def main(opt, ps):
#ralo = 36
#rahi = 42
#declo = -1.25
#dechi = 1.25
#width = 7
ralo = 37.5
rahi = 41.5
declo = -1.5
dechi = 2.5
width = 2.5
rl,rh = 39,40
dl,dh = 0,1
roipoly = np.array([(rl,dl),(rl,dh),(rh,dh),(rh,dl)])
ra = (ralo + rahi ) / 2.
dec = (declo + dechi) / 2.
bandnum = 1
band = 'w%i' % bandnum
plt.figure(figsize=(12,12))
#basedir = '/project/projectdirs/bigboss'
#wisedatadir = os.path.join(basedir, 'data', 'wise')
wisedatadirs = ['/clusterfs/riemann/raid007/bosswork/boss/wise_level1b',
'/clusterfs/riemann/raid000/bosswork/boss/wise1ext']
wisecatdir = '/home/boss/products/NULL/wise/trunk/fits/'
ofn = 'wise-images-overlapping.fits'
if os.path.exists(ofn):
print 'File exists:', ofn
T = fits_table(ofn)
print 'Found', len(T), 'images overlapping'
print 'Reading WCS headers...'
wcses = []
T.filename = [fn.strip() for fn in T.filename]
for fn in T.filename:
wcs = anwcs(fn, 0)
wcses.append(wcs)
else:
TT = []
for d in wisedatadirs:
ifn = os.path.join(d, 'WISE-index-L1b.fits') #'index-allsky-astr-L1b.fits')
T = fits_table(ifn, columns=['ra','dec','scan_id','frame_num'])
print 'Read', len(T), 'from WISE index', ifn
I = np.flatnonzero((T.ra > ralo) * (T.ra < rahi) * (T.dec > declo) * (T.dec < dechi))
print len(I), 'overlap RA,Dec box'
T.cut(I)
fns = []
for sid,fnum in zip(T.scan_id, T.frame_num):
print 'scan,frame', sid, fnum
fn = get_l1b_file(d, sid, fnum, bandnum)
print '-->', fn
assert(os.path.exists(fn))
fns.append(fn)
T.filename = np.array(fns)
TT.append(T)
T = merge_tables(TT)
wcses = []
corners = []
ii = []
for i in range(len(T)):
wcs = anwcs(T.filename[i], 0)
W,H = wcs.get_width(), wcs.get_height()
rd = []
for x,y in [(1,1),(1,H),(W,H),(W,1)]:
rd.append(wcs.pixelxy2radec(x,y))
rd = np.array(rd)
if polygons_intersect(roipoly, rd):
wcses.append(wcs)
corners.append(rd)
ii.append(i)
print 'Found', len(wcses), 'overlapping'
I = np.array(ii)
T.cut(I)
outlines = corners
corners = np.vstack(corners)
nin = sum([1 if point_in_poly(ra,dec,ol) else 0 for ol in outlines])
print 'Number of images containing RA,Dec,', ra,dec, 'is', nin
r0,r1 = corners[:,0].min(), corners[:,0].max()
d0,d1 = corners[:,1].min(), corners[:,1].max()
print 'RA,Dec extent', r0,r1, d0,d1
T.writeto(ofn)
print 'Wrote', ofn
# MAGIC 2.75: approximate pixel scale, "/pix
S = int(3600. / 2.75)
print 'Coadd size', S
cowcs = anwcs_create_box(ra, dec, 1., S, S)
if False:
print 'Plotting map...'
plot = Plotstuff(outformat='png', ra=ra, dec=dec, width=width, size=(800,800))
out = plot.outline
plot.color = 'white'
plot.alpha = 0.07
plot.apply_settings()
for wcs in wcses:
out.wcs = wcs
out.fill = False
plot.plot('outline')
out.fill = True
plot.plot('outline')
plot.color = 'gray'
plot.alpha = 1.0
plot.lw = 1
plot.plot_grid(1, 1, 1, 1)
plot.color = 'red'
plot.lw = 3
plot.alpha = 0.75
out.wcs = cowcs
out.fill = False
plot.plot('outline')
if opt.sources:
rd = plot.radec
plot_radec_set_filename(rd, opt.sources)
plot.plot('radec')
pfn = ps.getnext()
plot.write(pfn)
print 'Wrote', pfn
# Re-sort by distance to RA,Dec center...
#I = np.argsort(np.hypot(T.ra - ra, T.dec - dec))
#T.cut(I)
# IF YOU DO THIS, MUST ALSO RE-SORT 'wcses'!
if opt.sources:
# Look at a radius this big, in arcsec, around each source position.
# 15" = about 6 WISE pixels
Wrad = 15. / 3600.
# Look for SDSS objects within this radius; Wrad + a margin
Srad = Wrad + 5./3600.
S = fits_table(opt.sources)
print 'Read', len(S), 'sources from', opt.sources
groups,singles = cluster_radec(S.ra, S.dec, Wrad, singles=True)
print 'Source clusters:', groups
print 'Singletons:', singles
tractors = []
sdss = DR9(basedir='data-dr9')
sband = 'r'
for i in singles:
r,d = S.ra[i],S.dec[i]
print 'Source', i, 'at', r,d
fn = sdss.retrieve('photoObj', S.run[i], S.camcol[i], S.field[i], band=sband)
print 'Reading', fn
oo = fits_table(fn)
print 'Got', len(oo)
cat1,obj1,I = get_tractor_sources_dr9(None, None, None, bandname=sband,
objs=oo, radecrad=(r,d,Srad), bands=[],
nanomaggies=True, extrabands=[band],
fixedComposites=True,
getobjs=True, getobjinds=True)
print 'Got', len(cat1), 'SDSS sources nearby'
# Find images that overlap?
ims = []
for j,wcs in enumerate(wcses):
print 'Filename', T.filename[j]
ok,x,y = wcs.radec2pixelxy(r,d)
print 'WCS', j, '-> x,y:', x,y
if not anwcs_radec_is_inside_image(wcs, r, d):
continue
tim = wise.read_wise_level1b(
T.filename[j].replace('-int-1b.fits',''),
nanomaggies=True, mask_gz=True, unc_gz=True,
sipwcs=True, constantInvvar=True, radecrad=(r,d,Wrad))
ims.append(tim)
print 'Found', len(ims), 'images containing this source'
tr = Tractor(ims, cat1)
tractors.append(tr)
if len(groups):
# TODO!
assert(False)
sys.exit(0)
# Find additional SDSS sources nearby = within R pixels radius.
R = 30.
