"""
## GAMSSOURCE: https://www.gams.com/latest/gamslib_ml/libhtml/gamslib_weapons.html
## LICENSETYPE: Demo
## MODELTYPE: NLP
## KEYWORDS: nonlinear programming, assignment problem, military application, nlp test problem
Weapons Assignment (WEAPONS)
This model determines an assignment of weapons to targets in order
to inflict maximum damage at minimal cost. This is a classic
NLP test problem.
Bracken, J, and McCormick, G P, Chapter 2. In Selected Applications of
Nonlinear Programming. John Wiley and Sons, New York, 1968, pp. 22-27.
"""
from __future__ import annotations
import os
import pandas as pd
from gamspy import (
Card,
Container,
Equation,
Model,
Parameter,
Problem,
Product,
Sense,
Set,
Sum,
Variable,
)
def main():
m = Container(
system_directory=os.getenv("SYSTEM_DIRECTORY", None),
)
td_data = pd.DataFrame(
[
["icbm", "2", 0.05],
["icbm", "6", 0.15],
["icbm", "7", 0.10],
["icbm", "8", 0.15],
["icbm", "9", 0.20],
["icbm", "18", 0.05],
["mrbm-1", "1", 0.16],
["mrbm-1", "2", 0.17],
["mrbm-1", "3", 0.15],
["mrbm-1", "4", 0.16],
["mrbm-1", "5", 0.15],
["mrbm-1", "6", 0.19],
["mrbm-1", "7", 0.19],
["mrbm-1", "8", 0.18],
["mrbm-1", "9", 0.20],
["mrbm-1", "10", 0.14],
["mrbm-1", "12", 0.02],
["mrbm-1", "14", 0.12],
["mrbm-1", "15", 0.13],
["mrbm-1", "16", 0.12],
["mrbm-1", "17", 0.15],
["mrbm-1", "18", 0.16],
["mrbm-1", "19", 0.15],
["mrbm-1", "20", 0.15],
["lr-bomber", "1", 0.04],
["lr-bomber", "2", 0.05],
["lr-bomber", "3", 0.04],
["lr-bomber", "4", 0.04],
["lr-bomber", "5", 0.04],
["lr-bomber", "6", 0.10],
["lr-bomber", "7", 0.08],
["lr-bomber", "8", 0.09],
["lr-bomber", "9", 0.08],
["lr-bomber", "10", 0.05],
["lr-bomber", "11", 0.01],
["lr-bomber", "12", 0.02],
["lr-bomber", "13", 0.01],
["lr-bomber", "14", 0.02],
["lr-bomber", "15", 0.03],
["lr-bomber", "16", 0.02],
["lr-bomber", "17", 0.05],
["lr-bomber", "18", 0.08],
["lr-bomber", "19", 0.07],
["lr-bomber", "20", 0.08],
["f-bomber", "10", 0.04],
["f-bomber", "11", 0.09],
["f-bomber", "12", 0.08],
["f-bomber", "13", 0.09],
["f-bomber", "14", 0.08],
["f-bomber", "15", 0.02],
["f-bomber", "16", 0.07],
["mrbm-2", "1", 0.08],
["mrbm-2", "2", 0.06],
["mrbm-2", "3", 0.08],
["mrbm-2", "4", 0.05],
["mrbm-2", "5", 0.05],
["mrbm-2", "6", 0.02],
["mrbm-2", "7", 0.02],
["mrbm-2", "10", 0.10],
["mrbm-2", "11", 0.05],
["mrbm-2", "12", 0.04],
["mrbm-2", "13", 0.09],
["mrbm-2", "14", 0.02],
["mrbm-2", "15", 0.01],
["mrbm-2", "16", 0.01],
]
)
wa_data = pd.DataFrame(
[
["icbm", 200],
["mrbm-1", 100],
["lr-bomber", 300],
["f-bomber", 150],
["mrbm-2", 250],
]
)
tm_data = pd.DataFrame(
[
["1", 30],
["6", 100],
["10", 40],
["14", 50],
["15", 70],
["16", 35],
["20", 10],
]
)
mv_data = pd.DataFrame(
[
["1", 60],
["2", 50],
["3", 50],
["4", 75],
["5", 40],
["6", 60],
["7", 35],
["8", 30],
["9", 25],
["10", 150],
["11", 30],
["12", 45],
["13", 125],
["14", 200],
["15", 200],
["16", 130],
["17", 100],
["18", 100],
["19", 100],
["20", 150],
]
)
# Sets
w = Set(
m,
name="w",
records=["icbm", "mrbm-1", "lr-bomber", "f-bomber", "mrbm-2"],
description="weapons",
)
t = Set(
m,
name="t",
records=[str(i) for i in range(1, 21)],
description="targets",
)
# Parameters
td = Parameter(
m, name="td", domain=[w, t], records=td_data, description="target data"
)
wa = Parameter(
m,
name="wa",
domain=w,
records=wa_data,
description="weapons availability",
)
tm = Parameter(
m,
name="tm",
domain=t,
records=tm_data,
description="minimum number of weapons per target",
)
mv = Parameter(
m,
name="mv",
domain=t,
records=mv_data,
description="military value of target",
)
# Variables
x = Variable(
m,
name="x",
domain=[w, t],
type="Positive",
description="weapons assignment",
)
prob = Variable(
m, name="prob", domain=t, description="probability for each target"
)
# Equations
maxw = Equation(m, name="maxw", domain=w, description="weapons balance")
minw = Equation(
m,
name="minw",
domain=t,
description="minimum number of weapons required per target",
)
probe = Equation(
m, name="probe", domain=t, description="probability definition"
)
maxw[w] = Sum(t.where[td[w, t]], x[w, t]) <= wa[w]
minw[t].where[tm[t]] = Sum(w.where[td[w, t]], x[w, t]) >= tm[t]
probe[t] = prob[t] == 1 - Product(
w.where[td[w, t]], (1 - td[w, t]) ** x[w, t]
)
_ = Sum(t, mv[t] * prob[t])
etd = Sum(
t, mv[t] * (1 - Product(w.where[td[w, t]], (1 - td[w, t]) ** x[w, t]))
)
war = Model(
m,
name="war",
equations=[maxw, minw],
problem=Problem.NLP,
sense=Sense.MAX,
objective=etd,
)
x.l[w, t].where[td[w, t]] = wa[w] / Card(t)
war.solve()
import math
assert math.isclose(war.objective_value, 1735.5696, rel_tol=0.001)
print(war.objective_value)
if __name__ == "__main__":
main()
