let us consider following situation,for instance we have given matrix and we want to make zero centered this matrix in columns,so
A=rand(4,3)
A =
0.6948 0.4387 0.1869
0.3171 0.3816 0.4898
0.9502 0.7655 0.4456
0.0344 0.7952 0.6463
now this two method works properly
A-repmat(mean(A),size(A,1),1)
ans =
0.1957 -0.1565 -0.2553
-0.1820 -0.2137 0.0476
0.4511 0.1703 0.0035
-0.4647 0.1999 0.2042
and also
bsxfun(@minus,A,mean(A))
ans =
0.1957 -0.1565 -0.2553
-0.1820 -0.2137 0.0476
0.4511 0.1703 0.0035
-0.4647 0.1999 0.2042
but somehow following method does not work
B=mean(A)
B =
0.4991 0.5953 0.4421
A-repmat(B,1,4)
ans =
0 0 0 0 0 0 0 0 0 0 0 0
i have tried transpose but
A-repmat(B,1,4)'
Error using -
Matrix dimensions must agree.
also i have tried following
A-repmat(B,1,3)'
Error using -
Matrix dimensions must agree.
>> A-repmat(B,1,3)
Error using -
Matrix dimensions must agree.
so what is problem of failure of this method?
You are not using proper syntax for the repmat function
In your example ,You need to create a matrix of size 4 x 3 using repmat
Now a call to repmat(A,k,j) repeats matrix A k times along the first dimension (i.e vertically) and j times along the second dimension (i.e, horizontally).
Here, you need to repeat matrix mean 4 times in first dimension and 1 time in second dimension .
Hence, the correct call to repmat is repmat(mean,4,1)
repmat(B,4,1)
ans =
0.4991 0.5953 0.4421
0.4991 0.5953 0.4421
0.4991 0.5953 0.4421
0.4991 0.5953 0.4421
It looks like you need to know why your method is failing
repmat(B,1,4) %// returns a 1x12 matrix while A is 3x4 matrix hence dimensions do not agree while using @minus
ans =
0.4991 0.5953 0.4421 0.4991 0.5953 0.4421 0.4991 0.5953 0.4421 0.4991 0.5953 0.4421