Calls the CPLEX MILP solver
Version02
Date: Dec 3, 2008, ver02
By: Hooman Askari
--------------------------------------------------------------------------
Inputs
--------------------------------------------------------------------------
code uploads the following variables from the disk
A:
matrix defining linear constraints
b_U:
upper bounds on the linear inequalities
C:
coeefficients in linear objective function
numBlocks:
total number of blocks
numOfPeriods:
total number of scheduling periods
--------------------------------------------------------------------------
outputs
--------------------------------------------------------------------------
Result:
This is saved in Solution.mat
a struct holding the CPLEX solution. the solution is in the
field x_k
--------------------------------------------------------------------------
--------------------------------------------------------------------------0001 % Calls the CPLEX MILP solver 0002 % Version02 0003 % Date: Dec 3, 2008, ver02 0004 % By: Hooman Askari 0005 %-------------------------------------------------------------------------- 0006 %Inputs 0007 %-------------------------------------------------------------------------- 0008 % code uploads the following variables from the disk 0009 % A: 0010 % matrix defining linear constraints 0011 % b_U: 0012 % upper bounds on the linear inequalities 0013 % C: 0014 % coeefficients in linear objective function 0015 % numBlocks: 0016 % total number of blocks 0017 % numOfPeriods: 0018 % total number of scheduling periods 0019 %-------------------------------------------------------------------------- 0020 % outputs 0021 %-------------------------------------------------------------------------- 0022 % Result: 0023 % This is saved in Solution.mat 0024 % a struct holding the CPLEX solution. the solution is in the 0025 % field x_k 0026 %-------------------------------------------------------------------------- 0027 % 0028 %-------------------------------------------------------------------------- 0029 function mip 0030 0031 % profile on 0032 0033 load A 0034 load b_U 0035 load inputToMILPCutsBlocks numBlocks numCuts numOfPeriods 0036 load C; 0037 load inputToMILPCutsBlocks xIP 0038 0039 % name used in TOMLAB solvers 0040 Name='Mine Schedule Mixed Integer Linear Programming'; 0041 0042 [m,n] = size(A) 0043 0044 % numberOfNonZeroElements = nnz(A) 0045 0046 % Make problem on standard form for mipSolve 0047 x_L = zeros(n,1); 0048 x_U = ones(n,1); 0049 x_0 = zeros(n,1); 0050 0051 fprintf('Mine Scheduling Problem. Variables %d. Constraints %d\n',n,m); 0052 0053 x_min = x_L; x_max = x_U; % f_Low = -1E7; 0054 f_Low =[]; % f_Low <= f_optimal must hold 0055 b_L = -inf*ones(m,1); % lower bound 0056 f_opt = []; % f_opt = -141278; 0057 0058 % Problem number not used 0059 nProblem = []; 0060 % Do not use any prior knowledge 0061 fIP = []; 0062 % Do not use any prior knowledge 0063 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 0064 % xIP = []; 0065 % Just define the Prob structure, not any permanent setup file 0066 setupFile = []; 0067 % The optimal integer solution is not known 0068 x_opt = []; 0069 % No variable priorities, largest fractional part will be used 0070 VarWeight = []; 0071 % Run with the knapsack heuristic 0072 KNAPSACK = 1; 0073 0074 % reserve constraint equality 0075 % set the lower bound to the upper bound to enforce the equality 0076 % b_L(numRows+1:end) = b_U(numRows+1:end); 0077 0078 % this is for the binary product equality 0079 b_L( (end - (numBlocks + numCuts) )+1 : end ) = ... 0080 b_U( (end - (numBlocks + numCuts) )+1 : end ); 0081 0082 idx_int = (1:numCuts*numOfPeriods)'; 0083 0084 % intVars[]; % no integer variable 0085 IntVars = [idx_int]; 0086 0087 % Assign routine for defining an MIP problem. 0088 % intVars hold the blocks that have ore and are used as binary 0089 % variables. 0090 Prob = mipAssign(c, A, b_L, b_U, x_L, x_U, x_0, Name, setupFile, ... 0091 nProblem, IntVars, VarWeight, KNAPSACK, fIP, xIP, ... 0092 f_Low, x_min, x_max, f_opt, x_opt); 0093 0094 % Prob = bincont2lin(Prob, idx_prod, idx_bin, idx_cont); 0095 0096 Prob.optParam.IterPrint = 1; % Set to 1 to see iterations zero otherwise. 0097 Prob.Solver.Alg = 2; % Depth First, then Breadth search 0098 Prob.PriLev = 3; 0099 % print iterations information while runnning 0100 Prob.PriLevOpt = 1; 0101 0102 % Calling driver routine tomRun to run the solver. 0103 % The 1 sets the print level after optimization. 0104 0105 0106 % emphasis on memory message 0107 Prob.MIP.cpxControl.NODEFILEIND = 2; 0108 Prob.MIP.cpxControl.MEMORYEMPHASIS = 1; 0109 0110 % Sets an absolute tolerance on the gap between the best integer 0111 % objective and the objective of the best node remaining. 0112 %instruct CPLEX to stop as soon as it has found a feasible integer 0113 % solution proved to be within X% percent of optimal. 0114 % here is set to 5% 0115 Prob.MIP.cpxControl.EPGAP = 0.1; 0116 0117 %Species the amount by which an integer variable can be different from 0118 %an integer and still be considered feasible. 0119 % default = 0.01 0120 Prob.MIP.cpxControl.EPMRK = 0.1; 0121 0122 % Description: Optimality tolerance. 0123 % Influences the reduced-cost tolerance for optimality. 0124 % This parameter governs how closely CPLEX must approach 0125 % the theoretically optimal solution. default 10e-6 0126 Prob.MIP.cpxControl.EPOPT = 10e-4; 0127 0128 % MIP node log display information. 0129 Prob.MIP.cpxControl.MIPDISPLAY = 4; 0130 0131 % MIP emphasis indicator. Emphasize feasibility over optimality 0132 Prob.MIP.cpxControl.MIPEMPHASIS = 1; 0133 0134 % Name of the file to write the CPLEX log information to. 0135 Prob.CPLEX.LogFile = 'logFile.txt'; 0136 0137 % sensitivity analysis 0138 % only for LP problems IntVars =[]; 0139 % saRequest.obj.index = 1:n; 0140 % Prob.CPLEX.sa = saRequest; 0141 0142 % exitFlag = checkFuncs(Prob, 'cplex', 1); 0143 0144 Prob.CPLEX.confgrps = ... 0145 cpxBuildConflict( length(x_L), length(b_L), 0, 0, 0, 'full' ); 0146 0147 Prob.CPLEX.conflictFile = 'cplexC.txt'; 0148 0149 %Result = tomRun('mipSolve', Prob, 1); 0150 Result = tomRun('cplex', Prob, 1); 0151 %Result = tomRun('xpress-mp', Prob, 1); 0152 %Result = tomRun('miqpBB', Prob, 1); 0153 %Result = tomRun('minlpBB', Prob, 1); 0154 %print results as print level 2 0155 0156 PriLev = 3; 0157 PrintResult(Result,PriLev); 0158 0159 % profile report; 0160 0161 f = Result.f_k; 0162 0163 save Solution Result f; 0164 0165 0166 end