rlog does not recognize multiple $est steps captured to separate files.
Closed this issue · 4 comments
GoogleCodeExporter commented
Per M. Dodds:
You may consider raiding the .ctl file for all of the $EST steps to determine
if a FILE statement was used, and perhaps collect those in the rlog() return
structure, adding a column to indicate which FILE the estimates came from. In
the event that no FILE was specified, that column would be “XXX.ext” and
otherwise look very similar to the current output. Or, perhaps a parameter
like “estfile” to PLOTR and the like that’s defaulted to
glue(run,”.ext”) which could be a character vector. That wouldn’t trip
up default use (no FILE specified in $EST), but would allow one to get into
detail about the estimation results for multiple $EST steps.
Original issue reported on code.google.com by bergs...@gmail.com
on 19 Feb 2013 at 3:23
GoogleCodeExporter commented
Similarly, rlog does not recognize multiple est steps captured in a single file.
Original comment by bergs...@gmail.com
on 14 Mar 2013 at 1:49
GoogleCodeExporter commented
I have gone through the code base exhaustively. I can't find anywhere that a
single $EST method is assumed, in the literal sense (but see below).
read.nmctl() will simply return multiple elements named $est if they are
present.
Granted, it is a pain to manipulate just one of the multiples, since you cannot
address it unambiguously by name. The same is true for multiple omegas, etc.
In such cases, we could possibly substitute a numbered list with the same name
(but that could break a lot of existing code).
iterations() does scavenge lines with "ITERATION NO." from the *.lst file. If
more than one $EST produces such lines, confusion could result. iterations() is
called by PLOTR() for standard diagnostic plotting after each run.
xyplotExt() digests the *.ext file, and possibly assumes too much. This is a
standalone function.
as.unilog.pxml also digests the *.ext file, and is in turn invoked by
omegacor(), sigmacor(), and especially rlog().
Original comment by bergs...@gmail.com
on 15 Apr 2013 at 3:15
GoogleCodeExporter commented
substituting numbered lists e.g. replace this ...
str(ctl)
List of 16
$ sizes : chr [1:2]
$ problem : chr [1:2]
$ input : chr [1:5]
$ data : chr [1:3]
$ subroutines: chr [1:2]
$ model : chr[1:3]
$ pk : chr [1:40]
$ des : chr [1:12]
$ error : chr [1:27]
$ theta : chr [1:11]
$ omega : chr [1:11]
$ omega : chr [1:3]
$ sigma : chr [1:4]
$ est : chr [1:3]
$ est : chr [1:3]
$ est : chr[1:3]
with this ...
str(ctl)
List of 13
$ sizes : chr [1:2]
$ problem : chr [1:2]
$ input : chr [1:5]
$ data : chr [1:3]
$ subroutines: chr [1:2]
$ model : chr[1:3]
$ pk : chr [1:40]
$ des : chr [1:12]
$ error : chr [1:27]
$ theta : chr [1:11]
$ omega : List of 2
..$ chr [1:11]
..$ chr [1:3]
$ sigma : chr [1:4]
$ est : List of 3
..$ : chr [1:3]
..$ : chr [1:3]
..$ : chr [1:3]
Original comment by bergs...@gmail.com
on 15 Apr 2013 at 4:56
GoogleCodeExporter commented
see xlog in version 5.38.
Original comment by bergs...@gmail.com
on 16 Apr 2013 at 9:53
- Changed state: Fixed