BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//132.216.98.100//NONSGML kigkonsult.se iCalcreator 2.20.4//
BEGIN:VEVENT
UID:20251122T001827EST-7321c0sxn8@132.216.98.100
DTSTAMP:20251122T051827Z
DESCRIPTION:LSLQ: An Iterative Method for Linear Least-Squares with an Erro
 r Minimization Property.\n\nAbstract:  We propose an iterative method name
 d LSLQ for solving linear least-squares problems of any shape. The method 
 is based on the Golub and Kahan process\, where the dominant cost is produ
 cts with the linear operator and its transpose. In the rank-deficient case
 \, LSLQ identifies the minimum-length least-squares solution. LSLQ is form
 ally equivalent to SYMMLQ applied to the normal equations\, so that the cu
 rrent estimate's Euclidean norm increases monotonically while the associat
 ed error norm decreases monotonically. We provide lower and upper bounds o
 n the error in the Euclidean norm along the LSLQ iterations. The upper bou
 nd translates to an upper bound on the error norm along the LSQR iteration
 s\, which was previously unavailable\, and provides an error-based stoppin
 g criterion involving a transition to the LSQR point. We report numerical 
 experiments on standard test problems and on a full-wave inversion problem
  arising from geophysics in which an approximate least-squares solution co
 rresponds to an approximate gradient of a relevant penalty function that i
 s to be minimized. Joint work with Ron Estrin and Michael A. Saunders (ICM
 E\, Stanford University).\n
DTSTART:20171023T200000Z
DTEND:20171023T210000Z
LOCATION:Room 1205\, Burnside Hall\, CA\, QC\, Montreal\, H3A 0B9\, 805 rue
  Sherbrooke Ouest
SUMMARY:Dominique Orban (Polytechnique Montréal)
URL:/mathstat/channels/event/dominique-orban-polytechn
 ique-montreal-279359
END:VEVENT
END:VCALENDAR