def combine_scinv(self):
from glob import glob
outfile=self.combined_file()
stdout.write("Will write to file: %s\n" % outfile)
dir=self.output_dir()
pattern = self.scinv_file_pattern()
pattern=path_join(dir, pattern)
flist = glob(pattern)
datalist = []
idmin = 0
for f in flist:
print f
tdata = esutil.io.read(f)
data = numpy_util.add_fields(tdata, [('zid','i4')])
data['zid'] = idmin + numpy.arange(data.size)
idmin += data.size
print data['zid'].min(), data['zid'].max()
datalist.append(data)
hdr = esutil.sfile.read_header(flist[0])
print 'combining data'
data = numpy_util.combine_arrlist(datalist)
print 'writing file: %s' % outfile
esutil.sfile.write(data, outfile, header=hdr)
def _load_data(self):
from esutil.numpy_util import combine_arrlist
alldata = []
for ccd in xrange(1, 62 + 1):
datai = files.read_fits_output(run=self.run, ftype="psf", psfnum=self.psfnum, shnum=self.shnum, ccd=ccd)
if datai is None:
fname = files.get_output_path(run=self.run, ftype="psf", psfnum=self.psfnum, shnum=self.shnum, ccd=ccd)
raise ValueError("error reading: %s")
alldata.append(datai)
data = combine_arrlist(alldata)
flags_name = self.model + "_flags"
w, = where(data[flags_name] == 0)
if w.size == 0:
raise ValueError("none with flags==0")
data = data[w]
self._data = data
def go(self):
for psfnum in files.PSFNUMS:
for shnum in files.SHNUMS:
exp_data=[]
for ccd in files.CCDS:
pipe=self.load_single(psfnum,shnum,ccd)
ccd_data_i=self.get_matched_struct(pipe)
w,=where(ccd_data_i['flags'] == 0)
ccd_data_i=ccd_data_i[w]
self.set_use(ccd_data_i)
exp_data.append( ccd_data_i )
exp_data=combine_arrlist(exp_data)
self.write_data(exp_data, psfnum, shnum)
def read_exposure_data(dir,exposurename):
out_dtype=[('ccd','i1'),
('x','f4'),('y','f4'),
('psf_flags','i4'),
('e1','f4'),('e2','f4'),
('e1interp','f4'),('e2interp','f4')]
datalist=[]
for ccd in xrange(1,1+62):
psf_file = path_join(dir,'%s_%02d_psf.fits' % (exposurename,ccd))
psf=esutil.io.read(psf_file)
psfe1 = psf['shapelets'][:,3]*sqrt(2)
psfe2 = -psf['shapelets'][:,4]*sqrt(2)
shear_file = path_join(dir,'%s_%02d_shear.fits' % (exposurename,ccd))
shear=esutil.io.read(shear_file)
e1interp = shear['interp_psf_coeffs'][:,3]*sqrt(2)
e2interp = -shear['interp_psf_coeffs'][:,4]*sqrt(2)
mpsf,mshear=numpy_util.match(psf['id'], shear['id'])
tdata = numpy.zeros(mpsf.size, dtype=out_dtype)
tdata['ccd'] = ccd
tdata['x'] = psf['x'][mpsf]
tdata['y'] = psf['y'][mpsf]
tdata['e1'] = psfe1[mpsf]
tdata['e2'] = psfe2[mpsf]
tdata['psf_flags'] = psf['psf_flags'][mpsf]
tdata['e1interp'] = e1interp[mshear]
tdata['e2interp'] = e2interp[mshear]
datalist.append(tdata)
data = numpy_util.combine_arrlist(datalist)
return data
def _load_data(self):
from esutil.numpy_util import combine_arrlist
ntot=len(self['shnums'])*len(self['psfnums'])*len(self['ccds'])
itot=1
if self['progress']:
from progressbar import ProgressBar
prog=ProgressBar(width=70, color='green')
datalist=[]
for shnum in self['shnums']:
shlist=[]
for psfnum in self['psfnums']:
for ccd in self['ccds']:
if self['progress']:
prog.update(frac=float(itot)/ntot)
itot += 1
data0=self.read_one(shnum,psfnum,ccd)
if data0 is not None:
if self['setname'] is not None:
data=self.select(data0)
else:
data=data0
if data is not None:
datalist.append(data)
del data0
if len(datalist) == 0:
if self['verbose']:
print 'no data read or passed cuts'
self._data=None
else:
self._data=combine_arrlist(datalist)
def read(fileobj, **keywords):
"""
Name:
io.read
Usage:
import esutil
data = esutil.io.read(
filename/fileobject,
typ=None,
ext=0,
rows=None, fields=None, columns=None,
header=False,
combine=False,
view=None,
lower=False, upper=False,
noroot=True, seproot=False,
verbose=False,
ensure_native=False)
Purpose:
Provide a single interface to read from a variety of file types.
Supports reading from a list of files.
Inputs:
filename/fileobject:
File name or an open file object. Can also be a sequence. If a
sequence is input, the return value will, by default, be a list of
results. If the return types are numpy arrays, one can send the
combine=True (the default) keyword to combine them into a single
array as long as the data types match.
