Jammology Physics of self-driven particles Toward solution of all jams Katsuhiro Nishinari Faculty of Engineering, University of Tokyo Outline Introduction of “Jammology” Self-driven particles, methodology Simple traffic model for ants and molecular motors Conclusions Jams Everywhere School, Herd, Flock, etc. What are self-driven particles (SDP)? Vehicles, ants, pedestrians, molecular motors… Non-Newtonian particles, which do not satisfy three laws of motion. ex. 1) Action Reaction, “force” is psychological 2) Sudden change of motion D. Helbing, Rev. Mod. Phys. vol.73 (2001) p.1067. D. Chowdhury, L. Santen and A. Schadschneider,Phys. Rep. vol.329 (2000) p.199. Jammology=Collective dynamics of SDP Text book of “Jammology” Conventional mechanics, or statistical physics cannot be directly applicable. Rule-based approach (e.g., CA model) Numerical computations Exactly solvable models (ASEP,ZRP) Subjects of Jammology Vehicles car, bus, bicycle, airplane,etc. Humans Swarm, animals, ant, bee, cockroach, fly, bird,fish,etc. Internet packet transportation Jams in human body Blood, Kinesin, ribosome, etc. Infectious disease, forest fire, money, etc. Conventional theory of Jam = Queuing theory Out In Service Breakdown of balance of in and out causes Jam. What is NOT considered in Queuing theory Exclusion effect of finite volume of SDP ASEP model can deal the exclusion! ASEP=A toy model for jam ASEP(Asymmetric Simple Exclusion Process) Rule:move forward if the front is empty t 0 1 0 1 1 0 0 1 0 1 1 1 0 0 0 t 1 0 0 1 1 0 1 0 0 1 1 1 0 1 0 0 This is an exactly solvable model, i.e., we can calculate density distribution, flux, etc in the stationary state. Who considered ASEP? Macdonald & Gibbs, Biopolymers, vol.6 (1968) p.1. Protein composition process of Ribosomes on mRNA pr This research has not been recognized until recently. Fundamental diagram of ASEP with periodic boundary condition In the stationary state of TASEP, flow J - density relation is J 1 v 1 1 4 q (1 ) 2 • Flow-density・ ・ ・ ・ ・Particle-hole symmetry • Velocity-density・ ・ ・ ・ ・monotonic decrease M.Kanai, K.Nishinari and T.Tokihiro, J. Phys. A: Math. Gen., vol.39 (2006) pp.9071-9079.. Ultradiscrete method reveals the relation between different traffic models! J.Phys.A, vol.31 (1998) p.5439 ut u xx 2uu x Ultradiscrete method ASEP(Rule 184) Burgers equation CA model Macroscopic model Euler-Lagrange transformation Phys.Rev.Lett., vol.90 (2003) p.088701 OV model Car-following model Toward solution of all kind of jams! Vehicular traffic cars, bus, trains,… Pedestrians Jams in our body Traffic in ant-trail Ants drop a chemical (generically called pheromone) as they crawl forward. Other sniffing ants pick up the smell of the pheromone and follow the trail. with periodic boundary conditions Ant trail traffic models and experiments Experiments and theory 1) M. Burd, D. Archer, N. Aranwela and D.J. Stradling, American Natur. (2002) 2) I.D.Couzin and N.R.Franks, Proc.R.Soc.Lond.B (2002) 3) A. Dussutour, V. Fourcassie, D.Helbing and J.L. Deneubourg, Nature (2004) Differential equations E.M.Rauch, M.M.Millonas and D.R.Chialvo, Phys.Lett.A (1995) Ant Langevin type equation pheromonal field ( x, t ) d 2 x(t ) dx(t ) U ( ( x)) (t ) 2 dt dt ( x, t ) D 2 ( x, t ) f ( x, t ) g ( x) t f : evaporation rate (x ): ant density at x Ant trail CA model One lane, uni-directional flow Dynamics: 1. Ants movement 2. Update Pheromone(creation & diffusion) q Parameters: q < Q, f q f Q f f D. Chowdhury, V. Guttal, K. Nishinari and A. Schadschneider, J.Phys.A:Math.Gen., Vol. 35 (2002) pp.L573-L577. Bus Route Model Bus operation system=In fact the ant CA! The dynamics is the same as the ant model Q Q q Loose cluster formation = buses bunching up together f f f f Modeling of pedestrians Basic features of collective behaviours of pedestrians 1) Arch formation at exit 2) Oscillation of flow at bottleneck 3) Lane formation of counterflow at corridor