SearcharxivSearch

arXiv subjects

Toshiaki Omori

Publications and source records attributed to Toshiaki Omori.

4 recordsLinked to original sources

Eigenvalues of the Laplacian on the Goldberg-Coxeter constructions for $3$- and $4$-valent graphs

We are concerned with spectral problems of the Goldberg-Coxeter construction for $3$- and $4$-valent finite graphs. The Goldberg-Coxeter constructions $\mathrm{GC}_{k,l}(X)$ of a finite $3$- or $4$-valent graph $X$ are considered as "subdivisions" of $X$, whose number of vertices are increasing at order $O(k^2+l^2)$, nevertheless which have bounded girth. It is shown that the first (resp. the last) $o(k^2)$ eigenvalues of the combinatorial Laplacian on $\mathrm{GC}_{k,0}(X)$ tend to $0$ (resp. tend to $6$ or $8$ in the $3$- or $4$-valent case, respectively) as $k$ goes to infinity. A concrete estimate for the first several eigenvalues of $\mathrm{GC}_{k,l}(X)$ by those of $X$ is also obtained for general $k$ and $l$. It is also shown that the specific values always appear as eigenvalues of $\mathrm{GC}_{2k,0}(X)$ with large multiplicities almost independently to the structure of the initial $X$. In contrast, some dependency of the graph structure of $X$ on the multiplicity of the specific values is also studied.

math.CO

A Discrete Surface Theory

In the present paper, we propose a new discrete surface theory on 3-valent embedded graphs in the 3-dimensional Euclidean space which are not necessarily discretization or approximation of smooth surfaces. The Gauss curvature and the mean curvature of discrete surfaces are defined which satisfy properties corresponding to the classical surface theory. We also discuss the convergence of a family of subdivided discrete surfaces of a given 3-valent discrete surface by using the Goldberg-Coxeter construction. Although discrete surfaces in general have no corresponding smooth surfaces, we may find one as the limit.

math.DG

Neural network model with discrete and continuous information representation

An associative memory model and a neural network model with a Mexican-hat type interaction are the two most typical attractor networks used in the artificial neural network models. The associative memory model has discretely distributed fixed-point attractors, and achieves a discrete information representation. On the other hand, a neural network model with a Mexican-hat type interaction uses a line attractor to achieves a continuous information representation, which can be seen in the working memory in the prefrontal cortex and columnar activity in the visual cortex. In the present study, we propose a neural network model that achieves discrete and continuous information representation. We use a statistical-mechanical analysis to find that a localized retrieval phase exists in the proposed model, where the memory pattern is retrieved in the localized subpopulation of the network. In the localized retrieval phase, the discrete and continuous information representation is achieved by using the orthogonality of the memory patterns and the neutral stability of fixed points along the positions of the localized retrieval. The obtained phase diagram suggests that the antiferromagnetic interaction and the external field are important for generating the localized retrieval phase.

q-bio.NC

Statistical mechanics of lossy compression for non-monotonic multilayer perceptrons

A lossy data compression scheme for uniformly biased Boolean messages is investigated via statistical mechanics techniques. We utilize tree-like committee machine (committee tree) and tree-like parity machine (parity tree) whose transfer functions are non-monotonic. The scheme performance at the infinite code length limit is analyzed using the replica method. Both committee and parity treelike networks are shown to saturate the Shannon bound. The AT stability of the Replica Symmetric solution is analyzed, and the tuning of the non-monotonic transfer function is also discussed.

cond-mat.stat-mech