arrays - matlab数组创建随机连续的

Question

我想创建一个 4 维数组，每行包含随机数量的连续元素。这些应该始终从第一列开始并以随机列结束。 示例:

array(:,:,1,1) = [ 1 1 1 0 0 0;
                   1 1 0 0 0 0;
                   1 1 1 1 1 0;
                   ...         ]

可以用 3 个 for 循环做到这一点，但效率很低:

array = zeros(n,n,n,n);
for i= 1:n
   for j = 1:n
      for k =1:n
         rows = ceil(n*rand());
         array(k,1:rows,j,i) = 1;
      end
   end
end

有人能找到更好的解决方案吗？谢谢!!

Answer 1

简单直接的方法(虽然可能不是真正随机的)

rows = 8; %%// Number of rows
cols = 7; %%// Number of columns
ch3 = 3; %%// Number of elements in the 3rd dimension
ch4 = 2; %%// Number of elements in the 4th dimension

array = sort(round(rand(rows,cols,ch3,ch4)),2,'descend')

Bsxfun 方法(如下面的基准测试结果所示，速度要快得多，而且是真正随机的)

%%// Sizes
rows = 8; %%// Number of rows
cols = 7; %%// Number of columns
ch3 = 3; %%// Number of elements in the 3rd dimension
ch4 = 2; %%// Number of elements in the 4th dimension

%%// Get a 2D array with every row starting with a one
rows2 = rows*ch3*ch4;
a1 = reshape(1:rows2*cols,rows2,[]);
col = randi(cols,[1 rows2]);
b1 = bsxfun(@plus,(col-1)*rows2,1:rows2)';%//'
out = bsxfun(@le,a1,b1);

%%// Rearrange those to 4D
a2 = reshape(out',[rows*cols ch3 ch4]);%//'
a3 = reshape(a2,cols,rows,ch3,ch4);
array = permute(a3,[2 1 3 4]);

基准测试结果

我们正在将上述两种方法与 Rody 的方法进行比较。

Datasize I:
rows = 80;  %// Number of rows
cols = 70;  %// Number of columns
ch3  = 30;  %// Number of elements in the 3rd dimension
ch4  = 2; %// Number of elements in the 4th dimension

Results:
Elapsed time with SORT approach is:   0.0083445sec
Elapsed time with BSXFUN approach is: 0.0021sec
Elapsed time with RODY approach is:   0.0063026sec

Datasize II:
rows = 80;  %// Number of rows
cols = 70;  %// Number of columns
ch3  = 30;  %// Number of elements in the 3rd dimension
ch4  = 20; %// Number of elements in the 4th dimension

Results:
Elapsed time with SORT approach is:   0.07875sec
Elapsed time with BSXFUN approach is: 0.012329sec
Elapsed time with RODY approach is:   0.055937sec


Datasize III:
rows = 800;  %// Number of rows
cols = 70;  %// Number of columns
ch3  = 30;  %// Number of elements in the 3rd dimension
ch4  = 20; %// Number of elements in the 4th dimension

Results:
Elapsed time with SORT approach is:   0.87257sec
Elapsed time with BSXFUN approach is: 0.17624sec
Elapsed time with RODY approach is:   0.57786sec


Datasize IV:
rows = 800;  %// Number of rows
cols = 140;  %// Number of columns
ch3  = 30;  %// Number of elements in the 3rd dimension
ch4  = 20; %// Number of elements in the 4th dimension

Results:
Elapsed time with SORT approach is:   1.8508sec
Elapsed time with BSXFUN approach is: 0.35349sec
Elapsed time with RODY approach is:   0.71918sec

从这些发现得出的结论是，bsxfun 看起来是可行的方法，除非您想要处理数十亿个元素，正如 Rody 解决方案产生的内存基准测试结果所暗示的那样。