Improve doc
This commit is contained in:
Arthur Bit-Monnot
parent
13e732b68d
commit
a3839de2e5
5 changed files with 85 additions and 68 deletions
69
README.md
69
README.md
|
|
@ -19,7 +19,6 @@ To run the program in IntelliJ, you can
|
|||
- Select `Run Main.main()`. It should complain that some arguments are missing.
|
||||
- Give it the expected command line arguments : `Run > Edit Configuration`, then fill in the `Program arguments` text box.
|
||||
|
||||
|
||||
## Working on the command line (Gradle)
|
||||
|
||||
Compilation instructions are given for Linux. On Windows you can use the `gradlew.bat` script (but you are on your own).
|
||||
|
|
@ -44,72 +43,8 @@ This is especially useful if you want to run it on another machine.
|
|||
❯ java -jar build/libs/JSP.jar --solver basic --instance ft06
|
||||
```
|
||||
|
||||
The command line above indicates that we want to solve the instance named`ft06` with the `basic` solver. It should give an output like the following :
|
||||
```
|
||||
basic
|
||||
instance size best runtime makespan ecart
|
||||
ft06 6x6 55 1 60 9.1
|
||||
AVG - - 1.0 - 9.1
|
||||
```
|
||||
|
||||
Fields in the result view are the following :
|
||||
- `instance`: name of the instance
|
||||
- `size`: size of the instance `{num-jobs}x{num-tasks}`
|
||||
- `best`: best known resultfor this instance
|
||||
- `runtime`: time taken by the solver in milliseconds (rounded)
|
||||
- `makespan`: makespan of the solution
|
||||
- `ecart`: normalized distance to the best result: `100 * (makespan - best) / best`
|
||||
## Documentation
|
||||
|
||||
One can also specify multiple solvers (below `basic` and `random`) and instances (below `ft06`, `ft10` and `ft20`) for simultaneous testing:
|
||||
Documentation for this project is available here : [https://insa-4ir-meta-heuristiques.github.io/doc/](https://insa-4ir-meta-heuristiques.github.io/doc/)
|
||||
|
||||
```
|
||||
❯ java -jar build/libs/JSP.jar --solver basic random --instance ft06 ft10 ft20
|
||||
|
||||
basic random
|
||||
instance size best runtime makespan ecart runtime makespan ecart
|
||||
ft06 6x6 55 1 60 9.1 999 55 0.0
|
||||
ft10 10x10 930 0 1319 41.8 999 1209 30.0
|
||||
ft20 20x5 1165 0 1672 43.5 999 1529 31.2
|
||||
AVG - - 0.3 - 31.5 999.0 - 20.4
|
||||
```
|
||||
Here the last line give the average `runtime` and `ecart` for each solver.
|
||||
|
||||
```
|
||||
sage: jsp-solver [-h] [-t TIMEOUT] --solver SOLVER [SOLVER ...]
|
||||
--instance INSTANCE [INSTANCE ...]
|
||||
|
||||
Solves jobshop problems.
|
||||
|
||||
named arguments:
|
||||
-h, --help show this help message and exit
|
||||
-t TIMEOUT, --timeout TIMEOUT
|
||||
Solver timeout in seconds for each instance.
|
||||
(default: 1)
|
||||
--solver SOLVER [SOLVER ...]
|
||||
Solver(s) to use (space separated if more than
|
||||
one)
|
||||
--instance INSTANCE [INSTANCE ...]
|
||||
Instance(s) to solve (space separated if more
|
||||
than one). All instances starting with the given
|
||||
String will be selected. (e.g. "ft" will select
|
||||
the instances ft06, ft10 and ft20.
|
||||
```
|
||||
|
||||
|
||||
## IDE Support
|
||||
|
||||
Most IDEs should provide support for importing gradle projects. However, our experience has been best with IntelliJ so far and we would recommend it.
|
||||
|
||||
#### IntelliJ
|
||||
|
||||
IntelliJ has worked best, with out of the box support for the import of gradle projets:
|
||||
https://www.jetbrains.com/help/idea/gradle.html#gradle_import_project_start
|
||||
|
||||
|
||||
#### Eclipse
|
||||
|
||||
Most technical problems have been related to the use of Eclipse so we don't recommend using it unless you have a good reason to.
