CODEWITHSUNDEEP
pythonconstraint-programmingor-toolscp-sat
Stradivari
Stradivari

Reputation: 2766

Ceiling in CP-SAT

In a scheduling problem. I want to assign vacations proportionally to the days worked. Right now I have a working solution using multiple boolean flags but it doesn't scale very well.

from ortools.sat.python import cp_model
from math import ceil

model = cp_model.CpModel()
solver = cp_model.CpSolver()

days = model.NewIntVar(0, 365, 'days')
vacations = model.NewIntVar(0, 30, 'vacations')

for i in range(365 + 1):
    flag_bool = model.NewBoolVar(f'days == {i}')
    model.Add(days == i).OnlyEnforceIf(flag_bool)
    model.Add(days != i).OnlyEnforceIf(flag_bool.Not())
    model.Add(vacations == ceil(i * 30 / 365)).OnlyEnforceIf(flag_bool)

# test
model.Add(days == 300)

status = solver.Solve(model)

print(solver.Value(days), solver.Value(vacations))

Any suggestions?

Edit: A more general question would be if there is a better way to implement a precalculated arbitrary mapping of one variable to another.

Upvotes: 3

Views: 568

Answers (1)

Stradivari
Stradivari

Reputation: 2766

Solution following Laurent's suggestion:

ceil(days * 30 / 365) == (days * 30 + 364) // 365

Therefore

from ortools.sat.python import cp_model

model = cp_model.CpModel()
solver = cp_model.CpSolver()

days = model.NewIntVar(0, 365, 'days')
vacations = model.NewIntVar(0, 30, 'vacations')
tmp = model.NewIntVar(364, 365 * 30 + 364, 'days*30+364')

model.Add(tmp == days * 30 + 364)
model.AddDivisionEquality(vacations, tmp, 365)

# test
model.Add(days == 300)

status = solver.Solve(model)

print(solver.Value(days), solver.Value(vacations))

Upvotes: 3

Related Questions