package examples;

/* --------------------------------------------------------------------------

* File: Etsp.java

* Version 12.9.0

* --------------------------------------------------------------------------

* Licensed Materials - Property of IBM

* 5725-A06 5725-A29 5724-Y48 5724-Y49 5724-Y54 5724-Y55 5655-Y21

* Copyright IBM Corporation 2001, 2019. All Rights Reserved.

*

* US Government Users Restricted Rights - Use, duplication or

* disclosure restricted by GSA ADP Schedule Contract with

* IBM Corp.

* --------------------------------------------------------------------------

*

* Etsp.java - Solving an earliness-tardiness scheduling problem

* using CPLEX linearization capabilities.

*

* A command line argument indicating the input data file is required

* to run this example.

*

* java Etsp ../../../examples/data/etsp.dat

*/

import ilog.concert.*;

import ilog.cplex.*;

public class Etsp {

static private int Horizon = 10000;

static private class Data {

int nJobs;

int nResources;

int[][] activityOnResource;

double[][] duration;

double[] dueDate;

double[] earlinessCost;

double[] tardinessCost;

Data(String filename) throws IloException, java.io.IOException,

InputDataReader.InputDataReaderException

{

InputDataReader reader = new InputDataReader(filename);

activityOnResource = reader.readIntArrayArray();

duration = reader.readDoubleArrayArray();

dueDate = reader.readDoubleArray();

earlinessCost = reader.readDoubleArray();

tardinessCost = reader.readDoubleArray();

nJobs = dueDate.length;

nResources = activityOnResource.length;

}

}

public static void main (String args[]) {

try {

String filename;

if ( args.length > 1) filename = args[0];

else filename = "./data/etsp.dat";

Data data = new Data(filename);

IloCplex cplex = new IloCplex();

// Create start variables

IloNumVar[][] s = new IloNumVar[data.nJobs][];

for (int j = 0; j < data.nJobs; j++)

s[j] = cplex.numVarArray(data.nResources, 0.0, Horizon);

// State precedence constraints

for (int j = 0; j < data.nJobs; j++) {

for (int i = 1; i < data.nResources; i++)

cplex.addGe(s[j][i], cplex.sum(s[j][i-1], data.duration[j][i-1]));

}

// State disjunctive constraints for each resource

for (int i = 0; i < data.nResources; i++) {

int end = data.nJobs - 1;

for (int j = 0; j < end; j++) {

int a = data.activityOnResource[i][j];

for (int k = j + 1; k < data.nJobs; k++) {

int b = data.activityOnResource[i][k];

cplex.add(cplex.or(

cplex.ge(s[j][a], cplex.sum(s[k][b], data.duration[k][b])),

cplex.ge(s[k][b], cplex.sum(s[j][a], data.duration[j][a]))

));

}

}

}

// The cost is the sum of earliness or tardiness costs of each job

int last = data.nResources - 1;

IloNumExpr costSum = cplex.numExpr();

for (int j = 0; j < data.nJobs; j++) {

double[] points = { data.dueDate[j] };

double[] slopes = { -data.earlinessCost[j],

data.tardinessCost[j] };

costSum = cplex.sum(costSum,

cplex.piecewiseLinear(

cplex.sum(s[j][last], data.duration[j][last]),

points, slopes, data.dueDate[j], 0)

);

}

cplex.addMinimize(costSum);

/*

* 控制 MIP 中速度、可行性、最优性和移动界限之间的折衷

* 0: 在使用缺省设置 BALANCED 时，CPLEX 致力于快速证明最佳解法，但要均衡在优化早期找到高质量可行解法

* 1: 在将此参数设置为 FEASIBILITY 时，CPLEX 将在优化问题时频繁生成更多可行的解法，牺牲一部分最优性证明速度

* 2: 在设置为 OPTIMALITY 时，早期阶段应用于查找可行的解法的工作量较少

* 3: 使用设置 BESTBOUND 时，将通过移动最佳界限值来更着重强调证明最优性，因此顺便检测可行的解法几乎成为成为偶然

* 4: 在参数设置为 HIDDENFEAS 时，MIP 优化器更努力查找非常难于找到的高质量可行的解法，因此在 FEASIBILITY 设置难于找到可接受质量的解法时考虑此设置。

*/

cplex.setParam(IloCplex.Param.Emphasis.MIP, 4);

if ( cplex.solve() ) {

System.out.println("Solution status: " + cplex.getStatus());

System.out.println(" Optimal Value = " + cplex.getObjValue());

}

cplex.end();

}

catch (IloException e) {

System.err.println("Concert exception caught: " + e);

}

catch (InputDataReader.InputDataReaderException ex) {

System.out.println("Data Error: " + ex);

}

catch (java.io.IOException ex) {

System.out.println("IO Error: " + ex);

}

}

}