#R = 50.
rad = R * 0.396 / 3600.
cats = []
objs = []
for run,camcol,field,r,d in zip(S.run, S.camcol, S.field, S.ra, S.dec):
fn = sdss.retrieve('photoObj', run, camcol, field, band=sband)
print 'Reading', fn
oo = fits_table(fn)
print 'Got', len(oo)
cat1,obj1,I = get_tractor_sources_dr9(None, None, None, bandname=sband,
objs=oo, radecrad=(r,d,rad), bands=[],
nanomaggies=True, extrabands=[band],
fixedComposites=True,
getobjs=True, getobjinds=True)
print 'Got', len(cat1), 'SDSS sources nearby'
cats.append(cat1)
objs.append(obj1[I])
# Merge into one big catalog.
cat = Catalog()
for c in cats:
for src in c:
cat.append(src)
S = merge_tables(objs)
print 'Merged catalog has', len(cat), 'entries'
print 'S table has', len(S)
assert(len(S) == len(cat))
if opt.ptsrc:
print 'Converting all sources to PointSources'
pcat = Catalog()
for src in cat:
ps = PointSource(src.getPosition(), src.getBrightness())
pcat.append(ps)
print 'PointSource catalog:', pcat
cat = pcat
# ??
WW = S
#WW = tabledata()
# cat = get_tractor_sources_dr9(None, None, None, bandname=sband,
# objs=S, bands=[], nanomaggies=True,
# extrabands=[band])
print 'Got', len(cat), 'tractor sources'
#cat = Catalog(*cat)
print cat
for src in cat:
print ' ', src
### FIXME -- match to WISE catalog to initialize mags?
# Initialize WISE mags to be at least detectable
# so that we identify the right pixel ROIs below.
#minbright = NanoMaggies.magToNanomaggies()
#minbright = 50.
minbright = 250.
cat.freezeParamsRecursive('*')
cat.thawPathsTo(band)
p0 = cat.getParams()
cat.setParams(np.maximum(minbright, p0))
print 'Set minimum W1 brightness:'
for src in cat:
print ' ', src
# Cut images that don't overlap.
ii = []
for i,wcs in enumerate(wcses):
isin = False
for r,d in zip(S.ra, S.dec):
if anwcs_radec_is_inside_image(wcs, r, d):
isin = True
break
if isin:
ii.append(i)
T.cut(np.array(ii))
print 'Cut to', len(T), 'images containing sources'
else:
wfn = 'wise-sources-nearby.fits'
if os.path.exists(wfn):
print 'Reading existing file', wfn
W = fits_table(wfn)
print 'Got', len(W), 'with range RA', W.ra.min(), W.ra.max(), ', Dec', W.dec.min(), W.dec.max()
else:
# Range of WISE slices (inclusive) containing this Dec range.
ws0, ws1 = 26,27
WW = []
for w in range(ws0, ws1+1):
fn = os.path.join(wisecatdir, 'wise-allsky-cat-part%02i-radec.fits' % w)
print 'Searching for sources in', fn
W = fits_table(fn)
I = np.flatnonzero((W.ra >= r0) * (W.ra <= r1) * (W.dec >= d0) * (W.dec <= d1))
fn = os.path.join(wisecatdir, 'wise-allsky-cat-part%02i.fits' % w)
print 'Reading', len(I), 'rows from', fn
W = fits_table(fn, rows=I)
print 'Cut to', len(W), 'sources in range'
WW.append(W)
W = merge_tables(WW)
del WW
print 'Total of', len(W)
W.writeto(wfn)
print 'wrote', wfn
# DEBUG
W.cut((W.ra >= rl) * (W.ra <= rh) * (W.dec >= dl) * (W.dec <= dh))
print 'Cut to', len(W), 'in the central region'
print 'Creating', len(W), 'Tractor sources'
cat = Catalog()
for i in range(len(W)):
w1 = W.w1mpro[i]
nm = NanoMaggies.magToNanomaggies(w1)
cat.append(PointSource(RaDecPos(W.ra[i], W.dec[i]), NanoMaggies(w1=nm)))
WW = W
cat.freezeParamsRecursive('*')
cat.thawPathsTo(band)
cat0 = cat.getParams()
br0 = [src.getBrightness().copy() for src in cat]
nm0 = np.array([b.getBand(band) for b in br0])
WW.nm0 = nm0
w1psf = wise.get_psf_model(bandnum, opt.pixpsf)
# Create fake image in the "coadd" footprint in order to find overlapping
# sources.
H,W = int(cowcs.imageh), int(cowcs.imagew)
# MAGIC -- sigma a bit smaller than typical images (4.0-ish)
sig = 3.5
# typical zeropoint
zp = 20.752
faketim = Image(data=np.zeros((H,W), np.float32),
invvar=np.zeros((H,W), np.float32) + (1./sig**2),
psf=w1psf, wcs=ConstantFitsWcs(cowcs), sky=ConstantSky(0.),
photocal = LinearPhotoCal(NanoMaggies.zeropointToScale(zp),
band=band),
#photocal=LinearPhotoCal(1., band=band),
name='fake')
minsb = 0.1 * sig
#minsb = 0.
# pc = faketim.getPhotoCal()
# print 'Source counts:'
# for src in cat:
# print ' ', src
# print '-->', pc.brightnessToCounts(src.getBrightness())
# print ' -->', [pc.brightnessToCounts(br) for br in src.getBrightnesses()]
# print 'Source pixel positions:'
# wcs = faketim.getWcs()
# for src in cat:
# print ' ', src
# print '--> x,y', wcs.positionToPixel(src.getPosition())
print 'Finding overlapping sources...'
t0 = Time()
tractor = Tractor([faketim], cat)
groups,L,fakemod = tractor.getOverlappingSources(0, minsb=minsb)
print 'Overlapping sources took', Time()-t0
print 'Got', len(groups), 'groups of sources'
nl = L.max()
gslices = find_objects(L, nl)
print 'unique labels:', np.unique(L)
# plt.clf()
# plt.imshow(fakemod, interpolation='nearest', origin='lower',
# vmin=0, vmax=sig*3.)
# plt.title('Fakemod')
# ps.savefig()
#
# for IM in [L, (L>0)]:
# plt.clf()
# plt.imshow(IM, interpolation='nearest', origin='lower')
# plt.gray()
# wcs = faketim.getWcs()
# xy = []
# for src in cat:
# x,y = wcs.positionToPixel(src.getPosition())
# xy.append((x,y))
# xy = np.array(xy)
# ax = plt.axis()
# plt.plot(xy[:,0], xy[:,1], 'r+')
# plt.title('Source groups')
# ps.savefig()
# Find sources touching each group's (rectangular) ROI
tgroups = {}
for i,gslice in enumerate(gslices):
gl = i+1
tg = np.unique(L[gslice])
tsrcs = []
for g in tg:
if not g in [gl,0]:
if g in groups:
tsrcs.extend(groups[g])
