5232
|
1 % GLPKMEX Parameters list |
|
2 % |
|
3 % This document describes all control parameters currently implemented |
|
4 % in the GLPK, an Octave interface for the GLPK library. Symbolic names |
|
5 % of control parameters and corresponding codes of GLPK are given on the |
|
6 % left. Types, default values, and descriptions are given on the right. |
|
7 % |
|
8 % ----------------------- |
|
9 % 1 Integer parameters |
|
10 % ----------------------- |
|
11 % |
|
12 % msglev type: integer, default: 1 (LPX_K_MSGLEV) |
|
13 % Level of messages output by solver routines: |
|
14 % 0 no output |
|
15 % 1 error messages only |
|
16 % 2 normal output |
|
17 % 3 full output (includes informational messages) |
|
18 % |
|
19 % scale type: integer, default: 1 (LPX_K_SCALE) |
|
20 % Scaling option: |
|
21 % 0 no scaling |
|
22 % 1 equilibration scaling |
|
23 % 2 geometric mean scaling, then equilibration scaling |
|
24 % |
|
25 % dual type: integer, default: 0 (LPX_K_DUAL) |
|
26 % Dual simplex option: |
|
27 % 0 do not use the dual simplex |
|
28 % 1 if initial basic solution is dual feasible, use the |
|
29 % dual simplex |
|
30 % |
|
31 % price type: integer, default: 1 (LPX_K_PRICE) |
|
32 % Pricing option (for both primal and dual simplex): |
|
33 % 0 textbook pricing |
|
34 % 1 steepest edge pricing |
|
35 % |
|
36 % round type: integer, default: 0 (LPX_K_ROUND) |
|
37 % Solution rounding option: |
|
38 % 0 report all primal and dual values "as is" |
|
39 % 1 replace tiny primal and dual values by exact zero |
|
40 % |
|
41 % itlim type: integer, default: -1 (LPX_K_ITLIM) |
|
42 % Simplex iterations limit. If this value is positive, it is |
|
43 % decreased by one each time when one simplex iteration has |
|
44 % been performed, and reaching zero value signals the solver |
|
45 % to stop the search. Negative value means no iterations limit. |
|
46 % |
|
47 % itcnt type: integer, initial: 0 (LPX_K_ITCNT) |
|
48 % Simplex iterations count.This count is increased by one |
|
49 % each time when one simplex iteration has beenperformed. |
|
50 % |
|
51 % outfrq type: integer, default: 200 (LPX_K_OUTFRQ) |
|
52 % Output frequency, in iterations. This parameter specifies |
|
53 % how frequently the solver sends information about the solution |
|
54 % to the standard output. |
|
55 % |
|
56 % branch type: integer, default: 2 (LPX_K_BRANCH) |
|
57 % Branching heuristic option (for MIP only): |
|
58 % 0 branch on the first variable |
|
59 % 1 branch on the last variable |
|
60 % 2 branch using a heuristic by Driebeck and Tomlin |
|
61 % |
|
62 % btrack type: integer, default: 2 (LPX_K_BTRACK) |
|
63 % Backtracking heuristic option (for MIP only): |
|
64 % 0 depth first search |
|
65 % 1 breadth first search |
|
66 % 2 backtrack using the best projection heuristic |
|
67 % |
|
68 % presol type: int, default: 1 (LPX_K_PRESOL) |
|
69 % If this flag is set, the routine lpx_simplex solves the |
|
70 % problem using the built-in LP presolver. Otherwise the LP |
|
71 % presolver is not used. |
|
72 % |
5237
|
73 % lpsolver type: int, default: 1 |
|
74 % Select which solver to use: |
|
75 % 1 revised simplex method |
|
76 % 2 interior point method |
|
77 % If the problem is a MIP problem this flag will be ignored. |
|
78 % |
|
79 % save type: int, default 0 |
|
80 % If this parameter is nonzero, save a copy of the problem |
|
81 % problem in CPLEX LP format to the file "outpb.lp". There |
|
82 % is currently no way to change the name of the output file. |
5232
|
83 % |
|
84 % ----------------------- |
|
85 % 2 Real parameters |
|
86 % ----------------------- |
|
87 % |
|
88 % relax type: real, default: 0.07 (LPX_K_RELAX) |
|
89 % Relaxation parameter used in the ratio test. If it is zero, the |
