Large deviation principles for the Hopfield model and the Kac-Hopfield model

Summary We study the Kac version of the Hopfield model and prove a Lebowitz-Penrose theorem for the distribution of the overlap parameters. At the same time, we prove a large deviation principle for the standard Hopfield model with infinitely many patterns.Work partially supported by the Commission of the European Communities under contract No. SC1-CT91-0695     

Gibbs states of the Hopfield model in the regime of perfect memory

Summary We study the thermodynamic properties of the Hopfield model of an autoassociative memory. IfN denotes the number of neurons andM (N) the number of stored patterns, we prove the following results: IfM/N 0 asN , then there exists an infinite number of infinite volume Gibbs measures for all temperaturesT<1 concentrated on spin configurations that have overlap with exactly one specific pattern. Moreover, the measures induced on the overlap parameters are Dirac measures concentrated on a single point and the Gibbs measures on spin configurations are products of Bernoulli measures. IfM/N , asN for small enough, we show that for temperaturesT smaller than someT()<1, the induced measures can have support only on a disjoint union of balls around the previous points, but we cannot construct the infinite volume measures through convergent sequences of measures.Work partially supported by the Commission of the European Communities under contract No. SC1-CT91-0695     

The swapping algorithm for the Hopfield model with two patterns

The so-called swapping algorithm was designed to simulate from spin glass distributions, among others. In this note we show that it mixes rapidly, in a very simple disordered system, the Hopfield model with two patterns.     

Large deviations for Markov processes with mean field interaction and unbounded jumps

Summary An N-particle system with mean field interaction is considered. The large deviation estimates for the empirical distributions as N goes to infinity are obtained under conditions which are satisfied, by many interesting models including the first and the second Schlögl models.Supported partially by a scholarship from the Faculty of Graduate Studies and Research of Carleton University and the NSERC operating grant of D.A. Dawson     

Sharp asymptotics for Kawasaki dynamics on a finite box with open boundary

In this paper we study the metastable behavior of the lattice gas in two and three dimensions subject to Kawasaki dynamics in the limit of low temperature and low density. We consider the local version of the model, where particles live on a finite box and are created, respectively, annihilated at the boundary of the box in a way that reflects an infinite gas reservoir. We are interested in how the system nucleates, i.e., how it reaches a full box when it starts from an empty box. Our approach combines geometric and potential theoretic arguments. In two dimensions, we identify the full geometry of the set of critical droplets for the nucleation, compute the average nucleation time up to a multiplicative factor that tends to one, show that the nucleation time divided by its average converges to an exponential random variable, express the proportionality constant for the average nucleation time in terms of certain capacities associated with simple random walk, and compute the asymptotic behavior of this constant as the system size tends to infinity. In three dimensions, we obtain similar results but with less control over the geometry and the constant. A special feature of Kawasaki dynamics is that in the metastable regime particles move along the border of a droplet more rapidly than they arrive from the boundary of the box. The geometry of the critical droplet and the sharp asymptotics for the average nucleation time are highly sensitive to this motion.     

Random graph approach to Gibbs processes with pair interaction

This study was motivated by the observation that, in a broad class of cases, the distribution of classical Gibbs point processes in R ^d governed by pair potential, can be obtained as the equilibrium distribution of a Markov chain of point processes in R ^d. Our analysis of this Markov chain is based on its imbedding in an infinite random graph. A condition of ergodicity of the chain is given in terms of the absence of percolation in the graph, and this can be checked in simpler cases. The embedding also suggests a stochastic construction for the equilibrium distribution in question.These constructions (which can also be of independent interest) are related to Gibbs processes by means of the results obtained in a recent paper of R. V. Ambartzumian and H. S. Sukiasian [1] where the existence of a new class of stationary point processes in R ^d was established which have density (correlation) functions of the form % MathType!MTEF!2!1!+-% feaafiart1ev1aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn% hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr% 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq-Jc9% vqaqpepm0xbba9pwe9Q8fs0-yqaqpepae9pg0FirpepeKkFr0xfr-x% fr-xb9adbaqaaeGaciGaaiaabeqaamaabaabaaqdbaWexLMBb50ujb% qegi0BVTgib5gDPfxDHbacfiGae8NKbmOae8hkaGIae8hEaG3aaSba% a4qaaiaabgdaaeqaa0Gaaiilaiaac6cacaGGUaGaaiOlaiaacYcaca% WG4bWaaSbaa4qaaiaad6gaaeqaa0GaaiykaiaabccacqGH9aqpcaqG% GaGaamOyamaaCaaaoeqabaGaamOBaaaanmaarababaGaamiAaiaacI% cacaWG4bWaaSbaa4qaaiaadMgaaeqaaaqaaiaad6gaaeqaniabg+Gi% vdGaeyOeI0IaamiEamaaBaaaoeaacaWGQbaabeaaniaacMcaaaa!56B6!\[f(x_{\text{1}} ,...,x_n ){\text{ }} = {\text{ }}b^n \prod\nolimits_n {h(x_i } - x_j )\] (here and below, % MathType!MTEF!2!1!+-% feaafiart1ev1aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn% hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr% 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq-Jc9% vqaqpepm0xbba9pwe9Q8fs0-yqaqpepae9pg0FirpepeKkFr0xfr-x% fr-xb9adbaqaaeGaciGaaiaabeqaamaabaabaaqdbaWaaebaaeaada% WgaaGdbaGaamOBaaqabaaabeqab0Gaey4dIunaaaa!38CA!\[\prod {_n } \] denotes a product taken over all two-subsets {i, j} {1,..., n}).     

Aging for interacting diffusion processes

We study the aging phenomenon for a class of interacting diffusion processes {Xt(i),i∈Zd}. In this framework we see the effect of the lattice dimension d on aging, as well as that of the class of test functions f(Xt) considered. We further note the sensitivity of aging to specific details, when degenerate diffusions (such as super random walk, or parabolic Anderson model), are considered. We complement our study of systems on the infinite lattice, with that of their restriction to finite boxes. In the latter setting we consider different regimes in terms of box size scaling with time, as well as the effect that the choice of boundary conditions has on aging. The key tool for our analysis is the random walk representation for such diffusions.     

Finite clusters in high-density continuous percolation: Compression and sphericality

Summary A percolation process inR ^d is considered in which the sites are a Poisson process with intensity and the bond between each pair of sites is open if and only if the sites are within a fixed distancer of each other. The distribution of the number of sites in the clusterC of the origin is examined, and related to the geometry ofC. It is shown that when andk are large, there is a characteristic radius such that conditionally on |C|=k, the convex hull ofC closely approximates a ball of radius , with high probability. When the normal volumek/ thatk points would occupy is small, the cluster is compressed, in that the number of points per unit volume in this -ball is much greater than the ambient density . For larger normal volumes there is less compression. This can be compared to Bernoulli bond percolation on the square lattice in two dimensions, where an analog of this compression is known not to occur.Research supported by NSF grant number DMS-9006395     

Percolation on random Johnson–Mehl tessellations and related models

We make use of the recent proof that the critical probability for percolation on random Voronoi tessellations is 1/2 to prove the corresponding result for random Johnson–Mehl tessellations, as well as for two-dimensional slices of higher-dimensional Voronoi tessellations. Surprisingly, the proof is a little simpler for these more complicated models. B. Bollobás’s research was supported in part by NSF grants CCR-0225610 and DMS-0505550 and ARO grant W911NF-06-1-0076. O. Riordan’s research was supported by a Royal Society Research Fellowship.     

Some rigorous results on semiflexible polymers,I: Free and confined polymers

We introduce a class of models of semiflexible polymers. The latter are characterized by a strong rigidity, the correlation length associated with the gradient–gradient correlations, called the persistence length, being of the same order as the polymer length.     

Critical phenomenon for a percolation model

We consider a percolation model which consists of oriented lines placed randomly on the plane. The lines are of random length and at a random angle with respect to the horizontal axis and are placed according to a Poisson point process; the length, angle, and orientation being independent of the underlying Poisson process. We establish a critical behaviour of this model, i.e., percolation occurs for large intensity of the Poisson process and does not occur for smaller intensities. In the special case when the lines are of fixed unit length and are either oriented vertically up or oriented horizontally to the left, with probability p or (1-p), respectively, we obtain a lower bound on the critical intensity of percolation.     

