# How to implement the symmetric matrix product which only stored the upper or lower triangle matrix

## How to implement the symmetric matrix product which only stored the upper or lower triangle matrix

For example, matrix A and B are symmetric matrix, which only stored the upper or lower triangle matrix. There isnt blas function to implement the product of A and B, because the symm function need to store the matrix A and B fully.
So, who can tell me how to implement the symmetric matrix product which only stored the upper or lower triangle matrix.

4 posts / 0 new
For more complete information about compiler optimizations, see our Optimization Notice.

Do you mean kernlib TRSMUL/TRSMLU ? Those apparently calculate the product of a symmetric packed matrix with its transpose.

I dont know the kernlib.

I means that there are two symmetric matrix A and B, and are stored the Upper or Lower triangle matrix.

For example
1 2 3 4 2 7
A=2 4 5 B= 2 9 3
3 5 6 7 3 2
and A is sotred the Upper triangle matrix as A=[1 2 3 4 5 6]
B is sotred the Upper triangle matrix as B=[4 2 7 9 3 2]
So, which function is used to implement the A*B?

See here:

CernLib writeups