tgroups[gl] = tsrcs
# for i,gslice in enumerate(gslices):
# if not (i+1) in groups:
# continue
#
# plt.clf()
# plt.imshow(IM[gslice], interpolation='nearest', origin='lower')
# plt.gray()
# wcs = faketim.getWcs()
# xy = []
# y0,x0 = gslice[0].start, gslice[1].start
# for src in cat:
# x,y = wcs.positionToPixel(src.getPosition())
# xy.append((x-x0,y-y0))
# xy = np.array(xy)
#
# ax = plt.axis()
#
# plt.plot(xy[:,0], xy[:,1], 'r+')
#
# I = np.array(groups[i+1])
# if len(I):
# plt.plot(xy[I,0], xy[I,1], 'g.')
#
# I = np.array(tgroups[i+1])
# if len(I):
# plt.plot(xy[I,0], xy[I,1], 'gx')
#
# ps.savefig()
#
# plt.axis(ax)
# ps.savefig()
print 'Group size histogram:'
ng = Counter()
for g in groups.values():
ng[len(g)] += 1
kk = ng.keys()
kk.sort()
for k in kk:
print ' ', k, 'sources:', ng[k], 'groups'
nms = []
tims = []
allrois = {}
badrois = {}
if opt.threads:
mp = multiproc(opt.threads)
else:
mp = multiproc(1)
tims = mp.map(_read_l1b, T.filename)
for imi,tim in enumerate(tims):
tim.psf = w1psf
H,W = tim.shape
nin = 0
for src in cat:
x,y = tim.getWcs().positionToPixel(src.getPosition())
if x >= 0 and y >= 0 and x < W and y < H:
nin += 1
print 'Number of sources inside image:', nin
tractor = Tractor([tim], cat)
tractor.freezeParam('images')
### ??
cat.setParams(cat0)
pgroups = 0
pobjs = 0
for gi in range(len(gslices)):
gl = gi
# note, gslices is zero-indexed
gslice = gslices[gl]
gl += 1
if not gl in groups:
print 'Group', gl, 'not in groups array; skipping'
continue
gsrcs = groups[gl]
tsrcs = tgroups[gl]
# print 'Group number', (gi+1), 'of', len(Gorder), ', id', gl, ': sources', gsrcs
# print 'sources in groups touching slice:', tsrcs
# Convert from 'canonical' ROI to this image.
yl,yh = gslice[0].start, gslice[0].stop
xl,xh = gslice[1].start, gslice[1].stop
x0,y0 = W-1,H-1
x1,y1 = 0,0
for x,y in [(xl,yl),(xh-1,yl),(xh-1,yh-1),(xl,yh-1)]:
r,d = cowcs.pixelxy2radec(x+1, y+1)
x,y = tim.getWcs().positionToPixel(RaDecPos(r,d))
x = int(np.round(x))
y = int(np.round(y))
x = np.clip(x, 0, W-1)
y = np.clip(y, 0, H-1)
x0 = min(x0, x)
y0 = min(y0, y)
x1 = max(x1, x)
y1 = max(y1, y)
if x1 == x0 or y1 == y0:
print 'Gslice', gslice, 'is completely outside this image'
continue
gslice = (slice(y0,y1+1), slice(x0, x1+1))
if np.all(tim.getInvError()[gslice] == 0):
print 'This whole object group has invvar = 0.'
if not gl in badrois:
badrois[gl] = {}
badrois[gl][imi] = gslice
continue
if not gl in allrois:
allrois[gl] = {}
allrois[gl][imi] = gslice
if not opt.individual:
continue
fullcat = tractor.catalog
subcat = Catalog(*[fullcat[i] for i in gsrcs + tsrcs])
for i in range(len(tsrcs)):
subcat.freezeParam(len(gsrcs) + i)
tractor.catalog = subcat
print len(gsrcs), 'sources unfrozen; total', len(subcat)
pgroups += 1
pobjs += len(gsrcs)
t0 = Time()
tractor.optimize_forced_photometry(minsb=minsb, mindlnp=1.,
rois=[gslice])
print 'optimize_forced_photometry took', Time()-t0
tractor.catalog = fullcat
# mod = tractor.getModelImage(0, minsb=minsb)
# noise = np.random.normal(size=mod.shape)
# noise[tim.getInvError() == 0] = 0.
# nz = (tim.getInvError() > 0)
# noise[nz] *= (1./tim.getInvError()[nz])
# ima = dict(interpolation='nearest', origin='lower',
# vmin=tim.zr[0], vmax=tim.zr[1])
# imchi = dict(interpolation='nearest', origin='lower',
# vmin=-5, vmax=5)
# plt.clf()
# plt.subplot(2,2,1)
# plt.imshow(tim.getImage(), **ima)
# plt.gray()
# plt.subplot(2,2,2)
# plt.imshow(mod, **ima)
# plt.gray()
# plt.subplot(2,2,3)
# plt.imshow((tim.getImage() - mod) * tim.getInvError(), **imchi)
# plt.gray()
# plt.subplot(2,2,4)
# plt.imshow(mod + noise, **ima)
# plt.gray()
# plt.suptitle('W1, scan %s, frame %i' % (sid, fnum))
# ps.savefig()
if opt.individual:
print 'Photometered', pgroups, 'groups containing', pobjs, 'objects'
cat.thawPathsTo(band)
nm1 = np.array([src.getBrightness().getBand(band) for src in cat])
nms.append(nm1)
WW.nms = np.array(nms).T
fn = opt.output % imi
WW.writeto(fn)
print 'Wrote', fn
return dict(cat0=cat0, WW=WW, band=band, tims=tims,
allrois=allrois, badrois=badrois, groups=groups,
tgroups=tgroups, minsb=minsb,
gslices=gslices, cat=cat)
def simult_photom(cat0=None, WW=None, band=None, tims=None,
allrois=None, badrois=None, groups=None,
tgroups=None, minsb=None,
gslices=None, cat=None,
opt=None, ps=None):
def _plot_grid(ims, kwas):
N = len(ims)
C = int(np.ceil(np.sqrt(N)))
R = int(np.ceil(N / float(C)))
plt.clf()
for i,(im,kwa) in enumerate(zip(ims, kwas)):
plt.subplot(R,C, i+1)
#print 'plotting grid cell', i, 'img shape', im.shape
plt.imshow(im, **kwa)
plt.gray()
plt.xticks([]); plt.yticks([])
return R,C
def _plot_grid2(ims, cat, tims, kwas, ptype='mod'):
xys = []
stamps = []
for (img,mod,chi,roi),tim in zip(ims, tims):
if ptype == 'mod':
stamps.append(mod)
elif ptype == 'chi':
stamps.append(chi)
wcs = tim.getWcs()
y0,x0 = roi[0].start, roi[1].start
xy = []
for src in cat:
xi,yi = wcs.positionToPixel(src.getPosition())
xy.append((xi - x0, yi - y0))
xys.append(xy)
#print 'X,Y source positions in stamp of shape', stamps[-1].shape
#print ' ', xy
R,C = _plot_grid(stamps, kwas)
for i,xy in enumerate(xys):
plt.subplot(R, C, i+1)
ax = plt.axis()
xy = np.array(xy)
plt.plot(xy[:,0], xy[:,1], 'r+', lw=2)
plt.axis(ax)
# Simultaneous photometry
if opt.osources:
O = fits_table(opt.osources)
ocat = Catalog()
print 'Other catalog:'
for i in range(len(O)):
w1 = O.wiseflux[i, 0]
s = PointSource(RaDecPos(O.ra[i], O.dec[i]), NanoMaggies(w1=w1))
ocat.append(s)
print ocat
ocat.freezeParamsRecursive('*')
ocat.thawPathsTo(band)
