PlanetMath (more info)
 Math for the people, by the people. Sponsor PlanetMath
Encyclopedia | Requests | Forums | Docs | Wiki | Random | RSS  
Login
create new user
name:
pass:
forget your password?
Main Menu
[parent] Viewing Correction to 'singular value decomposition'
inverse of a nonsquare matrix? by igor

Correction id: 1389
Filed on: 2002-12-21 16:51:12
Status: Accepted on 2003-01-14 22:00:52
Type: Erratum

Correction text:
On the very last line you have "x = V S^-1 U^T b", but S is not necessarily a square matrix. I think I know what you mean S^-1 is the transpose of S
with all the diagonal elements inverted, right? But this is by no means
clear from the entry.

Comment from object owner akrowne:
Yup you're right. Re-using the [\Sigma \\ 0] form of S to fix this.
Discussion
Style: Expand: Order:
forum policy
actually... by akrowne on 2002-12-22 00:06:31
Actually no, it would really be wrong to call S transposed with the diagonal elements inverted "S^-1". I think U, S, and V there are versions that are "shrunk" so that they correspond to the true rank of A, as discussed in the paragraph before. What do you think?

Keep in mind that I didn't originally write this entry, I just adapted it from another source... (doesn't mean we shouldn't fix it, though =)


apk
[ reply | up ]

Interact
new correction | post message