1st Nordic Workshop on Statistical Physics: Biological, Complex and Nonequilibrium Systems
from
Wednesday, March 17, 2010 (9:00 AM)
to
Friday, March 19, 2010 (6:00 PM)
Monday, March 15, 2010
Tuesday, March 16, 2010
Wednesday, March 17, 2010
9:00 AM
Breakfast and Registration
Breakfast and Registration
9:00 AM  10:00 AM
10:00 AM
Opening
Opening
10:00 AM  10:15 AM
Room: Nordita Seminar Room 132:028
10:15 AM
DNA Analysis in Nanostructured Devices

Jonas Tegenfeldt
DNA Analysis in Nanostructured Devices
Jonas Tegenfeldt
10:15 AM  11:00 AM
Room: Nordita Seminar Room 132:028
We use standard staining protocols and epifluorescence microscopy to gain information on the local AT/GC ratio along large DNA molecules stretched in nanoscale channels[1]. Our development opens up a novel route to mapping of largescale genomic variations as well as fast identification of rare or single cells. With rising temperature, dark patches appear along the DNA corresponding to ATrich regions that lose in intensity due to local melting of the doublestranded helix thereby resulting in a “barcode” pattern along the DNA (Figure 1) much like Gbanding but with significantly improved resolution, currently on the order of 110kbp. Compared to standard techniques, such as pairedend sequencing and array comparative genomic hybridization (CGH), our technology may offer a simpler and quicker way to identify structural variations such as deletions, translocations, insertions and copy number variations on scales ranging from 1kbp and up[2] on the singlemolecule level. Furthermore, the resulting "barcode" may be used for identification of organisms, such as difficulttogrow fungi, bacteria and viruses. REFERENCES [1] Tegenfeldt, J.O., et al., The dynamics of genomic‐length DNA molecules in 100‐nm channels. Proceedings of the National Academy of Sciences of the United States of America, 2004. 101(30): p. 10979‐10983. [2] Stankiewicz, P. and J.R. Lupski, Genome architecture, rearrangements and genomic disorders. Trends in Genetics, 2002. 18(2): p. 74‐82.
11:00 AM
Coffee break
Coffee break
11:00 AM  11:30 AM
11:30 AM
Polymer Escape from Metastable Kramers Potential: Path Integral Hyperdynamics Study

Timo Ikonen
Polymer Escape from Metastable Kramers Potential: Path Integral Hyperdynamics Study
Timo Ikonen
11:30 AM  12:15 PM
Room: Nordita Seminar Room 132:028
We study the dynamics of flexible, semiflexible and selfavoiding polymer chains under the Kramers metastable potential. Due to thermal noise the polymers, initially in the metastable well, can cross the potential barrier, but these events are rare at low temperatures. To speed up the slow rate processes in computer simulations we employ the hyperdynamics method using the pathintegral representation of the relevant Langevin dynamics [1]. In this study, we extend the method for manyparticle systems with internal degrees of freedom, such as the polymer chain. We study the regime where the well size is comparable to chain length. We find that the flexible, semiflexible and selfavoiding chains exhibit qualitatively different behavior. For the flexible chain, the crossing rate decreases monotonically with the polymer length (L), while for the semiflexbile chain the rate saturates at a level that depends on the chain stiffness. For the selfavoiding chain, on the other hand, the rate varies nonmonotonically with L. For L less than Lm, the rate decreases as L increases, while for L > Lm, the rate increases approximately linearly with L. We attribute this behavior to the coiltostretch transition of the chain, which lowers the effective free energy barrier and enhances the crossing rate. This effect can be instrumental in efficient separation of biopolymers. [1] L. Y. Chen and N. J. M Horing, J. Chem. Phys. 126, 224103 (2007).
12:15 PM
A Stochastic Model of Anomalous Heat Transport

Carlos MejiaMonasterio
A Stochastic Model of Anomalous Heat Transport
Carlos MejiaMonasterio
12:15 PM  1:00 PM
Room: Nordita Seminar Room 132:028
Abstract: We investigate the nonequilibrium dynamics of a chain of harmonic oscillators in contact with two stochastic Langevin heat baths at different temperatures and undergoing random collisions between neighbours that exchange their momenta with a constant rate $\gamma$. By means of an appropriate continuum limit, we solve the equations for the covariance matrix to leading order in the stationary state, and derive exact expressions for the temperature profile and for the leading contribution of the energy current, which scales as $1/\sqrt{\gamma N}$. At finite times, we solve adiabatically the equation describing the time evolution of the temperature profile $T(y,t)$, obtaining that in the bulk of the system, $T(y)$ evolves according to the energy continuity equation, but with a spacetime scaling that is described by a fractional diffusion equation.
1:00 PM
Lunch
Lunch
1:00 PM  2:30 PM
2:30 PM
Interacting Random Walkers in OneDimansional Systems

Tobias Ambjörnsson
Interacting Random Walkers in OneDimansional Systems
Tobias Ambjörnsson
2:30 PM  3:15 PM
Room: Nordita Seminar Room 132:028
The problem of a single random walker has received a lot of attention in the science community during the last century. There is now an increasing amount of interest in the problem of INTERACTING random walkers (due to the strong connection of this problem to the fields of, for instance, biophysics, nanofluidics, and cell biology). In particular, much attention has been on the behavior of the nonequilibrium problem of interacting walkers in (quasi)one dimensional systems, so called singlefile diffusion. The quantities of main interest in such a system is the mean square displacement (MSD) of a (fluorescently) tagged particle. It has been found previously (theoretically and experimentally) that the MSD for a tagged particle in a single file system scales as t^(1/2) for long times (in the thermodynamic limit), rather than t as for unconstrained diffusion. In the talk three new singlefile results will be presented: 1) The problem of hardcore interacting particles in a FINITE system (box) is solved analytically using a Betheansatz, see Ref [1]. Analysis of our exact solution reveals three time regimes, where the t^(1/2)behaviour appears as an intermediate regime. 2) We recently introduced a procedure, which we refer as to as Harmonization, which maps the diffusive motion of any type of 1d shortrange singlefile system onto that of chain of harmonically coupled beads; the effective spring constant in the system is connected to the details of the potential between particles. The Harmonization procedure reproduces all known longtime results in the singlefile field with some backofthe envelope calculations and allow us to analytically solve the longtime behavior of more complicated singlefile systems. For instance, the tagged particle motion in a harmonic potential, in a timevarying force field and correlation functions between particles are calculated. 3) Finally, singlefile diffusion in a system where the particles have different diffusion constants is considered. By combining the Harmonization procedure with effective medium theory we derive analytic results for the MSD, and find that for certain types of distributions for the diffusion constants, the dynamics becomes ultraslow; the MSD scales as t^delta, with delta<1/2, [3]. [1] L. Lizana and T. Ambjornsson, Singlefile diffusion in a box, Phys. Rev. Lett. 100, 200601 (2008); Phys. Rev. E 80, 051103 (2009). [2] T. Ambjornsson, L. Lizana, A. Taloni, E. Barkai and M.A. Lomholt, Foundation of fractional Langevin equations: Harmonization of a manybody problem, submitted, Eprint: arXiv:0909.0881. [3]. M. A. Lomholt, L. Lizana and T. Ambjornsson, in preparation.
3:15 PM
LipidProtein Membranes out of Equilibrium

Michael Lomholt
LipidProtein Membranes out of Equilibrium
Michael Lomholt
3:15 PM  4:00 PM
Room: Nordita Seminar Room 132:028
Biological membranes are typically not in thermal equilibrium. Experiments on lipidprotein model membranes have revealed that protein activity influences the mechanical properties of the membrane. A possible explanation for this alteration is given by a theoretical model in which the active proteins act as forcedipoles on the surrounding medium. A prediction of this model is that the protein activity will modify the tension of the membrane. However, one has to be careful when one looks at the consequences of this tension modification for the fluctuation spectrum of the membrane shape.
4:00 PM
Coffee break
Coffee break
4:00 PM  4:30 PM
4:30 PM
Aging Dynamics is Trivial in Logarithmic Time