The diffusive phase of a model of self-interacting walks

Summary We consider simple random walk onZ ^d perturbed by a factor exp[T ^–P J _T], whereT is the length of the walk and . Forp=1 and dimensionsd2, we prove that this walk behaves diffusively for all – < <₀, with ₀ > 0. Ford>2 the diffusion constant is equal to 1, but ford=2 it is renormalized. Ford=1 andp=3/2, we prove diffusion for all real (positive or negative). Ford>2 the scaling limit is Brownian motion, but ford2 it is the Edwards model (with the wrong sign of the coupling when >0) which governs the limiting behaviour; the latter arises since for ,T ^–p J _T is the discrete self-intersection local time. This establishes existence of a diffusive phase for this model. Existence of a collapsed phase for a very closely related model has been proven in work of Bolthausen and Schmock.     

Mean-field lattice trees

We introduce a mean-field model of lattice trees based on embeddings into ^d of abstract trees having a critical Poisson offspring distribution. This model provides a combinatorial interpretation for the self-consistent mean-field model introduced previously by Derbez and Slade [9], and provides an alternative approach to work of Aldous. The scaling limit of the meanfield model is integrated super-Brownian excursion (ISE), in all dimensions. We also introduce a model of weakly self-avoiding lattice trees, in which an embedded tree receives a penaltye ^– for each self-intersection. The weakly self-avoiding lattice trees provide a natural interpolation between the mean-field model (=0), and the usual model of strictly self-avoiding lattice tress (=) which associates the uniform measure to the set of lattice trees of the same size.     

Two scales hydrodynamic limit for a model of malignant tumor cells

We consider a model introduced in [S. Luckhaus, L. Triolo, The continuum reaction-diffusion limit of a stochastic cellular growth model, Rend. Acc. Lincei (S.9) 15 (2004) 215-223] with two species (η and ξ) of particles, representing respectively malignant and normal cells. The basic motions of the η particles are independent random walks, scaled diffusively. The ξ particles move on a slower time scale and obey an exclusion rule among themselves and with the η particles. The competition between the two species is ruled by a coupled birth and death process. We prove convergence in the hydrodynamic limit to a system of two reaction-diffusion equations with measure valued initial data.     

The Legendre transform of two replicas of the Sherrington-Kirkpatrick spin glass model

Summary We prove a variational inequality linking the values of the free energy per site at different temperatures. This inequality is based on the Legendre transform of the free energy of two replicas of the system. We prove that equality holds when1/ and fails when 1/ <1. We deduce from this that the mean entropy per site of the uniform distribution with respect to the distribution of the coupling _i ¹ _i ² = _i between two replicas is null when 01/ and strictly positive when 1/ <1. We exhibit thus a new secondary critical phenomenon within the high temperature region 01. We given an interpretation of this phenomenon showing that the fluctuations of the law of the coupling with the interactions remains strong in the thermodynamic limit when>1/ . We also use our inequality numerically within the low temperature region to improve (slightly) the best previously known lower bounds for the free energy and the ground state energy per site.     

Percolation of even sites for random sequential adsorption

Consider random sequential adsorption on a red/blue chequerboard lattice with arrivals at rate 1 on the red squares and rate λ$\lambda$ on the blue squares. We prove that the critical value of λ$\lambda$, above which we get an infinite blue component, is finite and strictly greater than 1.     

Large scale properties of the IIIC for 2D percolation

We reinvestigate the 2D problem of the inhomogeneous incipient infinite cluster   where, in an independent percolation model, the density decays to _pc$p_c$ with an inverse power, λ$\lambda$, of the distance to the origin. Assuming the existence of critical exponents (as is known in the case of the triangular site lattice) if the power is less than 1/ν$1/\nu$, with ν$\nu$ the correlation length exponent, we demonstrate an infinite cluster with scale dimension given by _DH=2−βλ$D_H=2-\beta \lambda$. Further, we investigate the critical case _λc=1/ν${\lambda }_c=1/\nu$ and show that iterated logarithmic corrections will tip the balance between the possibility and impossibility of an infinite cluster.     

Gibbsianness of fermion random point fields

We consider fermion (or determinantal) random point fields on Euclidean space ℝ^d. Given a bounded, translation invariant, and positive definite integral operator J on L²(ℝ^d), we introduce a determinantal interaction for a system of particles moving on ℝ^d as follows: the n points located at x₁,· · ·,x_n ∈ ℝ^d have the potential energy given by where j(x−y) is the integral kernel function of the operator J. We show that the Gibbsian specification for this interaction is well-defined. When J is of finite range in addition, and for d≥2 if the intensity is small enough, we show that the fermion random point field corresponding to the operator J(I+J)⁻¹ is a Gibbs measure admitted to the specification.