Subject: paper update Resent-Date: Sun, 20 Feb 2000 14:32:45 -0500 Resent-From: Geoffrey Fox Resent-To: p_gcf@npac.syr.edu Date: Sat, 19 Feb 2000 20:43:00 -0600 From: "West, John E WES" To: "Turcotte, Louis WES" , gcf@npac.syr.edu first, thanks to both of you for your help with my paper draft. i learned a lot from that exercise. in response to several questions that gcf had in reading my last paper draft, i started poking around in my code. turns out i had a combination of several bugs that conspired to give answers that looked reasonable, but were ultimately wrong, invalid, whatever. i now believe that the application of stein's method in this way is not suitable to meet my goals, and that the results in that paper are completely invalid. more to the point, i'm not sure it is even possible to do what i was trying to do and keep the number of samples required to stastically ensure a fixed data quality within reason. the attached file contains a working note that details my thoughts so far. if you've read my paper, you can skip almost all of this note and go right to the section "Why this doesn't work" near the end. my plan at this time is to implement the heuristic work around i mention in my working note and to get on with the business of making switching functions. i still think this problem needs a good solving, but i've lost a lot of time here and think i should move on. hey, i only lost four months of work! if either of you have any thoughts, please pass them on. as always, thanks. jw. <> John E. West | hatter@wes.army.mil US Army Engineer Research and Development Center Information Technology Laboratory Vicksburg, MS --------------------------------------------------------------------- Name: PhD Working Note 1.doc PhD Working Note 1.doc Type: Microsoft Word Document (application/msword) Encoding: base64