Searcharxiv⌕ Search

arXiv subjects

Oliver Knill

Publications and source records attributed to Oliver Knill.

At least 109 records · Page 6Linked to original sources
arXiv2014

Characteristic Length and Clustering

We explore relations between various variational problems for graphs like Euler characteristic chi(G), characteristic length mu(G), mean clustering nu(G), inductive dimension iota(G), edge density epsilon(G), scale measure sigma(G), Hilbert action eta(G) and spectral complexity xi(G). A new insight in this note is that the local cluster coefficient C(x) in a finite simple graph can be written as a relative characteristic length L(x) of the unit sphere S(x) within the unit ball B(x) of a vertex. This relation L(x) = 2-C(x) will allow to study clustering in more general metric spaces like Riemannian manifolds or fractals. If eta is the average of scalar curvature s(x), a formula mu ~ 1+log(epsilon)/log(eta) of Newman, Watts and Strogatz relates mu with the edge density epsilon and average scalar curvature eta telling that large curvature correlates with small characteristic length. Experiments show that the statistical relation mu ~ log(1/nu) holds for random or deterministic constructed networks, indicating that small clustering is often associated to large characteristic lengths and lambda=mu/log(nu) can converge in some graph limits of networks. Mean clustering nu, edge density epsilon and curvature average eta therefore can relate with characteristic length mu on a statistical level. We also discovered experimentally that inductive dimension iota and cluster-length ratio lambda correlate strongly on Erdos-Renyi probability spaces.

cs.DM↗
arXiv2014

Curvature from Graph Colorings

Given a finite simple graph G=(V,E) with chromatic number c and chromatic polynomial C(x). Every vertex graph coloring f of G defines an index i_f(x) satisfying the Poincare-Hopf theorem sum_x i_f(x)=chi(G). As a variant to the index expectation result we prove that E[i_f(x)] is equal to curvature K(x) satisfying Gauss-Bonnet sum_x K(x) = χ(G), where the expectation is the average over the finite probability space containing the C(c) possible colorings with c colors, for which each coloring has the same probability.

math.CO↗
arXiv2014

Cauchy-Binet for Pseudo-Determinants

The pseudo-determinant Det(A) of a square matrix A is defined as the product of the nonzero eigenvalues of A. It is a basis-independent number which is up to a sign the first nonzero entry of the characteristic polynomial of A. We prove Det(F^T G) = sum_P det(F_P) det(G_P) for any two n times m matrices F,G. The sum to the right runs over all k times k minors of A, where k is determined by F and G. If F=G is the incidence matrix of a graph this directly implies the Kirchhoff tree theorem as L=F^T G is then the Laplacian and det^2(F_P) in {0,1} is equal to 1 if P is a rooted spanning tree. A consequence is the following Pythagorean theorem: for any self-adjoint matrix A of rank k, one has Det^2(A) = sum_P det^2(A_P), where det(A_P) runs over k times k minors of A. More generally, we prove the polynomial identity det(1+x F^T G) = sum_P x^{|P|} det(F_P) det(G_P) for classical determinants det, which holds for any two n times m matrices F,G and where the sum on the right is taken over all minors P, understanding the sum to be 1 if |P|=0. It implies the Pythagorean identity det(1+F^T F) = sum_P det^2(F_P) which holds for any n times m matrix F and sums again over all minors F_P. If applied to the incidence matrix F of a finite simple graph, it produces the Chebotarev-Shamis forest theorem telling that det(1+L) is the number of rooted spanning forests in the graph with Laplacian L.

math.RA↗
arXiv2014

If Archimedes would have known functions

These are notes and slides from a Pecha-Kucha talk given on March 6, 2013. The presentation tinkered with the question whether calculus on graphs could have emerged by the time of Archimedes, if the concept of a function would have been available 2300 years ago. The text first attempts to boil down discrete single and multivariable calculus to one page each, then presents the slides with additional remarks and finally includes 40 "calculus problems" in a discrete or so-called 'quantum calculus' setting. We also added some sample Mathematica code, gave a short overview over the emergence of the function concept in calculus and included comments on the development of calculus textbooks over time.

math.HO↗
arXiv2014

A notion of graph homeomorphism

We introduce a notion of graph homeomorphisms which uses the concept of dimension and homotopy for graphs. It preserves the dimension of a subbasis, cohomology and Euler characteristic. Connectivity and homotopy look as in classical topology. The Brouwer-Lefshetz fixed point leads to the following discretiszation of the Kakutani fixed point theorem: any graph homeomorphism T with nonzero Lefschetz number has a nontrivial invariant open set which is fixed by T.