# Keep track of params after simultaneous photometry...
cat.freezeParamsRecursive('*')
cat.thawPathsTo(band)
catsim = cat.getParams()
if opt.opt:
# ... and also after RA,Dec opt.
cat.thawPathsTo('ra','dec')
catopt = cat.getParams()
cat.freezeParamsRecursive('*')
cat.thawPathsTo(band)
for gi in range(len(gslices)):
gl = gi
gl += 1
if not gl in groups:
print 'Group', gl, 'not in groups array; skipping'
continue
gsrcs = groups[gl]
tsrcs = tgroups[gl]
print 'Group', gl
print 'gsrcs:', gsrcs
print 'tsrcs:', tsrcs
if (not gl in allrois) and (not gl in badrois):
print 'Group', gl, 'does not touch any images?'
continue
mytims = []
rois = []
if gl in allrois:
for imi,roi in allrois[gl].items():
mytims.append(tims[imi])
rois.append(roi)
mybadtims = []
mybadrois = []
if gl in badrois:
for imi,roi in badrois[gl].items():
mybadtims.append(tims[imi])
mybadrois.append(roi)
print 'Group', gl, 'touches', len(mytims), 'images and', len(mybadtims), 'bad ones'
tt = 'group %i: %i+%i sources' % (gl, len(gsrcs), len(tsrcs))
if len(mytims):
cat.setParams(catsim)
#print 'Restoring catsim:'
#cat.printThawedParams()
subcat = Catalog(*[cat[i] for i in gsrcs + tsrcs])
for i in range(len(tsrcs)):
subcat.freezeParam(len(gsrcs) + i)
tractor = Tractor(mytims, subcat)
tractor.freezeParam('images')
print len(gsrcs), 'sources unfrozen; total', len(subcat)
print 'Before fitting:'
for src in subcat[:len(gsrcs)]:
print ' ', src
t0 = Time()
ims0,ims1 = tractor.optimize_forced_photometry(minsb=minsb, mindlnp=1.,
rois=rois)
print 'optimize_forced_photometry took', Time()-t0
print 'After fitting:'
for src in subcat[:len(gsrcs)]:
print ' ', src
imas = [dict(interpolation='nearest', origin='lower',
vmin=tim.zr[0], vmax=tim.zr[1])
for tim in mytims]
imchi = dict(interpolation='nearest', origin='lower', vmin=-5, vmax=5)
imchis = [imchi] * len(mytims)
_plot_grid([img for (img, mod, chi, roi) in ims0], imas)
plt.suptitle('Data: ' + tt)
ps.savefig()
if ims1 is not None:
#_plot_grid([mod for (img, mod, chi, roi) in ims1], imas)
_plot_grid2(ims1, subcat, mytims, imas)
plt.suptitle('Forced-phot model: ' + tt)
ps.savefig()
#_plot_grid([chi for (img, mod, chi, roi) in ims1], imchis)
_plot_grid2(ims1, subcat, mytims, imchis, ptype='chi')
plt.suptitle('Forced-phot chi: ' + tt)
ps.savefig()
if opt.osources:
cc = tractor.catalog
tractor.catalog = ocat
nil,nil,ims3 = tractor.optimize_forced_photometry(minsb=minsb, rois=rois,
justims0=True)
tractor.catalog = cc
_plot_grid2(ims3, ocat, mytims, imas)
plt.suptitle("Schlegel's model: group %i" % gl)
ps.savefig()
_plot_grid2(ims3, ocat, mytims, imchis, ptype='chi')
plt.suptitle("Schlegel's chi: group %i" % gl)
ps.savefig()
if opt.opt:
op1 = ps.getnext()
op2 = ps.getnext()
#fits[gl] = (tractor, len(gsrcs), rois, op1, op2)
# print 'Plotting mods after simul photom'
# #_plot_grid([mod for (img, mod, chi, roi) in ims0], imas)
# _plot_grid2(ims0, subcat, mytims, imas)
# plt.suptitle('Initial model: ' + tt)
# ps.savefig()
#
# print 'Plotting chis after simul photom'
# #_plot_grid([chi for (img, mod, chi, roi) in ims0], imchis)
# _plot_grid2(ims0, subcat, mytims, imchis, ptype='chi')
# plt.suptitle('Initial chi: ' + tt)
# ps.savefig()
print 'After simultaneous photometry:'
subcat.printThawedParams()
# Copy updated params to "catsim"
catsim = cat.getParams()
#print 'Saving catsim:'
#cat.printThawedParams()
cat.freezeParamsRecursive('*')
cat.thawPathsTo(band)
WW.nmall = np.array([src.getBrightness().getBand(band) for src in cat])
if len(mytims) and opt.opt:
print 'Optimizing RA,Dec'
