Commit summary from repository starlink
------------------------------------
(Tim Jenness) smurf: Add SMO smoothing model to iterative map-maker
99.9% applications/smurf/libsmf/
(Tim Jenness) smurf: Check the bits in PLN and not multiply them
100.0% applications/smurf/libsmf/
(Tim Jenness) smurf: Assigning a negative number to a size_t does not work
100.0% applications/smurf/libsmf/
(Ed Chapin) smurf: some test code for alternative sample variance calc in maps
100.0% applications/smurf/libsmf/
(Ed Chapin) smurf: clean up needlessly complicated init in smf_model_create
100.0% applications/smurf/libsmf/
(Ed Chapin) smurf: replacing dead DKS with average now optional (dks.replacebad)
99.0% applications/smurf/libsmf/
(David Berry) smurf: heavier clipping when finding RMS gradient in dc step correction
100.0% applications/smurf/libsmf/
(David Berry) smurf: Fix bad quality indexing in step correction
100.0% applications/smurf/libsmf/
(David Berry) Revert "smurf: Correct indexing of quality array when DC steps found close to start or end"
100.0% applications/smurf/libsmf/
(David Berry) smurf: Correct indexing of quality array when DC steps found close to start or end
100.0% applications/smurf/libsmf/
(David Berry) smurf: Ignore previously flagged jumps when fitting data to a new jumps
100.0% applications/smurf/libsmf/
Commits from repository starlink
-----------------------------
commit 69c49eec09680db948b6599f8505c674c75e8a30
Author: Tim Jenness <[log in to unmask]>
Date: Fri May 28 15:42:20 2010 -1000
smurf: Add SMO smoothing model to iterative map-maker
SMO == smoothing model
Add boxcar smoothing via smf_boxcar1D.
Not really tested and bound to have problems with quality flags.
applications/smurf/defaults/smurf_makemap.def | 5 +
applications/smurf/libsmf/Makefile.am | 1 +
applications/smurf/libsmf/smf.h.source | 4 +
.../{smf_calcmodel_pln.c => smf_calcmodel_smo.c} | 86 +++++++++++++-----
applications/smurf/libsmf/smf_checkmem_dimm.c | 6 ++
applications/smurf/libsmf/smf_model_create.c | 16 ++++
applications/smurf/libsmf/smf_model_createHdr.c | 6 ++
applications/smurf/libsmf/smf_model_getname.c | 6 ++
applications/smurf/libsmf/smf_model_getptr.c | 6 ++
applications/smurf/libsmf/smf_model_gettype.c | 4 +
applications/smurf/libsmf/smf_typ.h | 5 +-
11 files changed, 123 insertions(+), 22 deletions(-)
commit af6e143cd301fdd2206f14da7dc1cc7fc68c668d
Author: Tim Jenness <[log in to unmask]>
Date: Fri May 28 15:41:11 2010 -1000
smurf: Check the bits in PLN and not multiply them
One of the if clauses checked the return result of multiplying
a quality with a bit rather than checking the logical AND.
applications/smurf/libsmf/smf_calcmodel_pln.c | 4 ++--
1 files changed, 2 insertions(+), 2 deletions(-)
commit 5cc9ace90da9ad5749c835366e29b4c92cbde67d
Author: Tim Jenness <[log in to unmask]>
Date: Fri May 28 15:39:01 2010 -1000
smurf: Assigning a negative number to a size_t does not work
"off" was a size_t so in this code below:
if( i < (window-1) ) off = -i/2;
else off = -window/2;
"off" was becoming very large and walking off the end of the array. I
have no real idea how boxcar smoothing was even working. "off" is
now a long and the above code changed to:
if( i < (window-1) ) off = -1*(long)i/2;
else off = -1*(long)window/2;
applications/smurf/libsmf/smf_boxcar1.cgen | 16 ++++++++++------
1 files changed, 10 insertions(+), 6 deletions(-)
commit 3b3dcde9ea46377c863068f11e375525bb15a534
Author: Ed Chapin <[log in to unmask]>
Date: Fri May 28 17:23:20 2010 -0700
smurf: some test code for alternative sample variance calc in maps
Per was looking into how the map sample variance is calculated. In
order to do his version we need to keep track of an additional sum,
the total of the mapweights^2 which requires an additional map-sized
buffer in smf_iteratemap and an extra parameter to smf_rebinmap1. To
turn this alternative method on, uncomment the line at the top of
smf_rebinmap1 that defines __SMF_REBINMAP__SAMPLE_STANDARD_DEVIATION.