|
90 % textbook ratio test is used. If it is non-zero (should be |
|
91 % positive), Harris' two-pass ratio test is used. In the latter |
|
92 % case on the first pass of the ratio test basic variables (in |
|
93 % the case of primal simplex) or reduced costs of non-basic |
|
94 % variables (in the case of dual simplex) are allowed to slightly |
|
95 % violate their bounds, but not more than (RELAX � TOLBND) or |
|
96 % (RELAX �TOLDJ) (thus, RELAX is a percentage of TOLBND or TOLDJ). |
|
97 % |
|
98 % tolbnd type: real, default: 10e-7 (LPX_K_TOLBND) |
|
99 % Relative tolerance used to check ifthe current basic solution |
|
100 % is primal feasible (Do not change this parameter without detailed |
|
101 % understanding its purpose). |
|
102 % |
|
103 % toldj type: real, default: 10e-7 (LPX_K_TOLDJ) |
|
104 % Absolute tolerance used to check if the current basic solution |
|
105 % is dual feasible (Do not change this parameter without detailed |
|
106 % understanding its purpose). |
|
107 % |
|
108 % tolpiv type: real, default: 10e-9 (LPX_K_TOLPIV) |
|
109 % Relative tolerance used to choose eligible pivotal elements of |
|
110 % the simplex table (Do not change this parameter without detailed |
|
111 % understanding its purpose). |
|
112 % |
|
113 % objll type: real, default: -DBL_MAX (LPX_K_OBJLL) |
|
114 % Lower limit of the objective function.If on the phase II the |
|
115 % objective function reaches this limit and continues decreasing, |
|
116 % the solver stops the search.(Used in the dual simplex only) |
|
117 % |
|
118 % objul type: real, default: +DBL_MAX (LPX_K_OBJUL) |
|
119 % Upper limit of the objective function. If on the phase II the |
|
120 % objective function reaches this limit and continues increasing, |
|
121 % the solver stops the search.(Used in the dual simplex only.) |
|
122 % |
|
123 % tmlim type: real, default: -1.0 (LPX_K_TMLIM) |
|
124 % Searching time limit, in seconds. If this value is positive, |
|
125 % it is decreased each time when one simplex iteration has been |
|
126 % performed by the amount of time spent for the iteration, and |
|
127 % reaching zero value signals the solver to stop the search. |
|
128 % Negative value means no time limit. |
|
129 % |
|
130 % outdly type: real, default: 0.0 (LPX_K_OUTDLY) |
|
131 % Output delay, in seconds. This parameter specifies how long |
|
132 % the solver should delay sending information about the solution |
|
133 % to the standard output. Non-positive value means no delay. |
|
134 % |
|
135 % tolint type: real, default: 10e-5 (LPX_K_TOLINT) |
|
136 % Relative tolerance used to check ifthe current basic solution is |
|
137 % integer feasible.(Do not change this parameter without detailed |
|
138 % understanding its purpose). |
|
139 % |
|
140 % tolobj type: real, default: 10e-7 (LPX_K_TOLOBJ) |
|
141 % Relative tolerance used to check if the value of the objective |
|
142 % function is not better than in the best known integer feasible |
|
143 % solution. (Do not change this parameter without detailed |
|
144 % understanding its purpose) |
|
145 % |
|
146 % |
|
147 % ----------------------- |
|
148 % 3 Octave Example |
|
149 % ----------------------- |
|
150 % |
|
151 % % Problem data |
|
152 % s=-1; |
|
153 % c=[10,6,4]'; |
|
154 % a=[1,1,1;... |
|
155 % 10,4,5;... |
|
156 % 2,2,6]; |
|
157 % b=[100,600,300]'; |
|
158 % ctype=['U','U','U']'; |
|
159 % lb=[0,0,0]'; |
|
160 % ub=[]; |
|
161 % vartype=['C','C','C']'; |
|
162 % |
|
163 % % Setting parameters |
|
164 % param.msglev=1; % error messages only |
|
165 % param.itlim=100; % Simplex iterations limit = 100 |
|
166 % |
|
167 % [xmin,fmin,status,lambda,extra]=glpkmex(s,c,a,b,ctype,lb,ub,vartype,param) |
|
168 % |
|
169 |