Models for evacuation Social force model (Continuous model) D.Helbing, I.Farkas and T.Vicsek, Nature (2000). Floor field model (CA model) C.Burstedde, K.Klauck, A.Schadschneider,J.Zittartz, Physica A (2001) Floor field CA Model C.Burstedde, K.Klauck, A.Schadschneider,J.Zittartz, Physica A, vol.295 (2001) p.507. Pedestrians in evacuation = herding behavior = long range interaction For computational efficiency, can we describe the behavior of pedestrians by using local interactions only? Idea: Footprints = Feromone Long range interaction is imitated by local interaction through „memory on a floor“. Details of FF model Floor is devided into cells (a cell=40*40 cm2) Exclusion principle in each cell Parallel update A person moves to one of nearest cells with the probability pij defined by „floor field(FF)“. Two kinds of FF is introduced in each cell: 1) Dynamic FF・・・footprints of persons 2) Static FF・・・Distance to an exit Dymanic FF (DFF) Number of footprints on each cell Leave a footprint at each cell whenever a person leave the cell Dynamics of DFF dissipation+diffusion dissipation・・・ diffusion・・・ 4 1 Herding behaviour = 1 choose the cell that has more footprints Store global information to local cells 2 Static FF (SFF) = Dijkstra metric Distance to the destination is recorded at each cell 100 80 60 40 20 0 0 20 40 60 80 One exit with a obstacle 100 Two exits with four obstacles This is done by Visibility Graph and Dijkstra method. K. Nishinari, A. Kirchner, A. Namazi and A. Schadschneider, IEICE Trans. Inf. Syst., Vol.E87-D (2004) p.726. Application of SFF Problem of “Zone partition” Which door is the nearest? By using SFF The ratio of an area to the total area determines the number of people who use the door in escaping from this room. Probability of movement pij exp( k D Dij ) exp( k s Sij ) exp( k I I ij ) S ij Dij I ij Distance between the cell (i,j) and a door. Number of footprints at the cell (i,j). Set I=1 if (i,j) is the previous direction of motion. Update procedure Initial: Calculate SFF 1. 2. 3. 4. 5. Update DFF (dissipation & diffusion) Calculate pij and determine the target cell Resolution of conflict Movement Add DFF +1 Resolution of conflict Parameter 1 [0,1] All of them cannot move. One of them can move. Meanings of parameters in the model kS kD pij exp( k D Dij ) exp( k s Sij ) kS large: Normal (kS small: Random walk) kD large: Panic kD / kS ・・・Panic degree (panic parameter) large: competition small: coorporation Simulations using inertia effect There is a minimum in the evacuation time when the effect of inertia is introduced. SFF is strongly disturbed. People become less flexible to form arches. (Do not care others!) People become flexible to avoid congestion. Simulation Example: Evacuation at Osaka-Sankei Hall Jams near exits. Hamburg airport in Germany Influence of Obstacle Placed an obstacle near exit. k S 10, k D 0, 0.3 2offset None Center 1offset If an obstacle is placed asymmetrically, total evacuation time is reduced! Intensive competition. D.Helbing, I.Farkas and T.Vicsek, Nature, vol.407 (2000) p.487. A.Kirchner, K.Nishinari, and A.Schadschneider,Phys. Rev. E, vol.67 (2003) p.056122. Conclusions • Traffic Jams everywhere = Jammology is interdisciplinary research among Math. , Physics and Engineering! Examples • Ant trail CA model is proposed by extending ASEP. The model is well analyzed by ZRP. • Non-monotonic variation of the average speed of the ants is confirmed by robots experiment. • Traffic jam in our body is related to diseases. • Modelling molecular motors Conclusions Ant trail CA model is proposed by extending ASEP. The model is well analyzed by ZRP. Non-monotonic variation of the average speed of the ants is confirmed by robots experiment. FF model is a local CA model with memory, which can emulating grobal behavior. FF model is quite efficient tool for simulating pedestrian behavior. Jammology = Math. , Physics and Engineering