Minimum norm least-squares solution to linear equation

The minimum-norm solution computed by

`lsqminnorm`

is of particular interest when several solutions exist. The equation*Ax = b*has many solutions whenever`A`

is underdetermined (fewer rows than columns) or of low rank.`lsqminnorm(A,B,tol)`

is typically more efficient than`pinv(A,tol)*B`

for computing minimum norm least-squares solutions to linear systems.`lsqminnorm`

uses the complete orthogonal decomposition (COD) to find a low-rank approximation of`A`

, while`pinv`

uses the singular value decomposition (SVD). Therefore, the results of`pinv`

and`lsqminnorm`

do not match exactly.For sparse matrices,

`lsqminnorm`

uses a different algorithm than for dense matrices, and therefore can produce different results.