applications/smurf/libsmf/smf.h.source | 3 +-
applications/smurf/libsmf/smf_calcmodel_ast.c | 3 ++
applications/smurf/libsmf/smf_iteratemap.c | 26 +++++++++++---
applications/smurf/libsmf/smf_rebinmap1.c | 42 +++++++++++++++++++---
applications/smurf/libsmf/smf_typ.h | 1 +
5 files changed, 62 insertions(+), 13 deletions(-)
commit 85e92e9077236ba65864c629d88c47d5dca6a20c
Author: Ed Chapin <[log in to unmask]>
Date: Fri May 28 12:55:08 2010 -0700
smurf: clean up needlessly complicated init in smf_model_create
Tim pointed out that we don't need thr ROW/COL index macros here...
applications/smurf/libsmf/smf_model_create.c | 24 ++++++++----------------
1 files changed, 8 insertions(+), 16 deletions(-)
commit 9adf5a838cfff7e595ac900ccea01887d37be1e6
Author: Ed Chapin <[log in to unmask]>
Date: Fri May 28 11:31:39 2010 -0700
smurf: replacing dead DKS with average now optional (dks.replacebad)
applications/smurf/defaults/smurf_makemap.def | 1 +
applications/smurf/examples/dimmconfig.lis | 3 +++
applications/smurf/libsmf/smf_clean_dksquid.c | 12 +++++++++++-
applications/smurf/libsmf/smf_model_create.c | 13 +++++++++++--
4 files changed, 26 insertions(+), 3 deletions(-)
commit 41718d3c2a2d4e2910399d3126598b3a16c824c2
Author: David Berry <[log in to unmask]>
Date: Fri May 28 13:10:09 2010 +0100
smurf: heavier clipping when finding RMS gradient in dc step correction
Clip at 2 sigma (not 3) when finding the RMS gradient, and
then correct for the heavy clipping using the factor appropriate
for pure Gaussian noise. heavier clipping does better in the
presence of lots of steps.
applications/smurf/libsmf/smf_fix_steps.c | 13 ++++++++++---
1 files changed, 10 insertions(+), 3 deletions(-)
commit 0ff6ed13d3b6584f620bcdb95ffbe052302d7366
Author: David Berry <[log in to unmask]>
Date: Fri May 28 12:52:29 2010 +0100
smurf: Fix bad quality indexing in step correction
This looks better... Fix a problem indexing the quality array when steps
are found close to the start or end of the time series.
applications/smurf/libsmf/smf_fix_steps.c | 18 ++++++++++--------
1 files changed, 10 insertions(+), 8 deletions(-)
commit 4c72d45f9a69e5fcf1f0508b6656cf1a569ff02c
Author: David Berry <[log in to unmask]>
Date: Fri May 28 12:32:23 2010 +0100
Revert "smurf: Correct indexing of quality array when DC steps found close
to start or end"
This reverts commit fca589ac61bcbe47872c07c529e9fc18b3e58871, which was
done in a rush prior to going on holiday and seems to break things.
applications/smurf/libsmf/smf_fix_steps.c | 47 ++++++++++++--------------
1 files changed, 22 insertions(+), 25 deletions(-)
commit fca589ac61bcbe47872c07c529e9fc18b3e58871
Author: David Berry <[log in to unmask]>
Date: Fri May 28 12:17:54 2010 +0100
smurf: Correct indexing of quality array when DC steps found close to start or end
applications/smurf/libsmf/smf_fix_steps.c | 47 ++++++++++++++------------
1 files changed, 25 insertions(+), 22 deletions(-)
commit 01b6e502b5c1e1c38bb88f296986e484db7b9df9
Author: David Berry <[log in to unmask]>
Date: Fri May 28 10:48:59 2010 +0100
smurf: Ignore previously flagged jumps when fitting data to a new jumps
Previously, the least squares fits that are used to determine the data
levels before and after a correlated DC step were including data that had
previously been flagged as a jump (i.e. the samples where the DC level is
moving from the lower level to the higher level). This could cause some
steps that have previously been fixed to be re-introduced (with a
different step height).
applications/smurf/libsmf/smf_fix_steps.c | 9 ++++++---
1 files changed, 6 insertions(+), 3 deletions(-)
|