Skip to content

This is the project work of the Combinatorial Decision Making and Optimization course of the Master's degree in Artificial Intelligence of Univeristy of Bologna.

Notifications You must be signed in to change notification settings

rrMat/VLSI-Multi-Paradigm-Solver

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

VLSI Multi Paradigm Solver

Very large-scale integration (VLSI) is the process of integrating or embedding hundreds of thousands of transistors on a single silicon semiconductor microchip.

output

Introduction

This is the project work of the Combinatorial Decision Making and Optimization course of the Master's degree in Artificial Intelligence of Univeristy of Bologna.

The group members are:

Name Surname Email
Alessandro Lombardini alessandr.lombardin3@studio.unibo.it
Giacomo Melacini giacomo.melacini@studio.unibo.it
Matteo Rossi Reich matteo.rossireich@studio.unibo.it
Lorenzo Tribuiani lorenzo.tribuiani@studio.unibo.it

Prerequisites

CP

  • Minizinc
    Install Minizinc IDE and add the bin folder to your System Path

  • Or-Tools

    1. Download the Or-Tools flat-zinc from Or-Tools Git Hub (or_tools_flatzinc_VisualStudio-64bit_v9.3.10497.zip).

    2. Extract the folder

    3. Go on Minizinc IDE and go in: Minizinc -> Preferences -> Solvers

    4. Under the list of solvers select Add new...

    5. In the opening tab insert the following:

      • Name: Or-Tools
      • Solver-ID: com.google.or-tools
      • Version: 9.3.1
      • Executable: path to the bin extracted
      • Solver library path: add a new folder called Or-Tools
      • Select -v flag
    6. Click Add

MIP

Install AMPL from AMPL website

Required Packages and installation

  1. Clone repository

  2. Open a terminal

  3. Create and activate a virtual environment with:

    python -m venv [name-of-venv]

    ./[name-of-venv]/Scripts/activate

  4. Install the requirements with:

    pip install -r requirements.txt

  5. Install MathSat and Z3 via pySMT following the instructions defined here https://pysmt.readthedocs.io/en/latest/getting_started.html

Usage

Once the project is installed, to run an execution give, from the root directory:

python main.py {CP, SAT, SMT, LP} ... [-h] [-v] [-i] [-p] [-r]

In the following table all the optional and positional argument (shared or selective from paradigm) are reported:

TABLE OF COMMANDS
Long argument Short argument Description Allowed params
OPTIONAL SHARED PARAMETERS
--verbose -v Toggle verbosity None
--ins -i Select instance to solve 1-40
omit to execute all
--print_img -p Toggle image saving None
--rotation -r Toggle solving with allowed rotation None
POSITIONAL ARGUMENTS
paradigm Select the solving paradigm CP | SAT | SMT | MIP
CP REQUIRED ARGUMENTS
--model -m Select CP model std | cml | syb (one or more)
--solver -s Select CP solver chuffed | gecode | or-tools (one or more)
SAT REQUIRED ARGUMENTS
--model -m Select SAT model SATModel | SATModelBorders
SAT OPTIONAL ARGUMENTS
--symmetry_breaking -sb Choose if the symmetry breaking constraint has to be used None
--encoding -e Select SAT encoding seq | np | bw | he
SMT REQUIRED ARGUMENTS
--model -m Select SMT model z3Py | z3Py_rotation | z3Py_parallel | pySMT_z3 | pySMT_msat
MIP REQUIRED ARGUMENTS
--mode -m Select MIP model std | strong (one or both)
--solver -s Select MIP solver gurobi | cplex | copt | highs | xpress (one or more)
MIP OPTIONAL ARGUMENTS
--ampl_dir -a If the AMPL installation directory is not in the system search path, specify the full path to the AMPL installation directory String with full path to AMPL installation dir

About

This is the project work of the Combinatorial Decision Making and Optimization course of the Master's degree in Artificial Intelligence of Univeristy of Bologna.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published