Visualisation of HUGE state spaces

1
Visualisation of HUGE state spaces

• Jan Friso Groote
• Frank van Ham

2
Models of behaviour (mCRL)
up
down
display
34
C(nNat)up.C(n1)
down.(C(n-1)?ngt0?error.d)
display(n).C(n)
3

Models of behaviour II
C(nNat)up.C(n1)
down.(C(n-1)?ngt0?error.C(0))
display(n).C(n)
ManSm read(m). Sk
read(k).inform(mk).Man System C(0) Man
C(0)
4
(No Transcript)
5
Visualize the structure of a state space
Visualiser.
6
Cluster nodes
Rank nodes
Start node
Back pointer
Phd of Frank van Ham
7
Draw tree as 3D object
8
Display with nodes and edges visible
9
Display with back pointers
10
Use color to highlight properties
Distance to the root
Probability to visit a node
11
0
1
2
k
n
Color on the values of variables
12
Color based on inform(n), n 0,1,2,3,4
13
Reduced behaviour modulo branching bisimulation
14
A larger example a modular hef system
start
start
stop
stop
start
start
stop
stop
Relay
15
The hef system with 2 legs
16
Lift system with three legs
17

Zoom into the deadlock
18
Hef system with 6 legs (500.000 states)
19
Hef system with 1.000.000 states
20
VLTS Benchmark site CWI/SEN2, INRIA/VASY
21
(No Transcript)
22
(No Transcript)
23
(No Transcript)
24
Research questions

• How to visualize parallel processes
• Relate full state space with the reduced
  statespace modula, say, branching bisimulation
• Introduce probabilities in the model to visualize
  the visiting probabilities of states more
  reliably
• Experience and use

25
Contact and downloads
J.F.Groote_at_tue.nl
FSM viewer www.win.tue.nl/fvham/fsm
(for Windows, Linux)
mCRL toolset www.cwi.nl/mcrl
(for Linux, Unix, MacOsX, Windows with Cygwin)
VLTS benchmark suite www.inrialpes.fr/vasy/cadp/re
sources/benchmark_bcg.html