitrary set of sets. D) For h(x)=t^|x| with some parameter t we have E[H]=1-f_H(t) with f_H(t)=1+f_0 t + cdots + f_d t^d+1 for the f-vector of H and L(x,y) = (1-f_W^-(x) cap W^-(y)(t)) and g(x,y)=omega(x) omega(y) (1-f_W^+(x) cap W^+(y)(t)). Now, the inverse of g is g^-1(x,y) = 1-f_W^-(x) cap W^-(y)(t)/t^dim(x cap y) and E[G] = 1-f_G(t)=sum_x,y g(x,y).

For a finite abstract simplicial complex G with n sets, define the n x n matrix K(x,y) which is the number of subsimplices in the intersection of x and y. We call it the counting matrix of G. Similarly as the connection matrix L which is L(x,y)=1 if x and y intersect and 0 else, the counting matrix K is unimodular. Actually, K is always in SL(n,Z). The inverse of K has the Green function entries K^(-1)(x,y)=w(x) w(y) |W^+(x) intersected W^+y|, where W^+(x) is the star of x, the sets in G which contain x and w(x)=(-1)^dim(x). The matrix K is always positive definite. The spectra of K and K^(-1) always agree so that the matrix Q=K-K^(-1) has the spectral symmetry spec(Q)=-spec(Q) and the zeta function z(s) summing l(k)^(-s) with eigenvalues l(k) of K satisfies the functional equation z(a+ib)=z(-a+ib). The energy theorem in this case tells that the sum of the matrix elements of K^(-1)(x,y) is equal to the number sets in G. In comparison, we had in the connection matrix case the identity that the sum of the matrix elements of L^(-1) is the Euler characteristic of G.

A finite abstract simplicial complex G defines a matrix L, where L(x,y)=1 if two simplicies x,y in G intersect and where L(x,y)=0 if they don't. This matrix is always unimodular so that the inverse g of L has integer entries g(x,y). In analogy to Laplacians on Euclidean spaces, these Green function entries define a potential energy between two simplices x,y. We prove that the total energy summing all matrix elements g(x,y) is equal to the Euler characteristic X(G) of G and that the number of positive minus the number of negative eigenvalues of L is equal to X(G).

Given a locally injective real function g on the vertex set V of a finite simple graph G=(V,E), we prove the Poincare-Hopf formula f_G(t) = 1+t sum_{x in V} f_{S_g(x)}(t), where S_g(x) = { y in S(x), g(y) less than g(x) } and f_G(t)=1+f_0 t + ... + f_{d} t^{d+1} is the f-function encoding the f-vector of a graph G, where f_k counts the number of k-dimensional cliques, complete sub-graphs, in G. The corresponding computation of f reduces the problem recursively to n tasks of graphs of half the size. For t=-1, the parametric Poincare-Hopf formula reduces to the classical Poincare-Hopf result X(G)=sum_x i_g(x), with integer indices i_g(x)=1-X(S_g(x)) and Euler characteristic X. In the new Poincare-Hopf formula, the indices are integer polynomials and the curvatures K_x(t) expressed as index expectations K_x(t) = E[i_x(t)] are polynomials with rational coefficients. Integrating the Poincare-Hopf formula over probability spaces of functions g gives Gauss-Bonnet formulas like f_G(t) = 1+sum_{x} F_{S(x)}(t), where F_G is the anti-derivative of f_G. A similar computation is done for the generating function f_{G,H}(t,s) = sum_{k,l} f_{k,l}(G,H) s^k t^l of the f-intersection matrix f_{k,l}(G,H) counting the number of intersections of k-simplices in G with l-simplices in H. Also here, the computation is reduced to 4 n^2 computations for graphs of half the size: f_{G,H}(t,s) = sum_{v,w} f_{B_g(v),B_g(w)}(t,s) - f_{B_g(v),S_g(w)}(t,s) - f_{S_g(v),B_g(w)}(t,s) + f_{S_g(v),S_g(w)}(t,s), where B_g(v)= S_g(v)+{v} is the unit ball of v.

In this note we revisit a "ring of graphs" Q in which the set of finite simple graphs N extend the role of the natural numbers and the signed graphs Z extend the role of the integers. We point out the existence of a norm which allows to complete Q to a real or complex Banach algebra R or C.