ddctl

A BDD-based model checker for CTL (Computational Tree Logic).

Directly using the model checker library

You can directly use the model checker by called the Model class like below: (Basically this is what the ddctl commandline tool does)

from model import *

m = Model(specfile=args.specfile, ectl_only=args.ectl_only, output=args.output)

print(m.ectlrepr, end=', ')
print(m.check())

# alternatively if you want you can print true or false as well instead of the 
# colorized message
print(m.check().status)

Command Line Tool - Usage

Help

$ ./ddctl -h

Model Checking

$ ./ddctl -s <specfile.json> [--ectl-only] [--output <outputfile.png>]

SPECFILE

The specification file is a json file with the following structure:

The JSON file should contain the kripke structure and the CTL formula you want to see model checked.

The output will give you $K \models \varphi$ or $K \not \models \varphi$, where $K$ is the kripke structure and $\varphi$ is the CTL formula.

No RELEASE operator yet

Example Specfile:

{
  "states" : ["a", "b", "c"],
  "init" : ["a"],
  "T" : {
    "a" : ["b", "c"],
    "b" : ["c"],
    "c" : ["a"]
  },
  "ap" : {
    "a" : ["p", "q"],
    "b" : ["q"]
  },
  "ctlf" : "(AG p) -> AF !q"
}

Note: labels are atomic propositions

Dependencies

Graphviz (for visualization)
PyDD - CuDD Python Binding (for BDDs)
PLY - Python Lex-Yacc (for parsing)

Getting Dependencies

Graphviz

If you don't have graphviz installed, you can install it with:

$ sudo apt-get install graphviz

Install python bindings:

pip3 install graphviz

CuDD

The project internally usescudd python bindings called PyDD for BDDs. The tar file for both cudd and PyDD are included in the lib folder. To build the project, you need to have python3, cython and python3-dev installed.

To get cython, run pip3 install cython
To get python3-dev, run sudo apt-get install python3-dev
1. we need this to build the PyDD library (it requires the python3-dev headers)
To build Cudd,
1. Go to the lib directory
2. Run tar -xvf cudd-3.1.0.tar.gz
3. Go to the cudd directory
4. Run ./configure --enable-dddmp --enable-obj --enable-shared
  1. If you want to build a static library, remove the --enable-shared flag.
  2. You can also change the installation directory by adding --prefix=/path/to/install/dir to the configure command.
5. Run make
  1. If you want to verify that the library is built correctly, run make check (but this is optional)
6. Run make install
7. You should see that the cudd library is installed in the directory specified by the --prefix flag, or in /usr/local/lib if you didn't specify a prefix.
  1. I would recommend that you install it in /usr or /usr/local so that you don't have to add the library to your LD_LIBRARY_PATH environment variable.
To build PyDD,
1. Go to the lib directory
2. Run tar -xvf PyDD-0.2.0.tar.gz
3. Go to the PyDD directory
4. Run python3 setup.py build
5. Run python3 setup.py install
6. You can verify this by opening your python interpreter and running import cudd
Once you have these installed, running the ddctl command as mentioned above in the Usage section should work.

raghudotcc/ddctl