Paolo Sibani
Aging Dynamics is Trivial in Logarithmic Time
Paolo Sibani
4:30 PM  5:15 PM
Room: Nordita Seminar Room 132:028
The dynamics of complex systems collectively known as glassy shares important phenomenological traits. I.e., a transition is generally observed from a timehomogeneous dynamical regime to an aging regime where physical changes occur intermittently and, on average, at a decreasing rate. It has been suggested that a global change of the independent time variable to its logarithm may render the aging dynamics homogeneous and thus trivialize it. In the talk this behavior is shown for experimental data from colloidal systems: the mean square displacement grows linearly in time at low densities but linearly in the logarithm of time at high densities. The intermittent nature of spatial fluctuations and the persistency of particle pairs is also discussed. A phenomenological oneparameter family of models is introduced which relies on the growth and collapse of strongly correlated clusters (“dynamic heterogeneities”). The full spectrum of colloidal behaviors is reperoduced by the model. In the limit where large clusters dominate the dynamics, intermittency induced by recordsize events occurs with rate ∝ 1/t, implying a homogeneous, logPoissonian process that qualitatively reproduces the experimental results. The crucial importance of recordsize fluctuations for colloidal dynamics is emphasized.
5:15 PM
Aggregation of Variables in Linear Systems

Martin N. Jacobi
Aggregation of Variables in Linear Systems
Martin N. Jacobi
5:15 PM  6:00 PM
Room: Nordita Seminar Room 132:028
I will present some new ideas for how to coarse grain linear dynamical systems through aggregation of variables. Both spectral methods and a very recent technique based on identification of ground states in a corresponding Potts glass model will be discussed. The methods are demonstrated by application to coarse graining of cellular automata and identification of the genetic code and higher level amino acid groups from DNA mutation statistics.
Thursday, March 18, 2010
9:00 AM
Lipids in Membranes Speak the Language of Curvature

Ole Mouritsen
Lipids in Membranes Speak the Language of Curvature
Ole Mouritsen
9:00 AM  9:45 AM
Room: Nordita Seminar Room 132:028
The physical properties of the lamellar lipidbilayer component of biological membranes is controlled by a host of thermodynamic forces leading to overall tensionless bilayers with a conspicuous lateral pressure profile and buildin curvaturestress instabilities that may be released locally or globally in terms morphological changes. In particular, the average molecular shape and the propensity of the different lipid and protein species for forming nonlamellar and curved structures are a source of structural transitions and control of biological function. I will discuss the effects of different lipids, sterols, and proteins on membrane structure and show how one can take advantage of the curvaturestress modulations brought about by specific molecular agents, such as fatty acids, lysolipids, and other amphiphilic solutes, to construct intelligent drugdelivery systems that function by enzymatic triggering of curvature.
9:45 AM
Complex Dynamics in Lipid Membranes

Olle Edholm
Erik Brandt
Complex Dynamics in Lipid Membranes
Olle Edholm
Erik Brandt
9:45 AM  10:30 AM
Room: Nordita Seminar Room 132:028
A biological lipid membrane may be viewed as a two dimensional (liquid crystal) fluid that is immersed in a three dimensional water solution. The system is further complicated by that the membrane is non flat, undergo time dependent undulations and have a thickness that fluctuates in time and space. This gives rise to complicated correlation functions in time and space. Experimentally some of these functions can be probed by inelastic scattering of neutrons or light and more recently by neutron spin echo experiments. Field dependent NMR relaxation experiments give also important information. We report here about molecular dynamics simulations that indicate that many of these correlation functions are stretched exponentials rather than ordinary exponentials and discuss different ways to interpret.
10:30 AM
Coffee break
Coffee break
10:30 AM  11:00 AM
11:00 AM
Diffusion Within Living Cells