math.GN↗
arXiv2013

The zeta function for circular graphs

We study the entire function zeta(n,s) which is the sum of l to the power -s, where l runs over the positive eigenvalues of the Laplacian of the circular graph C(n) with n vertices. We prove that the roots of zeta(n,s) converge for n to infinity to the line Re(s)=1/2 in the sense that for every compact subset K in the complement of this line, and large enough n, no root of the zeta function zeta(n,s) is in K. To prove this, we look at the Dirac zeta function, which uses the positive eigenvalues of the Dirac operator D=d+d^* of the circular graph, the square root of the Laplacian. We extend a Newton-Coates-Rolle type analysis for Riemann sums and use a derivative which has similarities with the Schwarzian derivative. As the zeta functions zeta(n,s) of the circular graphs are entire functions, the result does not say anything about the roots of the classical Riemann zeta function zeta(s), which is also the Dirac zeta function for the circle. Only for Re(s)>1, the values of zeta(n,s) converge suitably scaled to zeta(s). We also give a new solution to the discrete Basel problem which is giving expressions like zeta_n(2) = (n^2-1)/12 or zeta_n(4) = (n^2-1)(n^2+11)/45 which allows to re-derive the values of the classical Basel problem zeta(2) = pi^2/6 or zeta(4)=pi^4/90 in the continuum limit.

math.SP↗
arXiv2013

On quadratic orbital networks

These are some informal remarks on quadratic orbital networks over finite fields. We discuss connectivity, Euler characteristic, number of cliques, planarity, diameter and inductive dimension. We find a non-trivial disconnected graph for d=3. We prove that for d=1 generators, the Euler characteristic is always non-negative and for d=2 and large enough p the Euler characteristic is negative. While for d=1, all networks are planar, we suspect that for d larger or equal to 2 and large enough prime p, all networks are non-planar. As a consequence on bounds for the number of complete sub graphs of a fixed dimension, the inductive dimension of all these networks goes 1 as p goes to infinity.

math.DS↗
arXiv2013

Natural orbital networks

Given a finite set T of maps on a finite ring R, we look at the finite simple graph G=(V,E) with vertex set V=R and edge set E={(a,b) | exists t in T, b=t(a), b not equal to a}. An example is when R=Z_n and T consists of a finite set of quadratic maps T_i(x)=x^2+a_i. Graphs defined like that have a surprisingly rich structure. This holds especially in an algebraic set-up when T is generated by polynomials on Z_n. The characteristic path length L and the mean clustering coefficient C are interlinked by global-local quantity LC=-L/log(C) which often appears to have a limit for n to infinity like for two quadratic maps on a finite field Z_p. We see that for one quadratic map x^2+a, the probability to have connectedness goes to zero and for two quadratic maps, the probability goes to 1, for three different quadratic maps x^2+a,x^2+b,x^2+c on Z_p, we always appear to get a connected graph for all primes.

math.DS↗
arXiv2013

Dynamically generated Networks

Simple algebraic rules can produce complex networks with rich structures. These graphs are obtained when looking at a monoid operating on a ring. There are relations to dynamical systems theory and number theory. This document illustrates this class of networks introduced together with Montasser Ghachem. Besides showing off pictures, we look at elementary results related to the Chinese remainder theorem, the Collatz problem, the Artin constant, Fermat primes and Pierpont primes.

math.DS↗
arXiv2013

Counting rooted forests in a network

We use a recently found generalization of the Cauchy-Binet theorem to give a new proof of the Chebotarev-Shamis forest theorem telling that det(1+L) is the number of rooted spanning forests in a finite simple graph G with Laplacian L. More generally, we show that det(1+k L) is the number of rooted edge-k-colored spanning forests in G. If a forest with an even number of edges is called even, then det(1-L) is the difference between even and odd rooted spanning forests in G.

math.SP↗
arXiv2013

The Euler characteristic of an even-dimensional graph

We write the Euler characteristic X(G) of a four dimensional finite simple geometric graph G=(V,E) in terms of the Euler characteristic X(G(w)) of two-dimensional geometric subgraphs G(w). The Euler curvature K(x) of a four dimensional graph satisfying the Gauss-Bonnet relation sum_x K(x) = X(G) can so be rewritten as an average 1-E[K(x,f)]/2 over a collection two dimensional "sectional graph curvatures" K(x,f) through x. Since scalar curvature, the average of all these two dimensional curvatures through a point, is the integrand of the Hilbert action, the integer 2-2 X(G) becomes an integral-geometrically defined Hilbert action functional. The result has an interpretation in the continuum for compact 4-manifolds M: the Euler curvature K(x), the integrand in the classical Gauss-Bonnet-Chern theorem, can be seen as an average over a probability space W of 1-K(x,w)/2 with curvatures K(x,w) of compact 2-manifolds M(w). Also here, the Euler characteristic has an interpretation of an exotic Hilbert action, in which sectional curvatures are replaced by surface curvatures of integral geometrically defined random two-dimensional sub-manifolds M(w) of M.

