(a) 先验算可行性:代入 v 有
- 第 1 行:−x4−x5=−2 成立;
- 第 2 行:x3+x4=2 成立;
- 第 3 行:2x4+x5+x6=4 成立;
- 第 4 行:2x2+x5+x7=3 成立;
且 v≥0,所以 v 可行。
再看是否为 BFS:只需存在一个 4 列组成的可逆子矩阵作为基,使得非基变量为 0 时得到 v。下题 (b) 给出的 B={4,5,6,7} 正好做到,因此 v 是一个基本可行解(并且是退化的,因为 x5=x6=0 但可作为基变量)。
(b) 取基变量 (x4,x5,x6,x7),令非基变量 x1=x2=x3=0,由 Ax=b 解得
- 第 2 行:x4=2;
- 第 1 行:−x4−x5=−2⇒x5=0;
- 第 3 行:2x4+x5+x6=4⇒x6=0;
- 第 4 行:x5+x7=3⇒x7=3。
得到 xB=(2,0,0,3)≥0,故 B 是可行基。
(c) 用 x1,x2,x3 参数化。由等式组:
- 从 x3+x4=2 得 x4=2−x3;
- 从第 1 行 x1−x2−x4−x5=−2 得 x5=x1−x2+x3;
- 从第 3 行 2x4+x5+x6=4 得 x6=x3−x1+x2;
- 从第 4 行 2x2+x5+x7=3 得 x7=3−x1−x2−x3。
于是原目标函数
max(x4+x5+x6+x7)=max(5−x1−x2),
并且约束 x≥0 等价于以下关于 x1,x2,x3 的不等式:
x1≥0,x2≥0,x3≥0,x3≤2,x1−x2+x3≥0,x3−x1+x2≥0,x1+x2+x3≤3.
这就是只含 x1,x2,x3 的 LP。
(d) 在 (c) 中,目标是最大化 5−x1−x2。由于 x1,x2≥0,故
5−x1−x2≤5.
取 x1=x2=0 且例如 x3=0(满足所有约束)即可达到目标值 5。
当 (x1,x2,x3)=(0,0,0) 时,对应
(x4,x5,x6,x7)=(2,0,0,3),
即点 v,其目标值为 2+0+0+3=5,达到上界,因此 v 最优。
英文解析
-
- (a) * * Prior calculation feasibility: Replace with v Yes
- Line 1: −x4−x5=−2 established;
- Line 2: x3+x4=2 established;
- Line 3: 2x4+x5+x6=4 is established;
- Line 4: 2x2+x5+x7=3 established;
And v≥0, so vworks.
Look again at whether it is BFS: it is only necessary to have a reversible submatrix consisting of 4 columns as the base, so that when the non-basic variable is 0, v is obtained. The B={4,5,6,7} given in question (b) below does just that, so v is a basically feasible solution (and is degenerate, because x5=x6=0 can be used as a base variable).
-
- (b) * * Take the base variable (x4,x5,x6,x7), so that the non-base variable x1=x2=x3=0is solved by Ax=b
- Line 2: x4=2;
- Line 1: −x4−x5=−2⇒x5=0;
- Line 3: 2x4+x5+x6=4⇒x6=0;
- Line 4: x5+x7=3⇒x7=3.
You get xB=(2,0,0,3)≥0, so Bis a viable basis.
-
- (c) * * Parameterized with x1,x2,x3. By Equation Group:
-x4=2−x3 fromx3+x4=2;
-x5=x1−x2+x3 from line 1x1−x2−x4−x5=−2;
- Get x6=x3−x1+x2 from line 3 2x4+x5+x6=4;
- Get x7=3−x1−x2−x3 from line 4 2x2+x5+x7=3.
So the original objective function
max(x4+x5+x6+x7)=max(5−x1−x2),
and constrain x≥0 to be equivalent to the following inequality for x1,x2,x3:
x1≥0,x2≥0,x3≥0,x3≤2,x1−x2+x3≥0,x3−x1+x2≥0,x1+x2+x3≤3.
This is the LP with only x1,x2,x3.
-
- (d) * * In (c), the goal is to maximize 5−x1−x2. As a result of x1,x2≥0,
5−x1−x2≤5.
Take x1=x2=0 and for example x3=0 (meets all constraints) to reach the target value of 5.
When (x1,x2,x3)=(0,0,0), corresponds to
(x4,x5,x6,x7)=(2,0,0,3),
That is, v, the target value is 2+0+0+3=5, which reaches the upper limit, so vis the best.