Lene Oddershede
Diffusion Within Living Cells
Lene Oddershede
11:00 AM  11:45 AM
Room: Nordita Seminar Room 132:028
Using optical tweezers combined with image analysis we investigate motility of single proteins in membranes and of organelles inside living cellular organisms, one key issue being that the organisms are kept alive and healthy. Studies of two different biological systems will be presented: By specifically attaching a bead to a single protein, the lambdareceptor, which is a porin in the outer membrane of E. coli bacteria, we revealed its nanoscale diffusional motion and proposed a model that allows for extraction of the characteristic physical parameters including the diffusion constant. Surprisingly, the observed mobility is caused not only by thermal motion but in addition by an active motion associated with the metabolism of the organism. Connected to this, we show that antibiotics and antimicrobial peptides have a pronounced effect on single protein motility. The second biological system presented will be an S. pombe yeast cell, where the diffusion patterns of naturally occurring lipid granules have been uncovered using optical trapping and single particle tracking; the granules perform anomalous diffusion, with subdiffusion being most predominant at short timelags, and the biological functions giving motility footprints at longer timelags. The diffusional properties inside living yeast cells change during the cell cycle, and a novel maximal excursion method shows that the physical origin of the observed motility is probably fractional Brownian motion.
11:45 AM
Diffusion Controlled Reactions and Living Cell Biochemistry

Zoran Konkoli
Diffusion Controlled Reactions and Living Cell Biochemistry
Zoran Konkoli
11:45 AM  12:30 PM
Room: Nordita Seminar Room 132:028
The talk will discuss how Statistical Physics tools can be used to understand biochemistry of the living cell. Structures found in the living cell are rather special and to achieve such task techniques used in the field of Statistical Physics need to be slightly modified. A critical reflection is needed on which techniques to use and for what purpose. As an example the theory of diffusion controlled reactions will be reviewed with a purpose of using it for understanding spatiotemporal organization of the living cell. It will be argued that formalism of diffusion controlled reactions is a suitable framework for describing living cell and the scope and the limitations of such approach will be discussed. Informal discussion will be given around problems (and possible traps) one meets when trying to compute properties of biochemical reactions in the cell interior. For example, mean field calculations are routinely used to model cell biochemistry and there usage is rarely questioned. The validity of mean field equations will be critically reviewed. Some situations when these equations do not work will be mentioned (low dimension, fluctuation dominated kinetics). The generic features of spatio temporal organization of the living cell biochemistry will be discussed with particular emphasis on geometrical (spatial) features, ranging from shape of reactants towards spatial organization of intracellular reaction volumes. There is a great need for developing analysis tools that could help us understand intracellular organization and geometry and it will be argued that the theory of diffusion controlled reactions can be useful in the context. As an example, the brief overview of GeometryReaction InterPlay framework (GRIP) will be given. KEYWORDS: fluctuation dominated kinetics, diffusion controlled reactions, reactiongeometry interplay, reactions in restricted geometries, shape of reactants, shape of reaction volume, topology of pathway graph
12:30 PM
Lunch
Lunch
12:30 PM  2:00 PM
2:00 PM
Genetic Regulation in Time and Space

Mogens Høgh Jensen
Genetic Regulation in Time and Space
Mogens Høgh Jensen
2:00 PM  2:45 PM
Room: Nordita Seminar Room 132:028
Genetic circuits have been studied quite intensively in recent years. In particular, we have focussed on oscillatory patterns related to negative feedback loops inside single cells in eucaryotic systems [1,2]. In many cases, however, it is of interest to study how cells communicate with each other when cells are arranged in certain spatial structures, like biofilms and tissues. We have attacked this problem by means of a repressorlattice where single repressilators (closed feedback loops) are placed on a hexagonal lattice [3]. Such systems can be build without any internal frustration and can in most cases exhibit stable, oscillating states. Commensurability effects however play a role and may lead to internal frustration causing breaking of symmetries and solutions of many different phases. Eventually, also chaotic solutions may be present [3]. We discuss both situations of directed and bidirected interactions on the repressorlattice. [1] S. Pigolotti, S. Krishna and M.H. Jensen, "Oscillation patterns in negative feedback loops", Proc.Nat.Acad.Sci. 104, 65336537 (2007). [2] S. Pigolotti, S. Krishna and M.H. Jensen, "Symbolic dynamics of biological feedback networks", Phys. Rev. Lett. 102, 088710 (2009). [3] M.H. Jensen, S. Krishna and S. Pigolotti, "The RepressorLattice: Feedback, Commensurability, and Dynamical Frustration, Phys. Rev. Lett. 103, 118101 (2009).
3:15 PM
ALBANOVA COLLOQUIUM  Stochastic Thermodynamics: Theory and Experiments

Udo Seifert
ALBANOVA COLLOQUIUM  Stochastic Thermodynamics: Theory and Experiments
Udo Seifert
3:15 PM  4:15 PM
Room: Oskar Klein Lecture Hall (main building, 4th floor)
Stochastic thermodynamics provides a framework for describing small systems embedded in a heat bath and externally driven to nonequilibrium. Examples are colloidal particles in timedependent optical traps, single biomolecules manipulated by optical tweezers or AFM tips, and motor proteins driven by ATP excess. A firstlaw like energy balance allows to identify applied work and dissipated heat on the level of a single stochastic trajectory. Total entropy production includes not only this heat but also changes in entropy associated with the state of the small system. Within such a framework, exact results like an integral fluctuation theorem for total entropy production valid for any initial state, any timedependent driving and any length of trajectories can be proven [1]. These theoretical predictions have been illustrated and tested with experiments on a colloidal particle pushed by a periodically modulated laser towards a surface [2]. Key elements of this framework like a stochastic entropy can also be applied to athermal systems as experiments on an optically driven defect center in diamond show [3,4]. For mechanically driven nonequilibrium steady states, the violation of the fluctuationdissipation theorem can be quantified as an additive term directly related to broken detailed balance (rather than a multiplicative effective temperature) [5,6]. Integrated over time, a generalized Einstein relation appears which we have recently verified experimentally [7]. Finally, optimal protocols are derived which (i) minimize the work required to switch from one equilibrium state to another in finite time [8] and (ii) maximize the power of stochastic heat engines operating between two heat baths [9]. [1] U. Seifert, Phys. Rev. Lett. 95: 040602/14, 2005. [2] V. Blickle, T. Speck, L. Helden, U. Seifert, and C. Bechinger, Phys. Rev. Lett. 96: 070603/14, 2006. [3] S. Schuler, T. Speck, C. Tietz, J. Wrachtrup, and U. Seifert, Phys. Rev. Lett. 94: 180602/14, 2005. [4] C. Tietz, S. Schuler, T. Speck, U. Seifert, and J. Wrachtrup, Phys. Rev. Lett. 97: 050602/14, 2006. [5] T. Speck and U. Seifert, Europhys. Lett. 74: 391396, 2006. [6] U. Seifert and T. Speck, EPL, in press, 2010. [7] V. Blickle, T. Speck, C. Lutz, U. Seifert, and C. Bechinger. Phys. Rev. Lett., 210601/14, 2007. [8] T. Schmiedl and U. Seifert, Phys. Rev. Lett, 98: 108301/14, 2007. [9] T. Schmiedl and U. Seifert, EPL 81, 20003, 2008.
4:30 PM
Coffee break
Coffee break
4:30 PM  5:00 PM
5:00 PM
Fluctuation Theorems and Single Molecule Experiments

Alessandro Mossa
Fluctuation Theorems and Single Molecule Experiments
Alessandro Mossa
5:00 PM  5:45 PM
Room: Nordita Seminar Room 132:028
The manipulation of individual macromolecules made possible by experimental techniques such as optical tweezers or atomic force microscopy gives a unique insight into the nonequilibrum thermodynamics of small systems. Besides a general introduction about the theoretical and experimental framework, this talk is focused on two topics: the proper way of measuring the work applied to the system in a singlemolecule experiment, and a powerful generalization of Crooks fluctuation theorem that allows the exploration of misfolded and metastable states.
6:00 PM
Reception
Reception
6:00 PM  10:00 PM
Room: Nordita Main Building
Friday, March 19, 2010
9:00 AM
Equilibrium and NonEquilibrium Physics of Nucleosome Positioning

Ulrich Gerland
Equilibrium and NonEquilibrium Physics of Nucleosome Positioning
Ulrich Gerland
9:00 AM  9:45 AM
Room: Nordita Seminar Room 132:028
9:45 AM
Dynamics, Clustering and Collisions of Inertial Particles in Mixing Flows

Bernhard Mehlig
Dynamics, Clustering and Collisions of Inertial Particles in Mixing Flows
Bernhard Mehlig
9:45 AM  10:30 AM
Room: Nordita Seminar Room 132:028
We study the dynamics of small particles suspended in mixing ﬂows (e.g. microscopic water droplets in turbulent rain clouds). We describe how the particles move, cluster together, and collide. Our results enable us, for example, to address the question of how long it takes to rain from a vigorously turbulent rain cloud. The talk is based on the manuscripts appended below. Mehlig & Wilkinson, Phys. Rev. Lett. 92 (2004) 250602 Duncan, Mehlig, Ostlund & Wilkinson, Phys. Rev. Lett. 95 (2005) 165503 Arvedson, Mehlig, Wilkinson & Nakamura, Phys. Rev. Lett. 96 (2006) 030601 Wilkinson, Mehlig & Bezuglyy, Phys. Rev. Lett. 97 (2006) 048501 Gustavsson, Mehlig, Wilkinson & Uski, Phys. Rev. Lett. 101 (2008) 174503
10:30 AM
Coffee break
Coffee break
10:30 AM  11:00 AM
11:00 AM
Mesoscopic NonEquilibrium Thermodynamics

Dick Bedeaux
Mesoscopic NonEquilibrium Thermodynamics
Dick Bedeaux
11:00 AM  11:45 AM
Room: Nordita Seminar Room 132:028
Classical thermodynamics is a theory for a collection of molecules in equilibrium. What happens if the number of molecules in the system becomes smaller and smaller, and the system boundaries reflect conditions further and further away from equilibrium? Can we still use thermodynamics? This lecture aims to explain that the field of nonequilibrium thermodynamics can be extended to describe in a systematic manner even molecular behaviour far from equilibrium conditions. We start introducing the concept of internal variables, derive the law of mass action, and end illustrating the theory by applications to RNA stretching experiments and active transport by the CaATPase. We discuss that a thermodynamic theory is needed, also for molecules. References S. Kjelstrup, D. Bedeaux, Isabella Inzoli, JeanMarc Simon, Criteria for validity of thermodynamic equations from nonequilibrium molecular dynamics simulations, Energy, 33 (2008) 11851196 J.M. Rubi, D. Bedeaux and S. Kjelstrup, Thermodynamics for small molecule stretching experiments, J. Phys. Chem. B, 110 (2006) 1273312737 D. Bedeaux and S. Kjelstrup, The measurable heat flux that accompanies active transport by the Ca ATPase. Phys. Chem. Chem. Phys. 48 (2008) 73047317.
11:45 AM
Metabolic Networks, Information, Null Model, and Evolution

