我有一个nxm个由非负整数组成的矩阵。例如:

2 3 4 7 1
1 5 2 6 2
4 3 4 2 1
2 1 2 4 1
3 1 3 4 1
2 1 4 3 2
6 9 1 6 4

“投下炸弹”会使目标细胞及其所有八个邻居的数量减少一个,直到最小值为零。

x x x 
x X x
x x x

什么样的算法可以确定将所有细胞减少到零所需的最少炸弹数量?

B选项(因为我不是一个细心的读者)

事实上,问题的第一个版本并不是我要寻找的答案。我没有仔细阅读整个任务,有额外的约束条件,让我们说:

那么简单的问题是,当行中的序列必须是非递增的:

8 7 6 6 5是可能的输入序列

7 8 5 5 2是不可能的,因为7 -> 8在一个序列中增长。

也许为“简单”的问题找到答案会有助于为更难的问题找到解决方案。

PS:我相信当我们有几个相同的情况需要最少的炸弹来清除上面的线时,我们会选择在“左侧”使用最多炸弹的一个。还有什么证据是正确的吗?


当前回答

由于时间不够,我不得不停留在部分解决方案上,但希望即使是这个部分解决方案也能提供解决这个问题的潜在方法的一些见解。

当面对一个困难的问题时,我喜欢想出一些简单的问题来培养对问题空间的直觉。这里,我采取的第一步是将这个二维问题简化为一维问题。考虑一行字:

0 4 2 1 3 0 1

不管怎样,你知道你需要在4点附近炸4次才能把它降到0。因为左边是一个较低的数字,所以轰炸0或4比轰炸2没有任何好处。事实上,我相信(但缺乏严格的证明)轰炸2,直到4点降到0,至少和任何其他策略一样好,让4点降到0。从左到右,我们可以采用如下策略:

index = 1
while index < line_length
  while number_at_index(index - 1) > 0
    bomb(index)
  end
  index++
end
# take care of the end of the line
while number_at_index(index - 1) > 0
  bomb(index - 1)
end

几个轰炸命令示例:

0 4[2]1 3 0 1
0 3[1]0 3 0 1
0 2[0]0 3 0 1
0 1[0]0 3 0 1
0 0 0 0 3[0]1
0 0 0 0 2[0]0
0 0 0 0 1[0]0
0 0 0 0 0 0 0

4[2]1 3 2 1 5
3[1]0 3 2 1 5
2[0]0 3 2 1 5
1[0]0 3 2 1 5
0 0 0 3[2]1 5
0 0 0 2[1]0 5
0 0 0 1[0]0 5
0 0 0 0 0 0[5]
0 0 0 0 0 0[4]
0 0 0 0 0 0[3]
0 0 0 0 0 0[2]
0 0 0 0 0 0[1]
0 0 0 0 0 0 0

从一个需要以某种方式下降的数字开始是一个很有吸引力的想法,因为它突然变得可以找到一个解,就像一些人声称的那样,至少和所有其他解一样好。

The next step up in complexity where this search of at least as good is still feasible is on the edge of the board. It is clear to me that there is never any strict benefit to bomb the outer edge; you're better off bombing the spot one in and getting three other spaces for free. Given this, we can say that bombing the ring one inside of the edge is at least as good as bombing the edge. Moreover, we can combine this with the intuition that bombing the right one inside of the edge is actually the only way to get edge spaces down to 0. Even more, it is trivially simple to figure out the optimal strategy (in that it is at least as good as any other strategy) to get corner numbers down to 0. We put this all together and can get much closer to a solution in the 2-D space.

