We used to have a script to run all SLICOT benchmarks. This shall now be more automatic by calling such models from ReachabilityModels.jl.
In this issue we can track the progress for each model.
Tables:
- First table: runtime (in seconds) for different algorithms in
ReachabilityAnalysis.jl, and the reference runtime in HSCC'18 article (TAC).
- Second table: width of the final reach-set for the variables computed.
Motor
Settings: step-size: 1e-3, dim: 8, property: x5 < 0.45 || x5 > 0.6.
Runtime:
| Vars |
LGG09 |
HSCC18 |
| x5, -x5 |
|
0.5 |
| box |
|
3.85 |
Accuracy:
| Vars |
LGG09 |
HSCC18 |
| x5, -x5 |
|
- |
Building
Settings: step-size: 1e-3, dim: 48, property: x25 < 6e-3.
Runtime:
| Vars |
LGG09 |
HSCC18 |
| x25 |
|
1.71 |
| box |
|
66.5 |
Accuracy:
| Vars |
LGG09 |
HSCC18 |
| x25, -x25 |
|
|
| box |
|
|
PDE
Settings: step-size: 1e-3, dim: 84, property: y1 < 12.
Runtime:
| Vars |
LGG09 |
HSCC18 |
| y1 |
|
3.56 |
| box |
|
197 |
Accuracy:
| Vars |
LGG09 |
HSCC18 |
| y1, -y1 |
|
|
| box |
|
|
Heat
Settings: step-size: 1e-3, dim: 200, property: x133 < 0.1.
Runtime:
| Vars |
LGG09 |
HSCC18 |
| x133 |
0.47 |
8.28 |
| box |
98 |
1050 |
Accuracy:
| Vars |
LGG09 |
HSCC18 |
| x133, -x133 |
[-0.056793, 0.056862] |
|
ISS
Settings: step-size: 1e-3, dim: 270, property: -7e-4 <= y3 <= 7e-4.
Runtime:
| Vars |
LGG09 |
HSCC18 |
| y3 |
|
0.95 |
| box |
|
126 |
Accuracy:
| Vars |
LGG09 |
HSCC18 |
| y3, -y3 |
|
|
| box |
|
|
Beam
Settings: step-size: 1e-3, dim: 348, property: x89 < 2100.
Runtime:
| Vars |
LGG09 |
HSCC18 |
| x89 |
|
26.7 |
| box |
|
2960 |
Accuracy:
| Vars |
LGG09 |
HSCC18 |
| x89, -x89 |
|
|
| box |
|
|
MNA1
Settings: step-size: 1e-3, dim: 578, property: x1 < 0.5.
Runtime:
| Vars |
LGG09 |
HSCC18 |
| x1 |
|
90 |
| box |
|
580 |
Accuracy:
| Vars |
LGG09 |
HSCC18 |
| x1. -x1 |
|
|
| box |
|
|
FOM
Settings: step-size: 1e-3, dim: 1006, property: y1 < 185.
Runtime:
| Vars |
LGG09 |
HSCC18 |
| y1 |
|
3.67 |
| box |
|
1060 |
Accuracy:
| Vars |
LGG09 |
HSCC18 |
| y1, -y1 |
|
|
| box |
|
|
MNA5
Settings: step-size: 1e-3, dim: 10913, property: x1 < 0.2 & x2 < 0.15.
Runtime:
| Vars |
LGG09 |
HSCC18 |
| x1, x2 |
|
1220 |
| box |
|
TO |
Accuracy:
| Vars |
LGG09 |
HSCC18 |
| x1, -x1, x2, -x2 |
|
|
| box |
|
|
We used to have a script to run all SLICOT benchmarks. This shall now be more automatic by calling such models from
ReachabilityModels.jl.In this issue we can track the progress for each model.
Tables:
ReachabilityAnalysis.jl, and the reference runtime in HSCC'18 article (TAC).Motor
Settings: step-size: 1e-3, dim: 8, property:
x5 < 0.45 || x5 > 0.6.Runtime:
Accuracy:
Building
Settings: step-size: 1e-3, dim: 48, property:
x25 < 6e-3.Runtime:
Accuracy:
PDE
Settings: step-size: 1e-3, dim: 84, property:
y1 < 12.Runtime:
Accuracy:
Heat
Settings: step-size: 1e-3, dim: 200, property:
x133 < 0.1.Runtime:
Accuracy:
ISS
Settings: step-size: 1e-3, dim: 270, property:
-7e-4 <= y3 <= 7e-4.Runtime:
Accuracy:
Beam
Settings: step-size: 1e-3, dim: 348, property:
x89 < 2100.Runtime:
Accuracy:
MNA1
Settings: step-size: 1e-3, dim: 578, property:
x1 < 0.5.Runtime:
Accuracy:
FOM
Settings: step-size: 1e-3, dim: 1006, property:
y1 < 185.Runtime:
Accuracy:
MNA5
Settings: step-size: 1e-3, dim: 10913, property:
x1 < 0.2 & x2 < 0.15.Runtime:
Accuracy: