symmetricvectorization
Symmetricvectorization is the process of converting a symmetric matrix into a vector representation that exploits the matrix’s symmetry to remove redundancy. In various contexts it appears under different names, notably vech and svec, with conventions that differ about which triangle is stacked and whether off-diagonal entries are scaled.
One common form is vech, the vectorization of the lower triangular part of a symmetric n-by-n matrix
Another widely used form is svec, the symmetric vectorization that stacks the entries from the upper triangle
Relationship to ordinary vec: for any symmetric S, vec(S) = D vec(S) = D vech(S), where D is
Applications and considerations: symmetricvectorization reduces memory usage and can simplify optimization and algebraic manipulation by avoiding