alarsyo/spot
1
0
Fork 0
Code Issues Pull requests Projects Releases Wiki Activity

* src/tgbatest/randtgba.cc: Remplace the -O option by -A, reading

Browse source 
all algorithms from a file.  Use the emptiness_check_instantiator
syntax as name in the output.
* bench/emptchk/defs.in: DEfine ALGORITHMS here.
* bench/emptchk/ltl-human.sh, bench/emptchk/ltl-random.sh,
bench/emptchk/pml-clserv.sh, bench/emptchk/pml-clserv.sh: Use
$ALGORITHMS.
* src/misc/timer.cc: Truncate long keys in display.
This commit is contained in:
Alexandre Duret-Lutz 2005-02-18 12:28:42 +00:00
parent 3b3a196526
commit 6314b682ba
9 changed files with 144 additions and 105 deletions
Download patch file Download diff file Expand all files Collapse all files

64
bench/emptchk/README
View file

@ -175,51 +175,42 @@ This directory contains:
INTERPRETING THE RESULTS
==========================
Here are the short names for the algorithms used in the outputs.
ltl-*.sh tests use names from the left column, and pml-*.sh tests
use names from the right column.
Here are the short names for the algorithms presented in the outputs.
Cou99 Cou99
Cou99_shy- Cou99(shy !group)
Cou99_shy Cou99(shy group)
> Cou99_rem Cou99(poprem)
> Cou99_rem_shy- Cou99(poprem shy !group)
> Cou99_rem_shy Cou99(poprem shy group)
> CVWY90 CVWY90
CVWY90_bsh CVWY90(bsh=4K)
> GV04 GV04
> SE05 SE05
SE05_bsh SE05(bsh=4K)
> Tau03 Tau03
> Tau03_opt Tau03_opt
Cou99
Cou99(shy !group)
Cou99(shy group)
> Cou99(poprem) # called `Cou99' in the paper
> Cou99(poprem shy !group) # called `Cou99 Shy-' in the paper
> Cou99(poprem shy group) # called `Cou99 Shy' in the paper
> CVWY90
> GV04
> SE05
> Tau03
> Tau03_opt
Only the algorithms marked with a `>' have been shown in the paper.
`bsh' stands for `bit-state hashing'.
`Cou99_rem*' algorithms are using the `rem' field to remove
the SCC without recomputing the SCC as described in the paper.
The other `Cou99*' algorithms are not. (Beware that in the paper
we presented the `Cou99_rem*' variants and called them `Cou99*'.)
`Cou99(poprem*)' algorithms are using the `rem' field to remove the
SCC without recomputing the SCC as described in the paper. The
other `Cou99' algorithms are not. (Beware that in the paper we
presented the `Cou99(poprem*)' variants and called them `Cou99*'.)
The ltl-*.sh tests output look as follows:
| density: 0.001
| Emptiness check ratios
| CVWY90 5.5 4.4 6.3 25
| CVWY90_bsh 5.7 4.8 6.3 25
| Cou99 5.5 3.3 4.3 25
| Cou99_rem 5.5 3.0 4.3 25
| Cou99 18.9 9.6 10.4 29
| Cou99(shy !group) 16.7 16.3 25.7 29
| ...
(A) (B) (C) (D)
(A) (B) (C) (D)
|
| Accepting run ratios
| CVWY90 5.5 2.6
| Cou99 2.0 2.6
| Cou99_rem 2.0 2.1
| Cou99_rem_shy 1.2 2.1
| Cou99 8.6 13.5
| Cou99(shy !group) 7.3 12.2
| ...
(E) (F)
(E) (F)
(A) mean number of distinct states visited
expressed as a % of the number of state of the product space
@ -237,10 +228,10 @@ This directory contains:
The pml-*.sh tests output look as follows:
| Cou99 , 783, 2371, 5, 783, 4742, 237, no accepting run found
| Cou99_shy- , 783, 2371, 5, 783, 4742, 537, no accepting run found
| Cou99 , 92681, 391160, 1, 92681, 391160, 46471, no accepting run found
| Cou99(shy !group) , 92681, 391160, 1, 92681, 391160, 47148, no accepting run found
| ...
(G) (H) (I) (K) (L) (M) (N)
(G) (H) (I) (K) (L) (M) (N)
(G) Number of states in the product.
(H) Number of transitions in the product.
@ -260,5 +251,10 @@ This directory contains:
shown above. Try removing the `-1' option from the script, or toying
with randtgba itself.
CVWY90 and SE05 have bit-state hashing implementations. Edit the
file `algorithms' and add lines like `CVWY90(bsh=5M)' or
`SE05(bsh=512K)' to try these.
(The `bsh=' argument gives the hash table size in bytes.)
Besides randtgba, two other tools that you might find handy we
experimenting are src/ltltest/randltl and src/tgbatest/ltl2tgba.