MATLAB中最大值指标矩阵的生成

可能有更好的方法,我的第一个方法是:

D          = rand(8,4)

[val, sub] = max(D)    
ind        = sub2ind( size(D), sub, 1:4 )

res        = false( size(D) )
res( ind ) = true

M = D==repmat(max(D),size(D,1),1)

或者更优雅地说:

M = bsxfun(@eq, D, max(D))

根据评论,如果您想安全起见,抓住意外的非唯一最大值,请添加以下声明:

M( cumsum(M)>1 ) = false

max() 函数返回的索引)。

我已经为原始问题编写了一个扩展,它可以处理任意多维数组并沿任意指定维度搜索最大值。

我用它来解博弈论中的纳什均衡。希望其他人会觉得它有帮助。

A = rand([3 3 2]);
i = 1; % specify the dimension of A through which we find the maximum

% the following codes find the maximum number of each column of A
% and create a matrix M of the same size with A
% which puts 1 in the cell that contains maximum value, and 0 elsewhere.

[Amax pos] = max(A, [], i);
% pos is a now 1x3x3 matrix (the ith dimension is "shrinked" by the max function)

sub = cell(1, ndims(A));
[sub{:}] = ind2sub(size(pos), (1:length(pos(:)))');
sub{i} = pos(:);

ind = sub2ind(size(A), sub{:});
M = false(size(A));
M(ind) = true;

例子:

A(:,:,1)=

0.0292    0.4886    0.4588
0.9289    0.5785    0.9631
0.7303    0.2373    0.5468

A(:,:,2)=

0.5211    0.6241    0.3674
0.2316    0.6791    0.9880
0.4889    0.3955    0.0377

M(:,:,1)=

 0     0     0
 1     1     1
 0     0     0

 1     0     0
 0     1     1
 0     0     0