You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I've opened this feature request based off the back of a thread on slack with @andreasnoack . Since QR is not as reliable as SVD in determining the rank, we could have an additional kwarg where the user provides an expected dimension of the nullspace. Roughly
functionnullspace(A::AbstractMatrix; nullity::Int=0, kwargs...)
ifiszero(nullity) # default to SVDreturnnullspace(A;kwargs...)
elsereturnqr(A').Q[:,end+1-nullity:end]
endend
A small test
A =randn(3)randn(3)'
nullity =2# expected nullity
A*nullspace(A)
A*qr(A').Q[:,end+1-nullity:end]
The text was updated successfully, but these errors were encountered:
I've opened this feature request based off the back of a thread on slack with @andreasnoack . Since QR is not as reliable as SVD in determining the rank, we could have an additional kwarg where the user provides an expected dimension of the nullspace. Roughly
A small test
The text was updated successfully, but these errors were encountered: