lp_solve 连续调用“solve”;不正确
我需要解决一些简单的线性整数规划问题,我使用了 lp_solve 库。任务是获取线性函数的一些连续值的变量值,并对变量进行可能的简单(线性)约束(实际上,即使没有任何附加约束,我也遇到了问题)。例如,我有线性函数 4a + 5b。我感兴趣的第一个值是(函数值 - 变量值):
0 - (0, 0); 4 - (1, 0); 5 - (0, 1); 8 - (2, 0); 9 - (1, 1)
问题在于,在得到 8 - (2, 0)
后,lp_solve 在求解任务时返回 NUMFAILURE
代码 (5)并将其解析为0 - (0, 0)
...
如果我不使用连续调用“解决”函数,而只是从9开始,那么我会得到正确的答案(<代码>9 - (1, 1))。有人可以解释一下吗?代码如下。
#include <iostream>
#include <cstdio>
#include <lpsolve/lp_lib.h>
# if defined ERROR
# undef ERROR
# endif
# define ERROR() { fprintf(stderr, "Error\n"); exit(1); }
using std::cout;
using std::endl;
void print_res(REAL * vars, int size) {
cout << "(";
for (int i = 0; i < size - 1; ++i) {
cout << round(vars[i]) << ", ";
}
cout << round(vars[size - 1]) << ")";
}
int main()
{
lprec *lp;
int majorversion, minorversion, release, build, min = 0;
lp_solve_version(&majorversion, &minorversion, &release, &build);
const int l = 5; // number of iterations
const int dim = 2; // dimension ot current task
char p_data[] = "4 5"; // objective function: p(a, b) = 4a + 5b
if ((lp=make_lp(0, dim)) == NULL)
ERROR();
set_verbose(lp, CRITICAL);
if (!str_add_constraint(lp, p_data, GE, min)) // p(a, b) >= min
ERROR();
// objective function - p
if (!str_set_obj_fn(lp, p_data))
ERROR();
// work with integer non-negative variables
set_int(lp, 1, TRUE);
set_int(lp, 2, TRUE);
set_lowbo(lp, 1, 0);
set_lowbo(lp, 2, 0);
for (int i = 0; i < l; ++i) {
cout << "Status: " << solve(lp) << endl;
REAL vars[dim];
get_variables(lp, vars);
print_res(vars, dim);
// increase minimum value for p
min = round(get_objective(lp));
cout << ", p = " << min << endl;
if (!set_rh(lp, 1, min + 1))
ERROR();
}
return 0;
}
I need to solve some simple linear integer programming-like problem, I took lp_solve library. The task is to obtain variables' values for some consequtive values of linear function with possible simple (linear) constraints on variables (actually I encountered a problem even without any additional constraints). E.g. I have linear function 4a + 5b. First values I'm interested in are (function value - variables values):
0 - (0, 0); 4 - (1, 0); 5 - (0, 1); 8 - (2, 0); 9 - (1, 1)
The problem is that after getting 8 - (2, 0)
, lp_solve returns NUMFAILURE
code (5) while solving the task and resolves it to 0 - (0, 0)
…
If I do not use consequtive calls to 'solve' function and just start from 9 then I got right answer (9 - (1, 1)
). Would anyone please explain this? The code is following.
#include <iostream>
#include <cstdio>
#include <lpsolve/lp_lib.h>
# if defined ERROR
# undef ERROR
# endif
# define ERROR() { fprintf(stderr, "Error\n"); exit(1); }
using std::cout;
using std::endl;
void print_res(REAL * vars, int size) {
cout << "(";
for (int i = 0; i < size - 1; ++i) {
cout << round(vars[i]) << ", ";
}
cout << round(vars[size - 1]) << ")";
}
int main()
{
lprec *lp;
int majorversion, minorversion, release, build, min = 0;
lp_solve_version(&majorversion, &minorversion, &release, &build);
const int l = 5; // number of iterations
const int dim = 2; // dimension ot current task
char p_data[] = "4 5"; // objective function: p(a, b) = 4a + 5b
if ((lp=make_lp(0, dim)) == NULL)
ERROR();
set_verbose(lp, CRITICAL);
if (!str_add_constraint(lp, p_data, GE, min)) // p(a, b) >= min
ERROR();
// objective function - p
if (!str_set_obj_fn(lp, p_data))
ERROR();
// work with integer non-negative variables
set_int(lp, 1, TRUE);
set_int(lp, 2, TRUE);
set_lowbo(lp, 1, 0);
set_lowbo(lp, 2, 0);
for (int i = 0; i < l; ++i) {
cout << "Status: " << solve(lp) << endl;
REAL vars[dim];
get_variables(lp, vars);
print_res(vars, dim);
// increase minimum value for p
min = round(get_objective(lp));
cout << ", p = " << min << endl;
if (!set_rh(lp, 1, min + 1))
ERROR();
}
return 0;
}
如果你对这篇内容有疑问,欢迎到本站社区发帖提问 参与讨论,获取更多帮助,或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱,如果其他用户或者作者回复了您的评论,将不能在第一时间通知您!
发布评论