arXiv subjects
Publications and source records attributed to Mikko Alava.
We introduce a new local search algorithm for satisfiability problems. Usual approaches focus uniformly on unsatisfied clauses. The new method works by picking uniformly random variables in unsatisfied clauses. A Variable-based Focused Metropolis Search (V-FMS) is then applied to random 3-SAT. We show that it is quite comparable in performance to the clause-based FMS. Consequences for algorithmic design are discussed.
In order to use the advanced inference techniques available for Ising models, we transform complex data (real vectors) into binary strings, by local averaging and thresholding. This transformation introduces parameters, which must be varied to characterize the behaviour of the system. The approach is illustrated on financial data, using three inference methods -- equilibrium, synchronous and asynchronous inference -- to construct functional connections between stocks. We show that the traded volume information is enough to obtain well known results about financial markets, which use however the presumably richer price information: collective behaviour ("market mode") and strong interactions within industry sectors. Synchronous and asynchronous Ising inference methods give results which are coherent with equilibrium ones, and more detailed since the obtained interaction networks are directed.
We describe how the couplings in an asynchronous kinetic Ising model can be inferred. We consider two cases, one in which we know both the spin history and the update times and one in which we only know the spin history. For the first case, we show that one can average over all possible choices of update times to obtain a learning rule that depends only on spin correlations and can also be derived from the equations of motion for the correlations. For the second case, the same rule can be derived within a further decoupling approximation. We study all methods numerically for fully asymmetric Sherrington-Kirkpatrick models, varying the data length, system size, temperature, and external field. Good convergence is observed in accordance with the theoretical expectations.
The random 3-satisfiability (3-SAT) problem is in the unsatisfiable (UNSAT) phase when the clause density $α$ exceeds a critical value $α_s \approx 4.267$. However, rigorously proving the unsatisfiability of a given large 3-SAT instance is extremely difficult. In this paper we apply the mean-field theory of statistical physics to the unsatisfiability problem, and show that a specific type of UNSAT witnesses (Feige-Kim-Ofek witnesses) can in principle be constructed when the clause density $α> 19$. We then construct Feige-Kim-Ofek witnesses for single 3-SAT instances through a simple random sampling algorithm and a focused local search algorithm. The random sampling algorithm works only when $α$ scales at least linearly with the variable number $N$, but the focused local search algorithm works for clause densty $α> c N^{b}$ with $b \approx 0.59$ and prefactor $c \approx 8$. The exponent $b$ can be further decreased by enlarging the single parameter $S$ of the focused local search algorithm.
Cellular phones are now offering an ubiquitous means for scientists to observe life: how people act, move and respond to external influences. They can be utilized as measurement devices of individual persons and for groups of people of the social context and the related interactions. The picture of human life that emerges shows complexity, which is manifested in such data in properties of the spatiotemporal tracks of individuals. We extract from smartphone-based data for a set of persons important locations such as "home", "work" and so forth over fixed length time-slots covering the days in the data-set. This set of typical places is heavy-tailed, a power-law distribution with an exponent close to -1.7. To analyze the regularities and stochastic features present, the days are classified for each person into regular, personal patterns. To this are superimposed fluctuations for each day. This randomness is measured by "life" entropy, computed both before and after finding the clustering so as to subtract the contribution of a number of patterns. The main issue, that we then address, is how predictable individuals are in their mobility. The patterns and entropy are reflected in the predictability of the mobility of the life both individually and on average. We explore the simple approaches to guess the location from the typical behavior, and of exploiting the transition probabilities with time from location or activity A to B. The patterns allow an enhanced predictability, at least up to a few hours into the future from the current location. Such fixed habits are most clearly visible in the working-day length.
Network structures are reconstructed from dynamical data by respectively naive mean field (nMF) and Thouless-Anderson-Palmer (TAP) approximations. For TAP approximation, we use two methods to reconstruct the network: a) iteration method; b) casting the inference formula to a set of cubic equations and solving it directly. We investigate inference of the asymmetric Sherrington- Kirkpatrick (S-K) model using asynchronous update. The solutions of the sets cubic equation depend of temperature T in the S-K model, and a critical temperature Tc is found around 2.1. For T < Tc, the solutions of the cubic equation sets are composed of 1 real root and two conjugate complex roots while for T > Tc there are three real roots. The iteration method is convergent only if the cubic equations have three real solutions. The two methods give same results when the iteration method is convergent. Compared to nMF, TAP is somewhat better at low temperatures, but approaches the same performance as temperature increase. Both methods behave better for longer data length, but for improvement arises, TAP is well pronounced.
Elastic interfaces embedded in (quenched) random media exhibit meta-stability and stick-slip dynamics. These non-trivial dynamical features have been shown to be associated with cusp singularities of the coarse-grained disorder correlator. Here we show that annealed systems with many absorbing states and a conservation law but no quenched disorder exhibit identical cusps. On the other hand, similar non-conserved systems in the directed percolation class, are also shown to exhibit cusps, but of a different type. These results are obtained both by a recent method to explicitly measure disorder correlators and by defining an alternative new protocol, inspired by self-organized criticality, which opens the door to easily accessible experimental realizations.
A two-species spatially extended system of hosts and parasitoids is studied. There are two distinct kinds of coexistence; one with populations distributed homogeneously in space and another one with spatiotemporal patterns. In the latter case, there are noise-sustained oscillations in the population densities, whereas in the former one the densities are essentially constants in time with small fluctuations. We introduce several metrics to characterize the patterns and onset thereof. We also build a consistent sequence of corrections to the mean-field equations using a posteriori knowledge from simulations. These corrections both lead to better description of the dynamics and connect the patterns to it. The analysis is readily applicable to realistic systems, which we demonstrate by an example using an empirical metapopulation landscape.
Interacting populations often create complicated spatiotemporal behavior, and understanding it is a basic problem in the dynamics of spatial systems. We study the two-species case by simulations of a host--parasitoid model. In the case of co-existence, there are spatial patterns leading to noise-sustained oscillations. We introduce a new measure for the patterns, and explain the oscillations as a consequence of a timescale separation and noise. They are linked together with the patterns by letting the spreading rates depend on instantaneous population densities. Applications are discussed.
Cyclic dominance of three species is a commonly occurring interaction dynamics, often denoted the rock-paper-scissors (RPS) game. Such type of interactions is known to promote species coexistence. Here, we generalize recent results of Reichenbach et al. (e.g. Nature 448, 1046 (2007)) of a four-state variant of RPS. We show that spiral formation takes place only without a conservation law for the total density. Nevertheless, in general fast diffusion can destroy species coexistence. We also generalize the four-state model to slightly varying reaction rates. This is shown both analytically and numerically not to change pattern formation, or the effective wave length of the spirals, and therefore does not alter the qualitative properties of the cross-over to extinction.
We study anomalous scaling and multiscaling of two-dimensional crack profiles in the random fuse model using both periodic and open boundary conditions. Our large scale and extensively sampled numerical results reveal the importance of crack branching and coalescence of microcracks, which induce jumps in the solid-on-solid crack profiles. Removal of overhangs (jumps) in the crack profiles eliminates the multiscaling observed in earlier studies and reduces anomalous scaling. We find that the probability density distribution $p(Δh(\ell))$ of the height differences $Δh(\ell) = [h(x+\ell) - h(x)]$ of the crack profile obtained after removing the jumps in the profiles has the scaling form $p(Δh(\ell)) = <Δh^2(\ell)>^{-1/2} ~f(\frac{Δh(\ell)}{<Δh^2(\ell)>^{1/2}})$, and follows a Gaussian distribution even for small bin sizes $\ell$. The anomalous scaling can be summarized with the scaling relation $[\frac{<Δh^2(\ell)>^{1/2}}{<Δh^2(L/2)>^{1/2}}]^{1/ζ_{loc}} + \frac{(\ell-L/2)^2}{(L/2)^2} = 1$, where $<Δh^2(L/2)>^{1/2} \sim L^ζ$.
