Java-Gurobi学习笔记（二）-Bilinear

例子要求参考于Gurobi自带的案例Bilinear

代码如下

import gurobi.*;
/**
 * @Author：superNova
 * @ClassName：Bilinear
 * @Description：example from gurobi itself(Bilinear)
 * @Date：2022/09/03/21:24
 * @Email：shiyudawei@qq.com
 * @Blog：https://blog.csdn.net/shiyudawei?spm=1018.2226.3001.5343
 */

/* This example formulates and solves the following simple bilinear model :
maximize x
subject to x + y + z <= 10
x * y <= 2 ( bilinear inequality )
x * z + y * z == 1 ( bilinear equality )
x, y, z non - negative (x integral in second version )
*/

public class bilinear {
    public static void main(String[] args) {
        try {
            //creat Model
            GRBEnv grbEnv = new GRBEnv("grbenv");
            GRBModel grbModel = new GRBModel(grbEnv);

            //add variable
            GRBVar x = grbModel.addVar(0.0,GRB.INFINITY,0,GRB.CONTINUOUS,"x");
            GRBVar y = grbModel.addVar(0.0,GRB.INFINITY,0,GRB.CONTINUOUS,"y");
            GRBVar z = grbModel.addVar(0.0,GRB.INFINITY,0,GRB.CONTINUOUS,"z");

            GRBVar[] grbVars = {x,y,z};

            //set Object
            GRBLinExpr target = new GRBLinExpr();
            target.addTerm(1,x);
            grbModel.setObjective(target,GRB.MAXIMIZE);

            //create constraints
            /**
             * subject to x + y + z <= 10
             * x * y <= 2 ( bilinear inequality )
             * x * z + y * z == 1 ( bilinear equality )
             * x, y, z non - negative (x integral in second version )
             */
             GRBLinExpr expr1 = new GRBLinExpr();
             expr1.addTerms(new double[]{1, 1, 1},grbVars);
             grbModel.addConstr(expr1,GRB.LESS_EQUAL,2,"c0");

             GRBQuadExpr expr2 = new GRBQuadExpr();
             expr2.addTerm(1,x);
             expr2.addTerm(1,z);
             expr2.addTerm(1,y,z);
             grbModel.addQConstr(expr2,GRB.EQUAL,1,"c1");

             GRBQuadExpr expr3 = new GRBQuadExpr();
             //expr3.addTerms(new double[]{1,0,0},new GRBVar[]{x,z,y}, new GRBVar[]{y,y,y});
             expr3.addTerm(1,x,y);
             grbModel.addQConstr(expr3,GRB.LESS_EQUAL,2,"c2");

             //solve obj\
            try {
                grbModel.optimize();
                assert false;
            }catch (Exception e){
                System.out.println("Failed");
            }
            grbModel.set(GRB.IntParam.NonConvex,2);
            grbModel.optimize();
            System.out.println("最优解为：" + grbModel.get(GRB.DoubleAttr.ObjVal));
            for (int i = 0; i < grbVars.length; i++) {
                System.out.println(grbVars[i].get(GRB.StringAttr.VarName) + "=" + grbVars[i].get(GRB.DoubleAttr.X));

            }
            System.out.println("---------------------------------second version'result--------------------------");
            //second version requirements
            x.set(GRB.CharAttr.VType,GRB.INTEGER);
            grbModel.optimize();
            System.out.println("最优解为：" + grbModel.get(GRB.DoubleAttr.ObjVal));
            for (int i = 0; i < grbVars.length; i++) {
                System.out.println(grbVars[i].get(GRB.StringAttr.VarName) + "=" + grbVars[i].get(GRB.DoubleAttr.X));

            }
            grbModel.dispose();
            grbEnv.dispose();


        }catch (Exception e){
            e.printStackTrace();
        }
    }
}

输出结果

Optimal solution found (tolerance 1.00e-04)
Best objective 1.000000000000e+00, best bound 1.000000000000e+00, gap 0.0000%
最优解为：1.0
x=1.0
y=0.0
z=0.0

几点注意

 第一

这不是线性的问题，而是非线性的，所以在语法上，有一些区别，主要体现在：

GRBQuadExpr expr2 = new GRBQuadExpr();
expr2.addTerm(1,y,z);
grbModel.addQConstr();

第二

非线性问题会有下面的问题出现

Quadratic equality constraints are non-convex. Set NonConvex parameter to 2 to solve model

所以我们需要设置NonConvex参数为2 ，即

grbModel.set(GRB.IntParam.NonConvex,2);