Page 1 of 1
?LAQR5, ?LASQ3, ?LASQ4: some fixes in doc
Posted:
Mon Apr 11, 2016 9:38 amby zotkevich
?LAQR5 - in description of Z ILO and IHI should be replaced with ILOZ and IHIZ.
?LASQ3, ?LASQ4 - size of Z should be ( 4*N0 ) not ( 4*N )
W.B.R.
Alex Zotkevich, Intel Co.
Re: ?LAQR5, ?LASQ3, ?LASQ4: some fixes in doc
Posted:
Wed Jun 15, 2016 12:52 amby admin
Thank you for reporting the issue.
Corrected in revision 1756
Julie
Re: ?LAQR5, ?LASQ3, ?LASQ4: some fixes in doc
Posted:
Wed Jun 15, 2016 2:37 amby vchalupecky
?LAQR5 - in description of Z ILO and IHI should be replaced with ILOZ and IHIZ.
I'm not sure that's that the correct replacement. ILOZ and IHIZ are the rows of Z to which the transformation is applied.
Disentangling the column indices from the code is not easy but it seems to me that the correct range is (please check) KTOP:KBOT.
Re: ?LAQR5, ?LASQ3, ?LASQ4: some fixes in doc
Posted:
Fri Jun 17, 2016 3:11 pmby Julien Langou
The comment is correct. The subroutine does operate on the rows from ILOZ to IHIZ of Z. KBOT and KTOP are the first and last rows and columns of an isolated diagonal block upon which the QR sweep is to be applied. If WANTZ = .TRUE., then the QR Sweep orthogonal similarity transformation ("KBOT and KTOP") is accumulated into Z(ILOZ:IHIZ,ILOZ:IHIZ) from the right. (So from the right, means that the QR Sweep orthogonal similarity transformation works on the columns of Z.) Julien.
Re: ?LAQR5, ?LASQ3, ?LASQ4: some fixes in doc
Posted:
Mon Jun 20, 2016 8:34 pmby vchalupecky
Ok, thanks for the explanation and sorry for the noise.