|
||||
We have however configured gradle to allow generating an eclipse configuration like so :
|
||||
```
|
||||
./gradlew eclipseClean eclipse
|
||||
```
|
||||
|
|
@ -1,5 +1,7 @@
|
|||
# Summary
|
||||
|
||||
- [Welcome](./prelude.md)
|
||||
- [Instances](./instances.md)
|
||||
- [Encodings](./encodings.md)
|
||||
- [Solvers](./solvers.md)
|
||||
- [Entry Points](./entry-points.md)
|
||||
|
|
|
|||
74
doc/src/entry-points.md
Normal file
74
doc/src/entry-points.md
Normal file
|
|
@ -0,0 +1,74 @@
|
|||
# Entry Points
|
||||
|
||||
|
||||
## `jobshop.MainTest`
|
||||
|
||||
This is a java main class that you can use to run some small tests. It comes with some initial code to get you started when exploring the encodings.
|
||||
|
||||
## `jobshop.Main`
|
||||
|
||||
This is the principal entry point that aims at helping you running many solvers on many instances and comparing the results.
|
||||
It expects several command line arguments to specify which solvers to run and which instance to solve.
|
||||
|
||||
|
||||
```
|
||||
Usage: jsp-solver [-h] [-t TIMEOUT] --solver SOLVER [SOLVER ...]
|
||||
--instance INSTANCE [INSTANCE ...]
|
||||
|
||||
Solves jobshop problems.
|
||||
|
||||
named arguments:
|
||||
-h, --help show this help message and exit
|
||||
-t TIMEOUT, --timeout TIMEOUT
|
||||
Solver timeout in seconds for each instance.
|
||||
(default: 1)
|
||||
--solver SOLVER [SOLVER ...]
|
||||
Solver(s) to use (space separated if more than
|
||||
one)
|
||||
--instance INSTANCE [INSTANCE ...]
|
||||
Instance(s) to solve (space separated if more
|
||||
than one). All instances starting with the given
|
||||
String will be selected. (e.g. "ft" will select
|
||||
the instances ft06, ft10 and ft20.
|
||||
```
|
||||
|
||||
|
||||
### Example usage
|
||||
|
||||
```shell
|
||||
# Running the Main program with gradle
|
||||
❯ ./gradlew run --args="--solver basic --instance ft06"
|
||||
```
|
||||
|
||||
The command line above indicates that we want to solve the instance named`ft06` with the `basic` solver. It should give an output like the following :
|
||||
```
|
||||
basic
|
||||
instance size best runtime makespan ecart
|
||||
ft06 6x6 55 1 60 9.1
|
||||
AVG - - 1.0 - 9.1
|
||||
```
|
||||
|
||||
Fields in the result view are the following :
|
||||
- `instance`: name of the instance
|
||||
- `size`: size of the instance `{num-jobs}x{num-tasks}`
|
||||
- `best`: best known resultfor this instance
|
||||
- `runtime`: time taken by the solver in milliseconds (rounded)
|
||||
- `makespan`: makespan of the solution
|
||||
- `ecart`: normalized distance to the best result: `100 * (makespan - best) / best`
|
||||
|
||||
One can also specify multiple solvers (below `basic` and `random`) and instances (below `ft06`, `ft10` and `ft20`) for simultaneous testing:
|
||||
|
||||
```shell
|
||||
❯ ./gradlew run --args="--solver basic random --instance ft06 ft10 ft20"
|
||||
|
||||
basic random
|
||||
instance size best runtime makespan ecart runtime makespan ecart
|
||||
ft06 6x6 55 1 60 9.1 999 55 0.0
|
||||
ft10 10x10 930 0 1319 41.8 999 1209 30.0
|
||||
ft20 20x5 1165 0 1672 43.5 999 1529 31.2
|
||||
AVG - - 0.3 - 31.5 999.0 - 20.4
|
||||
```
|
||||
Here the last line give the average `runtime` and `ecart` for each solver.
|
||||
|
||||
|
||||
**Tip:** When selecting instance to solve, you can only provide a prefix to instance name. All instances that start with this prefix will be selected.
|
||||
1
doc/src/prelude.md
Normal file
1
doc/src/prelude.md
Normal file
|
|
@ -0,0 +1 @@
|
|||
# Welcome
|
||||
|
|
@ -24,4 +24,9 @@ It repeats this process until the deadline to produce a result is met and finall
|
|||
The greddy solver is not implemented yet.
|
||||
Its constructor accepts a parameter that specifies the priority that should be used to produce solutions.
|
||||
|
||||
## Descent Solver
|
||||
## Descent Solver
|
||||
|
||||
|
||||
### Neighborhoods
|
||||
|
||||
TODO
|
||||
Loading…
Reference in a new issue