Reputation: 23
Recently, I came across this behaviour in Matlab
and I am curious as to know why this happens.
a(1,:) = rand(4,1);
a(2,:) = rand(5,1);
This throws me Subscripted assignment dimension mismatch.
error whereas
a(1,:) = rand(4,1);
a(2,1:5) = rand(5,1);
adjusts the array to the max column size and appends zero to the shorter ones.
My question is: Why the former code snippet doesn't do what the latter does(which seems very logical to me)? I don't see any reason as to why the former code snippet should work the way it does. Or am I missing something?
Upvotes: 0
Views: 54
Reputation: 1580
With a(2,:) = rand(5,1);
, you are attempting to assign a 5x1 vector to a 4x1 matrix column. Hence the error.
With a(2,1:5) = rand(5,1);
, you are explicitly referencing a 5th row, which tells the Matlab engine to expand the matrix accordingly before attempting the assignment operation. Hence, this will succeed.
The way I see it, trying to put 5 values into a 4 row vector likely comes from an error in the code, so the user has to explicitly state that this is what he intends to do. Of course, I can only infer on the intent of Mathworks developers when they specifiy the behavior of their language.
Upvotes: 2