PTPNperfbound

A performance bound solver for Probabilistic Time Petri Nets

The solver computes:

1. Transitions maximum throughput bound
2. Transitions minimum cycle times (in case of live nets) and identifies the slowest subnet.

The computation is performed using Linear Programming techniques.

Structure of the repository

• dist: distributions (in Wheel format and zipped tar.gz)
• doc: documentation (see Solver modules)
• examples: PTPN model examples and results
• src: source files (solver modules)
• tests: source files (test scripts)
• LICENSE: license file
• README.md: this file
• README.txt: short explanation of the solver
• requirements.txt: Python libraries required to run the solver
• pyproject.toml: setup to package the Python modules and distribute the package

Dependences

• cplex: the optimization problems are solved using IBM CPLEX Studio or CPLEX Studio Community (up to 1000 variables/ contraints).

Besides other Python external libraries are needed, which can be installed with pip: all the dependencies are included in the requirements.txt).

How to install

The solver can currently run with Python3 (versions 3.8/3.9/3.10).

0. Clone or download this repository.

1. Install a virtual environment of Python3 (one of the versions indicated above):

$ python3 -m venv venv ...and activate it: $ source venv/bin/activate

2. Update the pip and setuptools versions:

python3 -m pip install --upgrade pip setuptools

3. Install the libraries required to run the solver:

$ pip install -r requirements.txt

4. Move to the PTPNperfbound directory and set the Python path environment variable to the current path:

export PYTHONPATH=.

5.(optional) The package can be installed from the PyPi repository using the command:

$ pip install PTPNperfbound

or you can install the library using the command:

$ pip install dist/PTPNperfbound-0.0.1-py3-none-any.whl

How to use

The solver can be run using a Command Line Inferface (CLI):

Usage: ptpnbound [OPTIONS] NAME TNAME

Options:
  -lp, --lpmodel TEXT          LP model files (CPLEX  models - lp format)
  -lpo, --lpoutput TEXT        Result files (CPLEX model results - xml format)
  -o, --output <TEXT TEXT>     Result file: <name format> (available formats:
                               pnml, dot)
  -v, --verbose                Print results to stdin
  --help                       Show this message and exit.

where NAME is the pathname of the PTPN model (.pnml) and TNAME is the name of the transition of reference for the bound computation.

In case the last installation step (5.) has not been performed, you can launch the CLI as follows:

python3 src/ptpnbound.py --help

References

S. Bernardi, J. Campos, "A min-max problem for the computation of the cycle time lower bound in interval-based Time Petri Nets," IEEE Transactions on Systems, Man, and Cybernetics: Systems, 43(5), September 2013.

Y. Emzivat, B. Delahaye, D. Lime, O.H. Roux, "Probabilistic Time Petri Nets", HAL Open Science, https://hal.science/hal-01590900.

T. Le Moigne, "ARPE clinical pathway identification algorithms", GitLab repository: https://gitlab.crans.org/bleizi/arpe-identification