Petter Minnhagen
Metabolic Networks, Information, Null Model, and Evolution
Petter Minnhagen
11:45 AM  12:30 PM
Room: Nordita Seminar Room 132:028
The metabolism in an organism is reduced to a network of substances. The resulting degreedistribution is power law like with an exponent about 2.2. In order to understand this, we use information theory to obtain a nullmodel defined by assigning equal probabilities to what is assumed to be the fundamental network possibilities. A stochastic variant of variational calculus is used to obtain the corresponding degree distribution for the nullmodel. The striking agreement implies that the null model catches the overall feature of the metabolic network. Using the network structure measures like clustering and assortativity, a small difference is identified as the only sign of any possible evolutionary pressure. However, this difference is only manifested in a slight difference in the degree distribution and seemingly not in any particular network design.
12:30 PM
Lunch
Lunch
12:30 PM  2:00 PM
2:00 PM
The Inverse Ising Problem: A Survey and some Empirical Results

Erik Aurell
(
KTH
)
The Inverse Ising Problem: A Survey and some Empirical Results
Erik Aurell
(
KTH
)
2:00 PM  2:45 PM
Room: Nordita Seminar Room 132:028
The "Inverse Ising Problem" refers to finding the parameters (the J_ij's and the h_i's) in an Ising model given the first and second moments (the magnitizations m_i and the correclation functions c_ij). This is of great interest in machine learning and data analysis whenever the data set and the number of variables is large, but the values taken by the variables can be taken to be "high" and "low". The maximum entropy distributions with given first and second moments then has the Ising form where the h_i's and J_ij's are Lagrange parameters. The last years have seen an explosion in interest in approximate but fast methods borrowed from statistical mechanics to learn such "maxentropy" models from correlation data. Some motivations have been e.g. inferring casual structures underlying observed gene expression, or inferring functional connectivities between neurons from multineuronal recordings, where measurements from hundreds of neurons are available today, and millions have been envisaged. Although methods borrowed from nonequilibrium may be more promising in applications, I will describe results using equilibrium statistical mechanics, and the testing ground will be mainly the SherringtonKirkpatrick spin glass. The methods discussed are simple meanfield, TAP, and the "Susceptibility Propagation" introduced by Mezard. One main message is that all these are sensitive to the accuracy of the correlation data themselves. There is hence a threeway tradeoff between computability, inference accuracy (given perfect data), and sensitivity to undersampling of the correlations. This is work done or in progess with John Hertz, Yasser Roudi, Mikko Alava, Hamed Mahmoudi, Aymeric Fouquier d'Herouel, Jarkko Salojärvi, Zeng HongLi and Charles Ollion. Similar results to ours on Susceptibility Propagation have been obtained by Enzo Marinari (paper available on arXiv.org).
2:45 PM
The Inverse Ising Model: Why and How

John Hertz
(
Nordita
)
The Inverse Ising Model: Why and How
John Hertz
(
Nordita
)
2:45 PM  3:30 PM
Room: Nordita Seminar Room 132:028
Ising models form a natural framework for modeling the distribution of multineuron spike patterns: Of all models that correctly describe the firing rates and pairwise firing correlations, the Ising model is the one of maximum entropy. The problem at hand here is an inverse one to that we usually encounter. Normally, one has a model with given couplings (Jij) and the task is to compute averages and correlation functions of the variables of the model. Here we are given the averages and correlations and the task is to find the couplings. In the simplest approach to this problem, one considers only the measured firing rates and equaltime pairwise firing correlations and tries to find the Ising model that has these statistics. In our work we have explored and compared a number of methods for doing this, using data from a realistic model network of spiking neurons. Several of these methods work remarkably well. This success is tempered, however, by our second set of findings. Using an informationtheoretic measure of the overall quality of fit, we find that, while the Ising model is a good description of the distribution of spike patterns for small populations of neurons (~ 10), it does worse and worse for larger and larger populations (for reasons that are not yet understood). Finally, I will describe some recent work, which extends the Ising approach to describe nonequaltime firing correlations.
3:30 PM
Discussion and Closing
Discussion and Closing
3:30 PM  5:00 PM
Room: Nordita Seminar Room 132:028