subcat = tractor.catalog
# Copy updated forced-phot params from catsim to catopt.
#print 'Saving subcat forced-phot params:'
#subcat.printThawedParams()
fphot = subcat.getParams()
cat.thawPathsTo('ra','dec')
cat.setParams(catopt)
#print 'Copying forced-phot results to catopt:'
cat.freezeParamsRecursive('*')
cat.thawPathsTo(band)
cat.freezeAllBut(*gsrcs)
#cat.printThawedParams()
NP = cat.numberOfParams()
cat.setParams(fphot[:NP])
#print 'Result:'
#print 'Restoring catopt:'
#cat.printThawedParams()
cat.freezeParamsRecursive('*')
cat.thawPathsTo(band)
NG = len(gsrcs)
for i in range(NG):
subcat[i].thawPathsTo('ra','dec')
p0 = subcat.getParams()
print 'Optimizing params:'
subcat.printThawedParams()
thetims = tractor.images
subimgs = []
for i,img in enumerate(thetims):
roi = rois[i]
y0 = roi[0].start
x0 = roi[1].start
subwcs = ShiftedWcs(img.wcs, x0, y0)
subimg = Image(data=img.data[roi], invvar=img.invvar[roi],
psf=img.psf, wcs=subwcs, sky=img.sky,
photocal=img.photocal, name=img.name)
subimgs.append(subimg)
tractor.images = Images(*subimgs)
while True:
dlnp,X,alpha = tractor.optimize()
print 'dlnp', dlnp
print 'alpha', alpha
if dlnp < 0.1:
break
p1 = subcat.getParams()
print 'Param changes:'
for nm,pp0,pp1 in zip(subcat.getParamNames(), p0, p1):
print ' ', nm, pp0, 'to', pp1, '; delta', pp1-pp0
cat.thawPathsTo('ra','dec')
catopt = cat.getParams()
print 'Saving catopt:'
cat.printThawedParams()
cat.freezeParamsRecursive('ra', 'dec')
tractor.images = thetims
nil,nil,ims2 = tractor.optimize_forced_photometry(minsb=minsb, rois=rois,
justims0=True)
print 'Plotting mods after RA,Dec opt'
#_plot_grid([mod for (img, mod, chi, roi) in ims2], imas)
_plot_grid2(ims2, subcat, mytims, imas)
plt.suptitle('RA,Dec-opt model: ' + tt)
plt.savefig(op1)
print 'Plotting chis after RA,Dec opt'
#_plot_grid([chi for (img, mod, chi, roi) in ims2], imchis)
_plot_grid2(ims2, subcat, mytims, imchis, ptype='chi')
plt.suptitle('RA,Dec-opt chi: ' + tt)
plt.savefig(op2)
N = len(mybadtims)
if N and False:
C = int(np.ceil(np.sqrt(N)))
R = int(np.ceil(N / float(C)))
plt.clf()
for i,(tim,roi) in enumerate(zip(mybadtims, mybadrois)):
plt.subplot(R,C, i+1)
plt.imshow(tim.getImage()[roi], interpolation='nearest', origin='lower',
vmin=tim.zr[0], vmax=tim.zr[1])
plt.gray()
plt.suptitle('Data in bad regions')
ps.savefig()
plt.clf()
for i,(tim,roi) in enumerate(zip(mybadtims, mybadrois)):
plt.subplot(R,C, i+1)
plt.imshow(tim.getInvError()[roi], interpolation='nearest', origin='lower')
plt.gray()
plt.suptitle('Inverr in bad regions')
ps.savefig()
if gi == 0 and opt.plotmask:
alltims = mybadtims+mytims
_plot_grid([tim.uncplane[roi] for tim,roi in zip(alltims,
mybadrois + rois)],
[dict(interpolation='nearest', origin='lower')]*len(alltims))
plt.suptitle('Uncertainty plane')
ps.savefig()
for bit,txt in [
(0 , 'static: excessively noisy due to high dark current alone'),
(1 , 'static: generally noisy [includes bit 0]'),
(2 , 'static: dead or very low responsivity'),
(3 , 'static: low responsivity or low dark current'),
(4 , 'static: high responsivity or high dark current'),
(5 , 'static: saturated anywhere in ramp'),
(6 , 'static: high, uncertain, or unreliable non-linearity'),
(7 , 'static: known broken hardware pixel or excessively noisy responsivity estimate [may include bit 1]'),
(9 , 'broken pixel or negative slope fit value'),
(10, 'saturated in sample read 1'),
(11, 'saturated in sample read 2'),
(12, 'saturated in sample read 3'),
(13, 'saturated in sample read 4'),
(14, 'saturated in sample read 5'),
(15, 'saturated in sample read 6'),
(16, 'saturated in sample read 7'),
(17, 'saturated in sample read 8'),
(18, 'saturated in sample read 9'),
(21, 'new/transient bad pixel from dynamic masking'),
(26, 'non-linearity correction unreliable'),
(27, 'contains cosmic-ray or outlier that cannot be classified (from temporal outlier rejection in multi-frame pipeline)'),
(28, 'contains positive or negative spike-outlier'),
]:
_plot_grid([tim.maskplane[roi] & (1 << bit)
for tim,roi in zip(alltims, mybadrois + rois)],
[dict(interpolation='nearest', origin='lower',
vmin=0, vmax=1)]*len(alltims))
plt.suptitle('Mask: ' + txt)
ps.savefig()
if opt.opt:
fn = opt.output % 998
WW.writeto(fn)
print 'Wrote', fn
cat.thawPathsTo('ra','dec')
cat.setParams(catopt)
WW.nmoptrd = np.array([src.getBrightness().getBand(band) for src in cat])
cat.freezeParamsRecursive(band, 'dec')
WW.raoptrd = np.array(cat.getParams())
cat.freezeParamsRecursive('ra')
cat.thawPathsTo('dec')
WW.decoptrd = np.array(cat.getParams())
cat.freezeParamsRecursive('dec')
fn = opt.output % 999
WW.writeto(fn)
print 'Wrote', fn
def _meisner_psf_models():
global plotslice
# Meisner's PSF models
# for band in [4]:
for band in [1, 2, 3, 4]:
print()
print('W%i' % band)
print()
#pix = fitsio.read('wise-psf-avg-pix.fits', ext=band-1)
pix = fitsio.read('wise-psf-avg-pix-bright.fits', ext=band - 1)
fit = fits_table('wise-psf-avg.fits', hdu=band)
#fit = fits_table('psf-allwise-con3.fits', hdu=band)
scale = 1.