math.GT↗
arXiv2013

Isospectral deformations of the Dirac operator

We give more details about an integrable system in which the Dirac operator D=d+d^* on a finite simple graph G or Riemannian manifold M is deformed using a Hamiltonian system D'=[B,h(D)] with B=d-d^* + i b. The deformed operator D(t) = d(t) + b(t) + d(t)^* defines a new exterior derivative d(t) and a new Dirac operator C(t) = d(t) + d(t)^* and Laplacian M(t) = d(t) d(t)^* + d(t)* d(t) and so a new distance on G or a new metric on M.

math-ph↗
arXiv2013

Illustrating Mathematics using 3D Printers

3D printing technology can help to visualize proofs in mathematics. In this document we aim to illustrate how 3D printing can help to visualize concepts and mathematical proofs. As already known to educators in ancient Greece, models allow to bring mathematics closer to the public. The new 3D printing technology makes the realization of such tools more accessible than ever. This is an updated version of a paper included in book Low-Cost 3D Printing for science, education and Sustainable Development, ICTP, 2013 edited by Carlo Fonda Enrique Canessa and Marco Zennaro.

math.HO↗
arXiv2013

An integrable evolution equation in geometry

We introduce an integrable Hamiltonian system which Lax deforms the Dirac operator D=d+d* on a finite simple graph or compact Riemannian manifold. We show that the nonlinear isospectral deformation always leads to an expansion of the original space, featuring a fast inflationary start. The nonlinear evolution leaves the Laplacian L=D^2 invariant so that linear Schroedinger or wave dynamics is not affected. The expansion has the following effects: a complex structure can develop and the nonlinear quantum mechanics asymptotically becomes the linear relativistic Dirac wave equation u''=Lu. While the later is not aware of the expansion of space and does not see the emerged complex structure, nor the larger non-commutative geometric setup, the nonlinear flow is affected by it. The natural Noether symmetries of quantum mechanics introduced here force to consider space as part of a larger complex geometry. The nonlinear evolution equation is a symmetry of quantum mechanics which still features super-symmetry, but it becomes clear why it is invisible: while the McKean-Singer formulas str(exp(i D(t) t)) = str(exp(-L t))=chi(G) still hold, the super-partners f,Df are orthogonal only at t=0 and become parallel or anti-parallel for |t| to infinity.

math.DS↗
arXiv2013

Thinking like Archimedes with a 3D printer

We illustrate Archimedes' method using models produced with 3D printers. This approach allowed us to create physical proofs of results known to Archimedes and illustrate ideas of a mathematician who is known both for his for his mechanical inventions as well as his breakthroughs in geometry and calculus. We use technology from the 21st century to trace intellectual achievements from the 3rd century BC. While we celebrate the 2300th birthday of Archimedes (287-212 BC) in 2013, we also live in an exciting time, where 3D printing is becoming popular and affordable.

math.HO↗
arXiv2013

The McKean-Singer Formula in Graph Theory

For any finite simple graph G=(V,E), the discrete Dirac operator D=d+d* and the Laplace-Beltrami operator L=d d* + d* d on the exterior algebra bundle Omega are finite v times v matrices, where dim(Omega) = v is the sum of the cardinalities v(k) of the set G(k) of complete subgraphs K(k) of G. We prove the McKean-Singer formula chi(G) = str(exp(-t L)) which holds for any complex time t, where chi(G) = str(1)= sum (-1)k v(k) is the Euler characteristic of G. The super trace of the heat kernel interpolates so the Euler-Poincare formula for t=0 with the Hodge theorem in the real limit t going to infinity. More generally, for any continuous complex valued function f satisfying f(0)=0, one has the formula chi(G) = str(exp(f(D))). This includes for example the Schroedinger evolutions chi(G) = str(cos(t D)) on the graph. After stating some general facts about the spectrum of D which includes statements about the complexity, the product of the non-zero eigenvalues as well as a perturbation result estimating the spectral difference of two graphs, we mention as a combinatorial consequence that the spectrum of D encodes the number of closed paths in the simplex space of a graph. McKean-Singer implies that the number of closed paths of length n starting at an even dimensional simplex is the same than the number of closed paths of length n starting at an odd dimensional simplex. We give a couple of worked out examples and see that McKean-Singer allows to find explicit pairs of non-isometric graphs which have isospectral Dirac operators.

math.CO↗