根据对角子的观察,我们可以肯定地说,我们知道从任何起始棋盘到所有角子都是0的棋盘的最佳策略。这是一个这样的板的例子(我借用了上面两个线性板的数字)。我用不同的方式标记了一些空间,我会解释为什么。

0 4 2 1 3 0 1 0
4 x x x x x x 4
2 y y y y y y 2
1 y y y y y y 1
3 y y y y y y 3
2 y y y y y y 2
1 y y y y y y 1
5 y y y y y y 5
0 4 2 1 3 0 1 0

你会注意到,最上面一行和我们之前看到的线性例子非常相似。回想一下我们之前的观察,将第一行全部降为0的最佳方法是破坏第二行(x行)。轰炸任何y行都无法清除顶部行,轰炸顶部行也没有比轰炸x行相应空间更多的好处。

我们可以从上面应用线性策略(轰炸x行上的相应空间),只关注第一行,不关注其他任何内容。大概是这样的:

0 4 2 1 3 0 1 0
4 x[x]x x x x 4
2 y y y y y y 2
1 y y y y y y 1
3 y y y y y y 3
2 y y y y y y 2
1 y y y y y y 1
5 y y y y y y 5
0 4 2 1 3 0 1 0

0 3 1 0 3 0 1 0
4 x[x]x x x x 4
2 y y y y y y 2
1 y y y y y y 1
3 y y y y y y 3
2 y y y y y y 2
1 y y y y y y 1
5 y y y y y y 5
0 4 2 1 3 0 1 0

0 2 0 0 3 0 1 0
4 x[x]x x x x 4
2 y y y y y y 2
1 y y y y y y 1
3 y y y y y y 3
2 y y y y y y 2
1 y y y y y y 1
5 y y y y y y 5
0 4 2 1 3 0 1 0

0 1 0 0 3 0 1 0
4 x[x]x x x x 4
2 y y y y y y 2
1 y y y y y y 1
3 y y y y y y 3
2 y y y y y y 2
1 y y y y y y 1
5 y y y y y y 5
0 4 2 1 3 0 1 0

0 0 0 0 3 0 1 0
4 x x x x x x 4
2 y y y y y y 2
1 y y y y y y 1
3 y y y y y y 3
2 y y y y y y 2
1 y y y y y y 1
5 y y y y y y 5
0 4 2 1 3 0 1 0

The flaw in this approach becomes very obvious in the final two bombings. It is clear, given that the only bomb sites that reduce the 4 figure in the first column in the second row are the first x and the y. The final two bombings are clearly inferior to just bombing the first x, which would have done the exact same (with regard to the first spot in the top row, which we have no other way of clearing). Since we have demonstrated that our current strategy is suboptimal, a modification in strategy is clearly needed.

在这一点上,我可以退一步,只关注一个角落。让我们考虑一下这个问题:

0 4 2 1
4 x y a
2 z . .
1 b . .

It is clear the only way to get the spaces with 4 down to zero are to bomb some combination of x, y, and z. With some acrobatics in my mind, I'm fairly sure the optimal solution is to bomb x three times and then a then b. Now it's a matter of figuring out how I reached that solution and if it reveals any intuition we can use to even solve this local problem. I notice that there's no bombing of y and z spaces. Attempting to find a corner where bombing those spaces makes sense yields a corner that looks like this:

0 4 2 5 0
4 x y a .
2 z . . .
5 b . . .
0 . . . .

对于这个问题,我很清楚,最优解决方案是轰炸y 5次,z 5次。让我们更进一步。

0 4 2 5 6 0 0
4 x y a . . .
2 z . . . . .
5 b . . . . .
6 . . . . . .
0 . . . . . .
0 . . . . . .

这里,最优解决方案是轰炸a和b 6次,然后x 4次。

现在它变成了一个如何将这些直觉转化为我们可以建立的原则的游戏。

希望能继续!

