Bindings for the 'PicoSAT' solver to solve Boolean satisfiability problems (SAT). The boolean satisfiability problem asks the question if a given boolean formula can be TRUE; i.e. does there exist an assignment of TRUE/FALSE for each variable such that the whole formula is TRUE? The package bundles 'PicoSAT' solver release 965 < http://www.fmv.jku.at/picosat/>.
R bindings to the PicoSAT solver release 965 by Armin Biere. The PicoSAT C code is distributed under a MIT style license and is bundled with this package.
devtools::install_github("dirkschumacher/rpicosat")
install.packages("rpicosat")
picosat_sat
can solve a SAT problem. The result is a data.frame
+ meta data, so you can use it with dplyr
et al.picosat_solution_status
applied to the result of picosat_sat
returns either PICOSAT_SATISFIABLE, PICOSAT_UNSATISFIABLE or PICOSAT_UNKNOWNThe following functions can be applied to solutions and make available some statistics generated by the PicoSAT
solver:
picosat_added_original_clauses
#clausespicosat_decisions
#decisionspicosat_propagations
#propagationspicosat_seconds
seconds spent in the C function picosat_sat
picosat_variables
#variablespicosat_visits
#visitsSuppose we want to test the following formula for satisfiability:
(A ⇒ B)∧(B ⇒ C)∧(C ⇒ A)
This can be formulated as a CNF (conjunctive normal form):
(¬A ∨ B)∧(¬B ∨ C)∧(¬C ∨ A)
In rpicosat
the problem is encoded as a list of integer vectors.
formula <- list(c(-1, 2),c(-2, 3),c(-3, 1))
library(rpicosat)res <- picosat_sat(formula)res#> Variables: 3#> Clauses: 3#> Solver status: PICOSAT_SATISFIABLE
Every result is also a data.frame
so you can process the results with packages like dplyr
.
as.data.frame(res)#> variable value#> 1 1 FALSE#> 2 2 FALSE#> 3 3 FALSE
We can also test for satisfiability if we assume that a certain literal is TRUE
or FALSE
picosat_sat(formula, c(1)) # assume A is TRUE#> Variables: 3#> Clauses: 3#> Solver status: PICOSAT_SATISFIABLE
picosat_sat(formula, c(1, -3)) # assume A is TRUE, but C is FALSE#> Solver status: PICOSAT_UNSATISFIABLE
This R package is licensed under MIT. The PicoSAT solver bundled in this package is licensed MIT as well: Copyright (c) Armin Biere, Johannes Kepler University.
There are numerous other packages providing bindings to PicoSAT. After writing this package I discovered another package on github providing bindings to PicoSAT in R.
covr::package_coverage()#> rpicosat Coverage: 39.09%#> src/picosat.c: 37.41%#> R/rpicosat.R: 80.00%#> src/init.c: 100.00%#> src/r_picosat.c: 100.00%