We study the scaling of two-dimensional crack roughness using large scale beam lattice systems. Our results indicate that the crack roughness obtained using beam lattice systems does not exhibit anomalous scaling in sharp contrast to the simulation results obtained using scalar fuse lattices. The local and global roughness exponents ($ζ_{loc}$ and $ζ$, respectively) are equal to each other, and the two-dimensional crack roughness exponent is estimated to be $ζ_{loc} = ζ= 0.64 \pm 0.02$. Removal of overhangs (jumps) in the crack profiles eliminates even the minute differences between the local and global roughness exponents. Furthermore, removing these jumps in the crack profile completely eliminates the multiscaling observed in other studies. We find that the probability density distribution $p(Δh(\ell))$ of the height differences $Δh(\ell) = [h(x+\ell) - h(x)]$ of the crack profile obtained after removing the jumps in the profiles follows a Gaussian distribution even for small window sizes ($\ell$).
We study the performance of stochastic local search algorithms for random instances of the $K$-satisfiability ($K$-SAT) problem. We introduce a new stochastic local search algorithm, ChainSAT, which moves in the energy landscape of a problem instance by {\em never going upwards} in energy. ChainSAT is a \emph{focused} algorithm in the sense that it considers only variables occurring in unsatisfied clauses. We show by extensive numerical investigations that ChainSAT and other focused algorithms solve large $K$-SAT instances almost surely in linear time, up to high clause-to-variable ratios $α$; for example, for K=4 we observe linear-time performance well beyond the recently postulated clustering and condensation transitions in the solution space. The performance of ChainSAT is a surprise given that by design the algorithm gets trapped into the first local energy minimum it encounters, yet no such minima are encountered. We also study the geometry of the solution space as accessed by stochastic local search algorithms.
Interfaces moving in a disordered medium exhibit stochastic velocity fluctuations obeying universal scaling relations related to the presence or absence of conservation laws. For fluid invasion of porous media, we show that the fluctuations of the velocity are governed by a geometry-dependent length scale arising from fluid conservation. This result is compared to the statistics resulting from a non-equilibrium (depinning) transition between a moving interface and a stationary, pinned one.
We study the intermittency and noise of dislocation systems undergoing shear deformation. Simulations of a simple two-dimensional discrete dislocation dynamics model indicate that the deformation rate exhibits a power spectrum scaling of the type $1/f^α$. The noise exponent is far away from a Lorentzian, with $α\approx 1.5$. This result is directly related to the way the durations of avalanches of plastic deformation activity scale with their size.
We investigate by exact optimization method properties of two- and three-dimensional systems of elastic lines in presence of splayed columnar disorder. The ground state of many lines is separable both in 2d and 3d leading to a random walk -like roughening in 2d and ballistic behavior in 3d. Furthermore, we find that in the case of pure splayed columnar disorder in contrast to point disorder there is no entanglement transition in 3d. Entanglement can be triggered by perturbing the pure splay system with point defects.
We investigate the properties of quantum annealing applied to the random field Ising model in one, two and three dimensions. The decay rate of the residual energy, defined as the energy excess from the ground state, is find to be $e_{res}\sim \log(N_{MC})^{-ζ}$ with $ζ$ in the range $2...6$, depending on the strength of the random field. Systems with ``large clusters'' are harder to optimize as measured by $ζ$. Our numerical results suggest that in the ordered phase $ζ=2$ whereas in the paramagnetic phase the annealing procedure can be tuned so that $ζ\to6$.
Collaboration networks are studied as an example of growing bipartite networks. These have been previously observed to have structure such as positive correlations between nearest-neighbour degrees. However, a detailed understanding of the origin of this phenomenon and the growth dynamics is lacking. Both of these are analyzed empirically and simulated using various models. A new one is presented, incorporating empirically necessary ingredients such as bipartiteness and sublinear preferential attachment. This, and a recently proposed model of team assembly both agree roughly with some empirical observations and fail in several others.