其他回答

对于更新后的问题,简单的贪心算法可以得到最优结果。

向单元格A[1,1]投掷A[0,0]炸弹,然后向单元格A[2,1]投掷A[1,0]炸弹,并继续向下此过程。要清除左下角,向单元格A[n -2,1]投掷max(A[n -1,0], A[n -2,0], A[n -3,0])炸弹。这将完全清除前3列。

用同样的方法清除第3、4、5列,然后是第6、7、8列,等等。

不幸的是,这并不能帮助找到最初问题的解决方案。


“更大”的问题(没有“非增加”约束)可能被证明是np困难的。这是证明的草图。

假设我们有一个度为3的平面图形。我们来求这个图的最小顶点覆盖。根据维基百科的文章,这个问题对于3次以下的平面图形是np困难的。这可以通过平面3SAT的简化来证明。平面3SAT的硬度由3SAT降低而成。这两个证明都在Erik Demaine教授最近的“算法下界”讲座(第7和第9讲)中提出。

如果我们分割原始图的一些边(图中左边的图),每条边都有偶数个额外的节点,结果图(图中右边的图)应该对原始顶点具有完全相同的最小顶点覆盖。这样的转换允许将图顶点对齐到网格上的任意位置。

如果我们将图顶点只放置在偶数行和列上(这样就不会有两条边与一个顶点形成锐角),在有边的地方插入“1”,在其他网格位置插入“0”,我们可以使用原始问题的任何解决方案来找到最小顶点覆盖。

这将是一个贪婪的方法:

计算一个阶为n X m的“score”矩阵,其中score[i][j]是如果位置(i,j)被炸毁,则矩阵中各点的总扣除额。(一个点的最高分数是9分,最低分数是0分) 逐行移动,找到并选择第一个得分最高的位置(例如(i,j))。 炸弹(i, j)。增加炸弹数量。 如果原矩阵的所有元素都不为零,则转到1。

但我怀疑这是否是最佳解决方案。

编辑:

我上面提到的贪心方法,虽然有效,但很可能不能给我们最优的解决方案。所以我想应该添加一些DP的元素。

我想我们可以同意,在任何时候,具有最高“分数”(分数[I][j] =总扣分,如果(I,j)被炸)的位置之一必须被瞄准。从这个假设开始,下面是新的方法:

NumOfBombs(M):(返回所需的最小炸弹数量)

给定一个矩阵M (n X M),如果M中的所有元素都为0,则返回0。 计算“分数”矩阵M。 设k个不同的位置P1 P2…Pk (1 <= k <= n*m),为m中得分最高的位置。 return (1 + min(NumOfBombs(M1), NumOfBombs(M2),…, NumOfBombs(Mk)) M1, M2,……,Mk是我们轰炸位置P1, P2,…, Pk。

此外,如果我们想在此基础上破坏位置的顺序,我们必须跟踪“min”的结果。

Well, suppose we number the board positions 1, 2, ..., n x m. Any sequence of bomb drops can be represented by a sequence of numbers in this set, where numbers can repeat. However, the effect on the board is the same regardless of what order you drop the bombs in, so really any choice of bomb drops can be represented as a list of n x m numbers, where the first number represents the number of bombs dropped on position 1, the second number represents the number of bombs dropped on position 2, etc. Let's call this list of n x m numbers the "key".

你可以试着先计算1个炸弹投下的所有板子状态,然后用这些来计算2个炸弹投下的所有板子状态,等等,直到你得到所有的0。但是在每一步中,您都将使用上面定义的键缓存状态,因此您可以在计算下一步时使用这些结果(一种“动态规划”方法)。

但是根据n、m的大小和网格中的数字,这种方法的内存需求可能会过多。一旦你计算了N + 1的所有结果,你就可以抛弃N个炸弹投掷的所有结果,所以这里有一些节省。当然,您不能以花费更长的时间为代价缓存任何东西——动态编程方法以内存换取速度。

Pólya说:“如果你不能解决一个问题,那么有一个更容易解决的问题:找到它。”

显然更简单的问题是一维问题(当网格是单行时)。让我们从最简单的算法开始——贪婪地轰炸最大的目标。什么时候会出问题?

