how_to_use
Differences
This shows you the differences between two versions of the page.
Both sides previous revisionPrevious revisionNext revision | Previous revisionNext revisionBoth sides next revision | ||
how_to_use [2018/01/24 13:11] – admin | how_to_use [2018/01/27 15:19] – liyong | ||
---|---|---|---|
Line 8: | Line 8: | ||
- | Below we demonstrate how to use our learner to learn the ω-regular language L=a< | + | Below we demonstrate how to use our learner to learn the ω-regular language L=a< |
| | ||
=== Prepare a BA file === | === Prepare a BA file === | ||
Line 16: | Line 16: | ||
a, | a, | ||
a, | a, | ||
- | b, | + | a, |
b, | b, | ||
[1] | [1] | ||
[2] | [2] | ||
</ | </ | ||
+ | In ROLL, one can create the above Büchi automaton as follows: | ||
+ | <code java> | ||
+ | // first create an alphabet | ||
+ | Alphabet alphabet = new Alphabet(); | ||
+ | alphabet.addLetter(' | ||
+ | alphabet.addLetter(' | ||
+ | // create an NBA with alphabet | ||
+ | NBA target = new NBA(alphabet); | ||
+ | target.createState(); | ||
+ | target.createState(); | ||
+ | target.createState(); | ||
+ | | ||
+ | // add transitions for NBA recognizing a^w + ab^w | ||
+ | int fst = 0, snd = 1, thd = 2; | ||
+ | target.getState(fst).addTransition(alphabet.indexOf(' | ||
+ | target.getState(fst).addTransition(alphabet.indexOf(' | ||
+ | target.getState(snd).addTransition(alphabet.indexOf(' | ||
+ | target.getState(thd).addTransition(alphabet.indexOf(' | ||
+ | target.setInitial(fst); | ||
+ | target.setFinal(snd); | ||
+ | target.setFinal(thd); | ||
+ | </ | ||
=== Prepare a HANOI file === | === Prepare a HANOI file === | ||
The input automaton should have only one initial state. | The input automaton should have only one initial state. | ||
Line 60: | Line 82: | ||
=== Call a learner and specify the BA construction === | === Call a learner and specify the BA construction === | ||
<code java> | <code java> | ||
- | // specify | + | // use the under-approximation method to construct a BA from an FDFA |
+ | options.approximation = Options.Approximation.UNDER; | ||
+ | // set NBA learner, here we use tree-based | ||
+ | options.algorithm = Options.Algorithm.SYNTACTIC; | ||
+ | options.structure = Options.Structure.TREE; | ||
LearnerBase< | LearnerBase< | ||
if(options.algorithm == Options.Algorithm.NBA_LDOLLAR) { | if(options.algorithm == Options.Algorithm.NBA_LDOLLAR) { | ||
+ | // input teacher as a membership oracle for the learner | ||
| | ||
}else if(options.algorithm == Options.Algorithm.PERIODIC | }else if(options.algorithm == Options.Algorithm.PERIODIC | ||
Line 108: | Line 135: | ||
options.log.println(" | options.log.println(" | ||
</ | </ | ||
+ | |||
+ | The complete code is the class roll.main.Executor. | ||
+ | |||
+ | ====== Set up ROLL in Eclipse ====== | ||
+ | This tutorial has been tested on the following distributions: | ||
+ | * Ubuntu 16.04 LTS, 64-bits | ||
+ | ===== Prerequisites ===== | ||
+ | |||
+ | |||
+ | * Java Development Kit (only tested with JDK 8.0) | ||
+ | * [[https:// | ||
+ | * Javacc plugin in Eclipse | ||
+ | * Maven Integration for Eclipse | ||
+ | * [[https:// | ||
+ | |||
+ | ===== Set up ROLL in Eclipse ===== | ||
+ | |||
+ | |||
+ | 1. Use following command to clone the repository to your local file system. | ||
+ | < | ||
+ | git clone https:// | ||
+ | </ | ||
+ | 2. Import roll-library as Maven Existing Projects. | ||
+ | |||
+ | You may have life cycle problem with Javacc in pom.xml but this can be fixed by the operations suggested by Eclipse. Then you should be able to run ROLL as a java application by choosing roll.main.ROLL as the main class. | ||
===== Acknowledgement ===== | ===== Acknowledgement ===== |
how_to_use.txt · Last modified: 2018/03/29 16:54 by liyong