Commit summary from repository starlink
------------------------------------
(Tim Jenness) smurf: Minor tweaks to smf_tophat1
100.0% applications/smurf/libsmf/
(Tim Jenness) smurf: Switch to smf_tophat1 from smf_boxcar1. Slower but more robust.
100.0% applications/smurf/libsmf/
(Tim Jenness) smurf: Fix precedence error when checking quality
100.0% applications/smurf/libsmf/
(Ed Chapin) smurf: add DKS cleaning to map-maker, quality arrays to DKS
5.4% applications/smurf/examples/
76.9% applications/smurf/libsmf/
17.6% applications/smurf/libsmurf/
(Ed Chapin) smurf: only clean with the DKS in sc2clean if requested
100.0% applications/smurf/libsmurf/
(Ed Chapin) smurf: more intelligent defaults for sc2clean
100.0% applications/smurf/defaults/
(David Berry) smurf: Smooth COM using a filter that avoids discontinuities at start and end
100.0% applications/smurf/libsmf/
(Tim Jenness) smurf: handle bad status in loop a bit better (sc2clean)
100.0% applications/smurf/libsmurf/
(Tim Jenness) smurf: clean dark squids in sc2clean before cleaning the bolometers
100.0% applications/smurf/libsmurf/
Commits from repository starlink
-----------------------------
commit 47c7775cbbdc820e69960cc7419f597486ed2f68
Author: Tim Jenness <[log in to unmask]>
Date: Thu Jun 10 17:16:35 2010 -1000
smurf: Minor tweaks to smf_tophat1
Check for bad values regardless in QUA clause.
applications/smurf/libsmf/smf_tophat1.cgen | 9 +++++----
1 files changed, 5 insertions(+), 4 deletions(-)
commit 17e50266ac6f3138ec794511a239d1b9f3dc997d
Author: Tim Jenness <[log in to unmask]>
Date: Thu Jun 10 17:15:53 2010 -1000
smurf: Switch to smf_tophat1 from smf_boxcar1. Slower but more robust.
applications/smurf/libsmf/smf_calcmodel_smo.c | 4 +++-
1 files changed, 3 insertions(+), 1 deletions(-)
commit d8786d2b5677520031acb16545a2b6a8a05bd1a9
Author: Tim Jenness <[log in to unmask]>
Date: Thu Jun 10 16:58:33 2010 -1000
smurf: Fix precedence error when checking quality
I was doing (NOT QUALITY) AND MASK instead of NOT (QUALITY AND MASK).
This makes a big difference when you have some jumps in the time
series.
Before this fix lots of jumps were being injected into the time
series where jumps were located and each iteration things were
getting worse.
applications/smurf/libsmf/smf_calcmodel_pln.c | 6 ++++--
applications/smurf/libsmf/smf_calcmodel_smo.c | 6 ++++--
2 files changed, 8 insertions(+), 4 deletions(-)
commit 9219db02e471b56d2320bf513a5aa70b6e66cee1
Author: Ed Chapin <[log in to unmask]>
Date: Thu Jun 10 17:05:27 2010 -0700
smurf: add DKS cleaning to map-maker, quality arrays to DKS
The default dimmconfig now has cleandk parameters defined for use
when the DKS model is turned on. I also added quality to DKS
to make it easier to keep track of what is going on -- it is
generated by smf_open_file and smf_concat_smfGroup.
