Array of struct manipulating
4 次查看(过去 30 天)
显示 更早的评论
I have array of struct
x.a
x.b
x.c
I want to do the following two operations
- x(:).a = x(:).a +10
- x(:).c = x(:).a / x(:).b
for all element without using for loop
6 个评论
Jan
2011-11-22
@Majed: Why do you want to avoid the FOR loop?! It would be much nicer and most likely more efficient than creating intermediate vectors, split them to a cell and distribute it over different fields again. The elements are independent from eachotehr, the calculations are independent, so a FOR loop is the straightest method.
回答(4 个)
David Young
2011-11-20
It depends whether your data is a structure of arrays or an array of structures. See this video for the difference (and google matlab structures and arrays for more information).
If you choose to use a structure of arrays, you can just do
x.a = x.a + 10;
etc. If you use an array of structures, it's more fiddly.
Jan
2011-11-21
Unfortunately you did not answer my question about the size and type of the fields and the dimensions of the struct. But let me guess at least - perhaps this helps:
x(1).a = 2;
x(2).a = 3;
x(1).b = 17;
x(2).b = 23;
result = [x.a] ./ [x.b]
Fangjun Jiang
2011-11-21
%%construct a structure array
M=3;
s=struct('a',repmat({1},M,1),'b',repmat({2},M,1),'c',repmat({3},M,1));
%process
N=length(s);
temp=[s.a]+10;
temp=mat2cell(temp,1, ones(N,1));
[s.a]=temp{:};
temp=[s.a]./[s.b];
temp=mat2cell(temp,1, ones(N,1));
[s.c]=temp{:};
0 个评论
Jan
2011-11-22
Fangjun's approach is nice and works without loops. But creating the temporary arrays wastes a lot of time. A simple FOR loop would be more efficient:
M = 3;
s0 = struct('a', repmat({1},M,1), 'b',repmat({2},M,1), 'c',repmat({3},M,1));
tic
for k = 1:1000
s = s0;
N = length(s);
temp = [s.a]+10;
temp = mat2cell(temp,1, ones(N,1));
[s.a] = temp{:};
temp = [s.a]./[s.b];
temp = mat2cell(temp,1, ones(N,1));
[s.c]= temp{:};
end
toc
% Elapsed time is 0.237474 seconds.
tic
for k = 1:1000
s = s0;
for i = 1:numel(s)
s(i).a = s(i).a + 10;
s(i).c = s(i).a / s(i).b;
end
end
toc
% Elapsed time is 0.024864 seconds.
For this tiny data set the FOR loop is 10 times faster. For larger data with N=300, I get at least a speedup factor of 3.
1 个评论
Fangjun Jiang
2011-11-22
Good point, Jan. However, when I changed M to be 300. The result is opposite.
Elapsed time is 3.575805 seconds.
Elapsed time is 5.151734 seconds.
Elapsed time is 3.550913 seconds.
Elapsed time is 5.179092 seconds.
Elapsed time is 3.552488 seconds.
Elapsed time is 5.137140 seconds.
另请参阅
类别
在 Help Center 和 File Exchange 中查找有关 Structures 的更多信息
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!