print('Pix shape', pix.shape)
h, w = pix.shape
xx, yy = np.meshgrid(np.arange(w), np.arange(h))
cx, cy = np.sum(xx * pix) / np.sum(pix), np.sum(yy * pix) / np.sum(pix)
print('Centroid:', cx, cy)
#S = 100
S = h / 2
slc = slice(h / 2 - S, h / 2 + S + 1), slice(w / 2 - S, w / 2 + S + 1)
plotss = 30
plotslice = slice(S - plotss,
-(S - plotss)), slice(S - plotss, -(S - plotss))
opix = pix
print('Original pixels sum:', opix.sum())
pix /= pix.sum()
pix = pix[slc]
print('Sliced pix sum', pix.sum())
psf = GaussianMixturePSF(fit.amp, fit.mean * scale, fit.var * scale**2)
psfmodel = psf.getPointSourcePatch(0., 0., radius=h / 2)
mod = psfmodel.patch
print('Orig mod sum', mod.sum())
mod = mod[slc]
print('Sliced mod sum', mod.sum())
print('Amps:', np.sum(psf.mog.amp))
_plot_psf(pix, mod, psf)
plt.suptitle('W%i: Orig' % band)
ps.savefig()
# Lanczos sub-sample
if band == 4:
lpix = _lanczos_subsample(opix, 2)
pix = lpix
h, w = pix.shape
#S = 140
slc = slice(h / 2 - S,
h / 2 + S + 1), slice(w / 2 - S, w / 2 + S + 1)
print('Resampled pix sum', pix.sum())
pix = pix[slc]
print('sliced pix sum', pix.sum())
psf = GaussianMixturePSF(fit.amp, fit.mean * scale,
fit.var * scale**2)
psfmodel = psf.getPointSourcePatch(0., 0., radius=h / 2)
mod = psfmodel.patch
print('Scaled mod sum', mod.sum())
mod = mod[slc]
print('Sliced mod sum', mod.sum())
_plot_psf(pix, mod, psf)
plt.suptitle('W%i: Scaled' % band)
ps.savefig()
plotslice = slice(S - plotss,
-(S - plotss)), slice(S - plotss, -(S - plotss))
psfx = GaussianMixturePSF.fromStamp(pix,
P0=(fit.amp, fit.mean * scale,
fit.var * scale**2))
psfmodel = psfx.getPointSourcePatch(0., 0., radius=h / 2)
mod = psfmodel.patch
print('From stamp: mod sum', mod.sum())
mod = mod[slc]
print('Sliced mod sum:', mod.sum())
_plot_psf(pix, mod, psfx)
plt.suptitle('W%i: fromStamp: %g = %s, res %.3f' %
(band, np.sum(psfx.mog.amp), ','.join(
['%.3f' % a
for a in psfx.mog.amp]), np.sum(pix - mod)))
ps.savefig()
print('Stamp-Fit PSF params:', psfx)
# class MyGaussianMixturePSF(GaussianMixturePSF):
# def getLogPrior(self):
# if np.any(self.mog.amp < 0.):
# return -np.inf
# for k in range(self.mog.K):
# if np.linalg.det(self.mog.var[k]) <= 0:
# return -np.inf
# return 0
#
# @property
# def amp(self):
# return self.mog.amp
#
# mypsf = MyGaussianMixturePSF(psfx.mog.amp, psfx.mog.mean, psfx.mog.var)
# mypsf.radius = sh/2
sh, sw = pix.shape
# Initialize from original fit params
psfx = psf
# Try concentric gaussian PSF
sigmas = []
for k in range(psfx.mog.K):
v = psfx.mog.var[k, :, :]
sigmas.append(np.sqrt(np.sqrt(np.abs(v[0, 0] * v[1, 1]))))
print('Initializing concentric Gaussian PSF with sigmas', sigmas)
gpsf = NCircularGaussianPSF(sigmas, psfx.mog.amp)
gpsf.radius = sh / 2
mypsf = gpsf
tim = Image(data=pix, invvar=1e6 * np.ones_like(pix), psf=mypsf)
tim.modelMinval = 1e-16
# xx,yy = np.meshgrid(np.arange(sw), np.arange(sh))
# cx,cy = np.sum(xx*pix)/np.sum(pix), np.sum(yy*pix)/np.sum(pix)
# print 'Centroid:', cx,cy
# print 'Pix midpoint:', sw/2, sh/2
cx, cy = sw / 2, sh / 2
src = PointSource(PixPos(cx, cy), Flux(1.0))
tractor = Tractor([tim], [src])
tim.freezeAllBut('psf')
# tractor.freezeParam('catalog')
src.freezeAllBut('pos')
# src.freezeAllBut('brightness')
# tractor.freezeParam('images')
# tractor.optimize_forced_photometry()
# tractor.thawParam('images')
# print 'Source flux after forced-photom fit:', src.getBrightness()
print('Optimizing Params:')
tractor.printThawedParams()
for i in range(200):
#dlnp,X,alpha = tractor.optimize(damp=0.1)
dlnp, X, alpha = tractor.optimize(damp=1)
print(i, 'dlnp %.3g' % dlnp, 'psf', gpsf)
if dlnp < 1e-6:
break
tractor.freezeParam('catalog')
gpsf.sigmas.stepsize = len(gpsf.sigmas) * [1e-6]
gpsf.weights.stepsize = len(gpsf.sigmas) * [1e-6]
for i in range(200):
#dlnp,X,alpha = tractor.optimize(damp=0.1)
dlnp, X, alpha = tractor.optimize(damp=1)
print(i, 'dlnp %.3g' % dlnp, 'psf', gpsf)
if dlnp < 1e-6:
break
print('PSF3(opt): flux', src.brightness)
print('PSF amps:', np.sum(mypsf.amp))
print('PSF amps * Source brightness:',
src.brightness.getValue() * np.sum(mypsf.amp))
print('pix sum:', pix.sum())
print('Optimize source:', src)
print('Optimized PSF:', mypsf)
mod = tractor.getModelImage(0)
print('Mod sum:', mod.sum())
_plot_psf(pix, mod, mypsf, flux=src.brightness.getValue())
plt.suptitle('W%i psf3 (opt): %g = %s, resid %.3f' %
(band, np.sum(mypsf.amp), ','.join(
['%.3f' % a for a in mypsf.amp]), np.sum(pix - mod)))
ps.savefig()
# Write before normalizing!
T = fits_table()
T.amp = mypsf.mog.amp
T.mean = mypsf.mog.mean
T.var = mypsf.mog.var
T.writeto('psf3-w%i.fits' % band)
T.writeto('psf3.fits', append=(band != 1))
mypsf.weights.setParams(
np.array(mypsf.weights.getParams()) /
sum(mypsf.weights.getParams()))
print('Normalized PSF weights:', mypsf)
mod = tractor.getModelImage(0)
print('Mod sum:', mod.sum())
_plot_psf(pix, mod, mypsf, flux=src.brightness.getValue())
plt.suptitle('W%i psf3 (opt): %g = %s, resid %.3f' %
(band, np.sum(mypsf.amp), ','.join(
['%.3f' % a for a in mypsf.amp]), np.sum(pix - mod)))
ps.savefig()
class MyGaussianPSF(NCircularGaussianPSF):
'''
A PSF model with strictly positive weights that sum to
unity.