applications/smurf/examples/dimmconfig.lis | 23 +++++
applications/smurf/libsmf/smf_clean_dksquid.c | 15 +---
applications/smurf/libsmf/smf_concat_smfGroup.c | 90 +++++++++++++++++----
applications/smurf/libsmf/smf_model_create.c | 34 +++++++-
applications/smurf/libsmf/smf_open_file.c | 32 +++++++-
applications/smurf/libsmf/smf_write_smfData.c | 22 ++++-
applications/smurf/libsmurf/smurf_sc2clean.c | 15 +---
7 files changed, 185 insertions(+), 46 deletions(-)
commit 9ad3c95c90fef1b001c8b25b66d16bd08483261e
Author: Ed Chapin <[log in to unmask]>
Date: Thu Jun 10 11:41:41 2010 -0700
smurf: only clean with the DKS in sc2clean if requested
applications/smurf/libsmurf/smurf_sc2clean.c | 8 +++++++-
1 files changed, 7 insertions(+), 1 deletions(-)
commit 4ffdefe2c1c10cf968fed264f9bda3477ff581ec
Author: Ed Chapin <[log in to unmask]>
Date: Thu Jun 10 11:38:32 2010 -0700
smurf: more intelligent defaults for sc2clean
If we turn on DKS cleaning (with dkclean=1) we will certainly want
to remove the average from them first, so set cleandk.order=0.
Also, since both the DKS and the bolo data are cleaned using the
same routine, we have two confusing/useless parameters cleandk.dkclean
(we clearly don't want to clean the dark squids with themselves!) and
cleandk.flagstat (we don't bother carrying around a quality array...
if we don't want to make a map out of data when the telescope was
stopped that is handled by the quality for the bolometers). So,
these are now separated and explicitly turned off for now...
applications/smurf/defaults/smurf_sc2clean.def | 12 +++++++++---
1 files changed, 9 insertions(+), 3 deletions(-)
commit e26391f62d2726215e21c611efb2f1cca24a973c
Author: David Berry <[log in to unmask]>
Date: Thu Jun 10 15:36:24 2010 +0100
smurf: Smooth COM using a filter that avoids discontinuities at start and end
I have found cases where smoothing the common mode using smf_boxcar1D can
introduce discontinuities at the start and end. For instance, the very
first smoothed COM value is the mean of COM values that come later (since
there are by definition no COM values earlier than the first one). So if
COM begins with a positive slope, the smoothed COM at the start is
biassed up above the slope, to some non-zero positive value. This means
there is a sudden leap at the start from zero to the non-zero value. When
COM is subtracted from the residuals, this leap gets transferred to the
residuals, which then causes the FLT model to introduce bad ringing.
The smf_boxcar1D function attempts to get round this by reducing the size
of the filter box at the start and end of the array. But this doesn't
work because the actual good COM signal does not start at the start of
the array - there is block of bad values first (the padded/apodised
region). By the time the good COM signal starts, the smf_boxcar1D filter
box has assumed it's full size, and so introduces the step mentioned above.
I've changed smf_calcmodel_com to smooth COM using the new routine
smf_tophat1D instead of smf_boxcar1D. The new routine performs the same
job as smf_boxcar1D but takes care to avoid the bias produced by an
assymetric dustribution of bad values in the filter box (it's
consequently slower than smf_boxcar1D).
applications/smurf/libsmf/Makefile.am | 2 +-
applications/smurf/libsmf/smf_calcmodel_com.c | 7 +-
applications/smurf/libsmf/smf_tophat1.cgen | 212 +++++++++++++++++++++++
3 files changed, 218 insertions(+), 3 deletions(-)
commit 71c87f8f5f5cc85d484d173df1926dcd024317ff
Author: Tim Jenness <[log in to unmask]>
Date: Wed Jun 9 18:50:11 2010 -1000
smurf: handle bad status in loop a bit better (sc2clean)
applications/smurf/libsmurf/smurf_sc2clean.c | 58 ++++++++++++-----------
1 files changed, 30 insertions(+), 28 deletions(-)
commit c92a596c2952ae9ba1b828157905647d760c8538
Author: Tim Jenness <[log in to unmask]>
Date: Wed Jun 9 18:46:31 2010 -1000
smurf: clean dark squids in sc2clean before cleaning the bolometers
Previously the dark squids were cleaned after they might be used.
applications/smurf/libsmurf/smurf_sc2clean.c | 19 +++++++++++--------
1 files changed, 11 insertions(+), 8 deletions(-)
|