solving the shortest path problem by physarum solver - modeling of the adaptive network of true...

43
Physarum polycephalum 1cm

Post on 21-Dec-2015

220 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Physarum polycephalum

1cm

Page 2: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Physarum polycephalum

Page 3: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Solving maze

Page 4: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Discrete Form

2N

jN

iN

1N

ijM

Page 5: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Discrete Form

1ijM

iN jN

2ijM

Page 6: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Discrete Form

2N

jN

iN

1N

ijM

Start point

Goal

1N

2N

iN )(tip Pressure

ijM ijL

)(tijD

)(tijQ Flux

Length

Conductivity

Page 7: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Modeling of the flux of the sol

)( jiij

ijij pp

L

DQ

iN )(tip Pressure

ijM ijL

)(tijD

)(tijQ Flux

Length

Conductivity

)(tip Electric Pressure

)( tij

ij

D

LElectric Resistance

)(tijQ Electric Current

Page 8: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Kirchhoff law

jN

2,1j0 iij Q

Page 9: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

1N

0I

FS2N

0IFS

Electric sources

0

0

0 2

0 1

I Q

I Q

ii

ii

Page 10: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Modeling of the tube growth

ijij rDijQD

dt

d || r :degenerate rate

(constant)

time

rijQ ||

ijD )(tijD

Page 11: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Model equations

Flux of sol Tube growth

ijij rDijQD

dt

d ||

2,1j

0

0

0 2

0 1

I Q

I Q

ii

ii

0 iij Q)( ji

ij

ijij pp

L

DQ

Page 12: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

T shape vessel

The tube at dead end disappears

Page 13: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Ring shape vessel

Only shortest tube remains

)( 212

112 LL

Stable equilibrium pointUnstable equilibrium point

Page 14: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Ring shape vessel

Both tubes remain

)( 212

112 LL

Stable equilibrium pointUnstable equilibrium point

Page 15: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Solving Maze

Page 16: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Solving Maze

スタート

ゴール

Page 17: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Solving Maze

スタート

ゴール

Page 18: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Apply for road navigation system

START

GOAL

Page 19: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

1. We can reproduce the adaptive network of the true slime mold.

2. “Physarum Solver” can find the shortest path.

Summary

1. Simulation with many food sources

Current Work

Page 20: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

That’s all. Thank you very much!!

Page 21: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Apply for road navigation system

START

GOAL

Page 22: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Ex 2. Shortest path on Weighted graph

food

food

Page 23: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Ex 2. Shortest path on Weighted graph

nyflexibilit

region

::

nyflexibilit

region

::

Page 24: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Ex 2. Shortest path on Weighted graph

ijij DrijQD

dt

d)(|| x

1

3)(xr

Page 25: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Solving Maze

START

GOAL

Does physarum have enough intelligence to solve the maze?

Page 26: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Shortest Path From Seattle to Houston

Page 27: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Ex. 3 Path choice

food

vessel

Physarum

Page 28: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Model Equation

ijijij rDQDdt

d

Model of Tube Growth

One(Short)

Both

180180

1

1

10

Page 29: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

)1(

Both tubes remain

Stable equilibrium pointUnstable equilibrium point

Simulation Result

Page 30: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

)1(

Almost all tube remain

数値シミュレーション

Page 31: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

)1( Simulation Result

Page 32: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

One tube remains (chose by initial condition )

Simulation Result )1(

Stable equilibrium pointUnstable equilibrium point

Page 33: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

One tube remains (chose by initial condition )

数値シミュレーション )1(

Page 34: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Simulation Result

One(initial condition)

One(Short)

Both

Both

180180

1

1

10

ijijij rDQDdt

d

Model of Tube Growth

Page 35: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Model Equation

Q

jiD

1

1

ijij rDQ

QD

dt

d

1||

||

01||

||

ijrDQ

Q

)3(

Page 36: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Simulation Result with each 0I

Flux

Number of the tube

Page 37: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Both tubes remain

)0.3(1||

||0

IrDQ

QD

dt

dij

ij

ijij

Simulation Result

Page 38: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Application

Page 39: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Application

Page 40: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Thank you for your attention!!

Page 41: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

a2 Q

Modeling of the flux of the sol

FluxPoiseuille Flow

ij

jiijij L

ppaQ

)(

8

4

8

4ij

ij

aD )( ji

ij

ijij pp

L

DQ

Page 42: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

Ex 2. Shortest path on Weighted graph

ijjiij

ij LrppL

D)(x

partlower

partupperr

3/11

3/23)(x

ijji

ij

ij LppL

Dr ˆˆ)(

x

ijijij DrQDdt

d)(x

Page 43: Solving the shortest path problem by Physarum Solver - Modeling of the Adaptive Network of True Slime Mold Hokkaido Univ. Atsushi Tero Hiroshima Univ

)0.2(1||

||0

IrDQ

QD

dt

dij

ij

ijij

One tube remains

Simulation Result