'''
def __init__(self, sigmas, weights):
ww = np.array(weights)
ww = np.log(ww)
super(MyGaussianPSF, self).__init__(sigmas, ww)
@staticmethod
def getNamedParams():
return dict(sigmas=0, logweights=1)
def __str__(self):
return ('MyGaussianPSF: sigmas [ ' +
', '.join(['%.3f' % s for s in self.mysigmas]) +
' ], weights [ ' +
', '.join(['%.3f' % w for w in self.myweights]) + ' ]')
@property
def myweights(self):
ww = np.exp(self.logweights.getAllParams())
ww /= ww.sum()
return ww
@property
def weights(self):
ww = np.exp(self.logweights.getParams())
wsum = np.sum(np.exp(self.logweights.getAllParams()))
return ww / wsum
if band == 4:
# HACK
mypsf = MyGaussianPSF([1.7, 6.4, 15.0], [0.333, 0.666, 0.1])
else:
mypsf = MyGaussianPSF(gpsf.sigmas, gpsf.amp)
mypsf.radius = sh / 2
tim.psf = mypsf
print('Optimizing Params:')
tractor.printThawedParams()
for i in range(200):
#dlnp,X,alpha = tractor.optimize(damp=0.1)
dlnp, X, alpha = tractor.optimize(damp=1)
print(i, 'dlnp %.3g' % dlnp, 'psf', mypsf)
if dlnp < 1e-6:
break
mypsf.sigmas.stepsize = len(mypsf.sigmas) * [1e-6]
mypsf.logweights.stepsize = len(mypsf.sigmas) * [1e-6]
for i in range(200):
#dlnp,X,alpha = tractor.optimize(damp=0.1)
dlnp, X, alpha = tractor.optimize(damp=1)
print(i, 'dlnp %.3g' % dlnp, 'psf', mypsf)
if dlnp < 1e-6:
break
print('PSF amps:', np.sum(mypsf.amp))
print('pix sum:', pix.sum())
print('Optimize source:', src)
print('Optimized PSF:', mypsf)
mod = tractor.getModelImage(0)
print('Mod sum:', mod.sum())
_plot_psf(pix, mod, mypsf, flux=src.brightness.getValue())
plt.suptitle('W%i psf3 (opt2): %g = %s, resid %.3f' %
(band, np.sum(mypsf.amp), ','.join(
['%.3f' % a for a in mypsf.amp]), np.sum(pix - mod)))
ps.savefig()
# Write
mog = mypsf.getMixtureOfGaussians()
T = fits_table()
T.amp = mog.amp
T.mean = mog.mean
T.var = mog.var
T.writeto('psf4.fits', append=(band != 1))
#_meisner_psf_models()
sys.exit(0)
psf2 = GaussianMixturePSF(mypsf.mog.amp[:2] / mypsf.mog.amp[:2].sum(),
mypsf.mog.mean[:2, :], mypsf.mog.var[:2, :, :])
psf2.radius = sh / 2
tim.psf = psf2
mod = tractor.getModelImage(0)
_plot_psf(lpix, mod, psf2, flux=src.brightness.getValue())
plt.suptitle('psf2 (init)')
ps.savefig()
src.freezeAllBut('brightness')
tractor.freezeParam('catalog')
for i in range(100):
print('Optimizing PSF:')
tractor.printThawedParams()
dlnp, X, alpha = tractor.optimize(damp=1.)
tractor.freezeParam('images')
tractor.thawParam('catalog')
print('Optimizing flux:')
tractor.printThawedParams()
tractor.optimize_forced_photometry()
tractor.thawParam('images')
def _meisner_psf_models():
global plotslice
# Meisner's PSF models
#for band in [4]:
for band in [1,2,3,4]:
print
print 'W%i' % band
print
#pix = fitsio.read('wise-psf-avg-pix.fits', ext=band-1)
pix = fitsio.read('wise-psf-avg-pix-bright.fits', ext=band-1)
fit = fits_table('wise-psf-avg.fits', hdu=band)
#fit = fits_table('psf-allwise-con3.fits', hdu=band)
scale = 1.
print 'Pix shape', pix.shape
h,w = pix.shape
xx,yy = np.meshgrid(np.arange(w), np.arange(h))
cx,cy = np.sum(xx*pix)/np.sum(pix), np.sum(yy*pix)/np.sum(pix)
print 'Centroid:', cx,cy
#S = 100
S = h/2
slc = slice(h/2-S, h/2+S+1), slice(w/2-S, w/2+S+1)
plotss = 30
plotslice = slice(S-plotss, -(S-plotss)), slice(S-plotss, -(S-plotss))
opix = pix
print 'Original pixels sum:', opix.sum()
pix /= pix.sum()
pix = pix[slc]
print 'Sliced pix sum', pix.sum()
psf = GaussianMixturePSF(fit.amp, fit.mean * scale, fit.var * scale**2)
psfmodel = psf.getPointSourcePatch(0., 0., radius=h/2)
mod = psfmodel.patch
print 'Orig mod sum', mod.sum()
mod = mod[slc]
print 'Sliced mod sum', mod.sum()
print 'Amps:', np.sum(psf.mog.amp)
_plot_psf(pix, mod, psf)
plt.suptitle('W%i: Orig' % band)
ps.savefig()
# Lanczos sub-sample
if band == 4:
lpix = _lanczos_subsample(opix, 2)
pix = lpix
h,w = pix.shape
#S = 140
slc = slice(h/2-S, h/2+S+1), slice(w/2-S, w/2+S+1)
print 'Resampled pix sum', pix.sum()
pix = pix[slc]
print 'sliced pix sum', pix.sum()
psf = GaussianMixturePSF(fit.amp, fit.mean * scale, fit.var * scale**2)
psfmodel = psf.getPointSourcePatch(0., 0., radius=h/2)
mod = psfmodel.patch
print 'Scaled mod sum', mod.sum()
mod = mod[slc]
print 'Sliced mod sum', mod.sum()
_plot_psf(pix, mod, psf)
plt.suptitle('W%i: Scaled' % band)
ps.savefig()
plotslice = slice(S-plotss,-(S-plotss)),slice(S-plotss,-(S-plotss))
psfx = GaussianMixturePSF.fromStamp(pix, P0=(fit.amp, fit.mean*scale, fit.var*scale**2))
psfmodel = psfx.getPointSourcePatch(0., 0., radius=h/2)
mod = psfmodel.patch
print 'From stamp: mod sum', mod.sum()
mod = mod[slc]
print 'Sliced mod sum:', mod.sum()
_plot_psf(pix, mod, psfx)
plt.suptitle('W%i: fromStamp: %g = %s, res %.3f' %
(band, np.sum(psfx.mog.amp), ','.join(['%.3f'%a for a in psfx.mog.amp]),
np.sum(pix - mod)))
ps.savefig()
print 'Stamp-Fit PSF params:', psfx
# class MyGaussianMixturePSF(GaussianMixturePSF):
# def getLogPrior(self):
# if np.any(self.mog.amp < 0.):
# return -np.inf
# for k in range(self.mog.K):
# if np.linalg.det(self.mog.var[k]) <= 0:
# return -np.inf
# return 0
#
# @property
# def amp(self):
# return self.mog.amp
#
# mypsf = MyGaussianMixturePSF(psfx.mog.amp, psfx.mog.mean, psfx.mog.var)
# mypsf.radius = sh/2
sh,sw = pix.shape
# Initialize from original fit params
psfx = psf
