Skip to content

Latest commit

 

History

History
38 lines (32 loc) · 2.06 KB

README.md

File metadata and controls

38 lines (32 loc) · 2.06 KB
===== SAT Solver =====

Made by Clément Moutet and Guillaume Cordonnier

1 ) How to use the program

To make the sat solver work, you just need to give him an input file beginning with

p cnf for the classic SAT, followed by the clauses in the Conjunctive Normal Form (Dimacs format)

p logic for the tseitin transformation. You can write complex clauses that will be transforme in a classic SAT problem and solved by our program. Here is how you should write these clauses
The AND is represented by /\ or & , the OR by / or | , the logical consequence by =>, and the not by !
Example : (5 & 4) => var2

To compile, just type make in the terminal to use the makefile. The generated file is called sat2 and is located in the /bin folder.
To give an input file, just lace it in the stdin or give it as an arument to sat2.
The way of giving argument is stipulated in the help :
Syntax: sat [option] [-flags] [] [-o ]
With one of these options:
--DPLL (by default), use DPLL algorithm
--rand DPLL with random acces
--MOMS MOMS heuristic
--DLIS DLIS heuristic
--help show this help
--hist open a prompt to parse history.slh (learned clauses)
And the following flags:
-w use the watched literals
-l use learned clauses
If the l flag is set, these others flags can be used too:
-i disable interactive mode
-d disable documentation file generation

  1. Strengths and weaknesses

We think one of the advantages of our solver is the diversity of the possibilities of resolution of sat. You can launch it in DPLL, DLIS, MOMS, or rand mode with or without watched literals, with or without learned clauses, witch gives us 16 different heuristics. Our code is very modular, so one can easily add new heuristics .

The weakness of our code is the optimisation of these heuristics. Indeed, always manipulating our code to make it modularisable made us abandon a lot of optimisations that we didn't made again because of the time.