给定11 11,贪婪算法对先炸毁哪个单元格无关。当然,中心单元格更好——它一次将所有三个单元格归零。这就提出了一种新的算法a,“炸弹最小化剩余的总和”。这个算法什么时候会出错?

给定1 1 2 11 1,算法A在轰炸第2,第3或第4单元格之间是无所谓的。但是轰炸第二个单元格,留下0 0 11 11比轰炸第三个单元格,留下10 10 10 10 1好。如何解决这个问题?轰炸第三个单元格的问题是,左边的功和右边的功必须分开做。

“炸弹使剩余的总和最小化,但使左边(我们轰炸的地方)的最小值和右边的最小值最大化”如何?叫这个算法b,这个算法什么时候出错?


编辑:在阅读了评论之后,我同意一个更有趣的问题将是改变一维问题,使其两端连接起来。很乐意看到这方面的进展。

评价函数,总和:

int f (int ** matrix, int width, int height, int x, int y)
{
    int m[3][3] = { 0 };

    m[1][1] = matrix[x][y];
    if (x > 0) m[0][1] = matrix[x-1][y];
    if (x < width-1) m[2][1] = matrix[x+1][y];

    if (y > 0)
    {
        m[1][0] = matrix[x][y-1];
        if (x > 0) m[0][0] = matrix[x-1][y-1];
        if (x < width-1) m[2][0] = matrix[x+1][y-1];
    }

    if (y < height-1)
    {
        m[1][2] = matrix[x][y+1];
        if (x > 0) m[0][2] = matrix[x-1][y+1];
        if (x < width-1) m[2][2] = matrix[x+1][y+1];
    }

    return m[0][0]+m[0][1]+m[0][2]+m[1][0]+m[1][1]+m[1][2]+m[2][0]+m[2][1]+m[2][2];
}

目标函数:

Point bestState (int ** matrix, int width, int height)
{
    Point p = new Point(0,0);
    int bestScore = 0;
    int b = 0;

    for (int i=0; i<width; i++)
        for (int j=0; j<height; j++)
        {
            b = f(matrix,width,height,i,j);

            if (b > bestScore)
            {
                bestScore = best;
                p = new Point(i,j);
            }
        }

    retunr p;
}

破坏功能:

void destroy (int ** matrix, int width, int height, Point p)
{
    int x = p.x;
    int y = p.y;

    if(matrix[x][y] > 0) matrix[x][y]--;
    if (x > 0) if(matrix[x-1][y] > 0) matrix[x-1][y]--;
    if (x < width-1) if(matrix[x+1][y] > 0) matrix[x+1][y]--;

    if (y > 0)
    {
        if(matrix[x][y-1] > 0) matrix[x][y-1]--;
        if (x > 0) if(matrix[x-1][y-1] > 0) matrix[x-1][y-1]--;
        if (x < width-1) if(matrix[x+1][y-1] > 0) matrix[x+1][y-1]--;
    }

    if (y < height-1)
    {
        if(matrix[x][y] > 0) matrix[x][y+1]--;
        if (x > 0) if(matrix[x-1][y+1] > 0) matrix[x-1][y+1]--;
        if (x < width-1) if(matrix[x+1][y+1] > 0) matrix[x+1][y+1]--;
    }
}

目标函数:

bool isGoal (int ** matrix, int width, int height)
{
    for (int i=0; i<width; i++)
        for (int j=0; j<height; j++)
            if (matrix[i][j] > 0)
                return false;
    return true;
}

线性最大化函数:

void solve (int ** matrix, int width, int height)
{
    while (!isGoal(matrix,width,height))
    {
        destroy(matrix,width,height, bestState(matrix,width,height));
    }
}

这不是最优的,但可以通过找到更好的评价函数来优化。

. .但是考虑到这个问题,我在想一个主要的问题是在0中间的某个点上得到废弃的数字,所以我要采取另一种方法。这是支配最小值为零,然后试图转义零,这导致一般的最小现有值(s)或这样