# Try concentric gaussian PSF
sigmas = []
for k in range(psfx.mog.K):
v = psfx.mog.var[k,:,:]
sigmas.append(np.sqrt(np.sqrt(np.abs(v[0,0] * v[1,1]))))
print 'Initializing concentric Gaussian PSF with sigmas', sigmas
gpsf = NCircularGaussianPSF(sigmas, psfx.mog.amp)
gpsf.radius = sh/2
mypsf = gpsf
tim = Image(data=pix, invvar=1e6 * np.ones_like(pix), psf=mypsf)
tim.modelMinval = 1e-16
# xx,yy = np.meshgrid(np.arange(sw), np.arange(sh))
# cx,cy = np.sum(xx*pix)/np.sum(pix), np.sum(yy*pix)/np.sum(pix)
# print 'Centroid:', cx,cy
# print 'Pix midpoint:', sw/2, sh/2
cx,cy = sw/2, sh/2
src = PointSource(PixPos(cx, cy), Flux(1.0))
tractor = Tractor([tim], [src])
tim.freezeAllBut('psf')
#tractor.freezeParam('catalog')
src.freezeAllBut('pos')
# src.freezeAllBut('brightness')
# tractor.freezeParam('images')
# tractor.optimize_forced_photometry()
# tractor.thawParam('images')
# print 'Source flux after forced-photom fit:', src.getBrightness()
print 'Optimizing Params:'
tractor.printThawedParams()
for i in range(200):
#dlnp,X,alpha = tractor.optimize(damp=0.1)
dlnp,X,alpha = tractor.optimize(damp=1)
print i,'dlnp %.3g' % dlnp, 'psf', gpsf
if dlnp < 1e-6:
break
tractor.freezeParam('catalog')
gpsf.sigmas.stepsize = len(gpsf.sigmas) * [1e-6]
gpsf.weights.stepsize = len(gpsf.sigmas) * [1e-6]
for i in range(200):
#dlnp,X,alpha = tractor.optimize(damp=0.1)
dlnp,X,alpha = tractor.optimize(damp=1)
print i,'dlnp %.3g' % dlnp, 'psf', gpsf
if dlnp < 1e-6:
break
print 'PSF3(opt): flux', src.brightness
print 'PSF amps:', np.sum(mypsf.amp)
print 'PSF amps * Source brightness:', src.brightness.getValue() * np.sum(mypsf.amp)
print 'pix sum:', pix.sum()
print 'Optimize source:', src
print 'Optimized PSF:', mypsf
mod = tractor.getModelImage(0)
print 'Mod sum:', mod.sum()
_plot_psf(pix, mod, mypsf, flux=src.brightness.getValue())
plt.suptitle('W%i psf3 (opt): %g = %s, resid %.3f' %
(band, np.sum(mypsf.amp), ','.join(['%.3f'%a for a in mypsf.amp]), np.sum(pix-mod)))
ps.savefig()
# Write before normalizing!
T = fits_table()
T.amp = mypsf.mog.amp
T.mean = mypsf.mog.mean
T.var = mypsf.mog.var
T.writeto('psf3-w%i.fits' % band)
T.writeto('psf3.fits', append=(band != 1))
mypsf.weights.setParams(np.array(mypsf.weights.getParams()) /
sum(mypsf.weights.getParams()))
print 'Normalized PSF weights:', mypsf
mod = tractor.getModelImage(0)
print 'Mod sum:', mod.sum()
_plot_psf(pix, mod, mypsf, flux=src.brightness.getValue())
plt.suptitle('W%i psf3 (opt): %g = %s, resid %.3f' %
(band, np.sum(mypsf.amp), ','.join(['%.3f'%a for a in mypsf.amp]), np.sum(pix-mod)))
ps.savefig()
class MyGaussianPSF(NCircularGaussianPSF):
'''
A PSF model with strictly positive weights that sum to
unity.
'''
def __init__(self, sigmas, weights):
ww = np.array(weights)
ww = np.log(ww)
super(MyGaussianPSF, self).__init__(sigmas, ww)
@staticmethod
def getNamedParams():
return dict(sigmas=0, logweights=1)
def __str__(self):
return ('MyGaussianPSF: sigmas [ ' +
', '.join(['%.3f'%s for s in self.mysigmas]) +
' ], weights [ ' +
', '.join(['%.3f'%w for w in self.myweights]) +
' ]')
@property
def myweights(self):
ww = np.exp(self.logweights.getAllParams())
ww /= ww.sum()
return ww
@property
def weights(self):
ww = np.exp(self.logweights.getParams())
wsum = np.sum(np.exp(self.logweights.getAllParams()))
return ww / wsum
if band == 4:
# HACK
mypsf = MyGaussianPSF([1.7, 6.4, 15.0], [0.333, 0.666, 0.1])
else:
mypsf = MyGaussianPSF(gpsf.sigmas, gpsf.amp)
mypsf.radius = sh/2
tim.psf = mypsf
print 'Optimizing Params:'
tractor.printThawedParams()
for i in range(200):
#dlnp,X,alpha = tractor.optimize(damp=0.1)
dlnp,X,alpha = tractor.optimize(damp=1)
print i,'dlnp %.3g' % dlnp, 'psf', mypsf
if dlnp < 1e-6:
break
mypsf.sigmas.stepsize = len(mypsf.sigmas) * [1e-6]
mypsf.logweights.stepsize = len(mypsf.sigmas) * [1e-6]
for i in range(200):
#dlnp,X,alpha = tractor.optimize(damp=0.1)
dlnp,X,alpha = tractor.optimize(damp=1)
print i,'dlnp %.3g' % dlnp, 'psf', mypsf
if dlnp < 1e-6:
break
print 'PSF amps:', np.sum(mypsf.amp)
print 'pix sum:', pix.sum()
print 'Optimize source:', src
print 'Optimized PSF:', mypsf
mod = tractor.getModelImage(0)
print 'Mod sum:', mod.sum()
_plot_psf(pix, mod, mypsf, flux=src.brightness.getValue())
plt.suptitle('W%i psf3 (opt2): %g = %s, resid %.3f' %
(band, np.sum(mypsf.amp), ','.join(['%.3f'%a for a in mypsf.amp]), np.sum(pix-mod)))
ps.savefig()
# Write
mog = mypsf.getMixtureOfGaussians()
T = fits_table()
T.amp = mog.amp
T.mean = mog.mean
T.var = mog.var
T.writeto('psf4.fits', append=(band != 1))
#_meisner_psf_models()
sys.exit(0)
psf2 = GaussianMixturePSF(mypsf.mog.amp[:2]/mypsf.mog.amp[:2].sum(),
mypsf.mog.mean[:2,:], mypsf.mog.var[:2,:,:])
psf2.radius = sh/2
tim.psf = psf2
mod = tractor.getModelImage(0)
_plot_psf(lpix, mod, psf2, flux=src.brightness.getValue())
plt.suptitle('psf2 (init)')
ps.savefig()
src.freezeAllBut('brightness')
tractor.freezeParam('catalog')
for i in range(100):
print 'Optimizing PSF:'
tractor.printThawedParams()
dlnp,X,alpha = tractor.optimize(damp=1.)
tractor.freezeParam('images')
tractor.thawParam('catalog')
print 'Optimizing flux:'
tractor.printThawedParams()
tractor.optimize_forced_photometry()
tractor.thawParam('images')