Keywords:
type:
A string describing the file type, see below. If this is not sent,
then the file type is determined from the file extension.
ext:
The file extension. If multiple extensions are supported by the
file type, such as for FITS, then use this keyword to select which
is to be read. Default is the first extension with data.
rows:
For numpy record-type files such as FITS binary tables or simple
REC files, setting this keyword will return a subset of the rows.
For FITS, this requires reading the entire file and selecting a
subset. For REC files only the requested rows are read from disk
by using the recfile package. Default is all rows.
fields=, columns=:
For numpy record-type files such as FITS binary tables or simple
REC files, return a subset of the columns or fields. The keywords
"fields" and "columns" are synonyms. For FITS, this requires
reading the entire file and selecting a subset. For REC files only
the requested rows are read from disk by using the recfile package.
Default is all columns.
header:
If True, and the file type supports header+data, return a tuple
(data, header). Can also be 'only' in which case only the header
is read and returned (rec and fits only for now). Default is
False.
combine: If a list of filenames/fileobjects is sent, the default
behavior is to return a list of data. If combine=True and the
data are numpy arrays, attempt to combine them into a single
array. Only works if the data types match. Default True
view: If the result is derived from a numpy array, set this to
pick the view. E.g. pyfits returns a special pyfits type for
binary table. You can request a simple numpy array with fields
by setting view=numpy.ndarray, or a numpy recarray type with
view=numpy.recarray
lower,upper: For FITS files, if true convert the case of the
fields to all lower or all upper. Certain FITS writers
tend to write all fields names as capitals which can result
in annoyance.
noroot: For XML files, do not return the root name as the base
name in the dictionary. Default is True
seproot: For XML files, return a tuple (data, rootname) instead of
just the data under the root.
ensure_native: For numpy arrays, make sure data is in native
byte ordering.
Currently Supported File Types:
fits
Flexible Image Transport System
rec
Simple ascii header followed by data in binary or text form. These
files can be written/read using the esutil.sfile module. REC files
support appending rows. Also supports reading sub-selections of
rows and columns.
xml
Extensible Markup Language
json
JavaScript Object Notation. Less flexible than XML but more useful
in most practical situations such as storing inhomogeneous data in
a portable way.
yaml
A nice, human readable markup language, especially useful
for configuration files. YAML stands for
YAML Ain't Markup Language
pyobj
A straight dump of an object to disk using it's repr(). Files are
written using pprint, read simply using eval(open(file).read()).
This is not secure so use with caution.
Revision History:
Use **keywords for input and for sending to all called methods. Much
more flexible when adding new keywords and file types.
2010
"""
verbose = keywords.get('verbose', False)
# If input is a sequence, read them all.
if isinstance(fileobj, (list,tuple)):
flist = fileobj
nfiles=len(flist)
if nfiles==1:
return read(flist[0], **keywords)
combine = keywords.get('combine', True)
# we want to default to verbose in terms of showing progress
verbose_progress = keywords.get('verbose', True)
# a list was given
alldata = []
for i,f in enumerate(flist):
if verbose_progress:
print("reading %d/%d %s" % (i+1,nfiles,f))
# note, only fields/columns is being passed on but not rows
# also note seproot is not being passed on
data = read(f, **keywords)
alldata.append(data)
if combine:
fn,fobj,type,fs = _get_fname_ftype_from_inputs(fileobj[0], **keywords)
if type == 'fits' or type == 'rec':
# this will only work if the all data has the
# same structure
alldata = numpy_util.combine_arrlist(alldata)
return alldata
# a scalar was input
fname,fobj,type,fs = _get_fname_ftype_from_inputs(fileobj, **keywords)
if fs == 'hdfs':
with hdfs.HDFSFile(fname, verbose=verbose) as hdfs_file:
data = hdfs_file.read(read, **keywords)
return data
else:
if verbose:
print("reading:",fname)
# pick the right reader based on type
try:
if type == 'fits':
data = read_fits(fobj, **keywords)
elif type == 'json':
data = json_util.read(fobj, **keywords)
elif type == 'yaml':
data = read_yaml(fobj, **keywords)
elif type == 'rec':
data = read_rec(fobj, **keywords)
elif type == 'xml':
data = read_xml(fobj, **keywords)
elif type == 'pyobj':
data = read_pyobj(fobj, **keywords)
else:
raise ValueError("Don't know about file type '%s'" % type)
finally:
pass
return data
def read_output_set(run, psfnums, shnums,
objtype=None,
s2n_field='s2n_w',
s2n_min=None,
s2n_max=None,
s2_max=None,
gsens_min=None,
gerr_max=None,
columns=None,
subtract_mean=False,
progress=False):
"""
Read some data based on the input.
Multiple files may be read. If files are missing they will be skipped
Note only a single shear number is expected but many psfnums can
be sent.
Only those with flags==0 are kept.
parameters
----------
run: string
run id
psfnums: integers
the psf numbers to read
shnums: integers
The shear numbers to read.
objtype: string, optional
optionally select only objects with this best-fit model
columns: optional
only return these columns
subtract_mean: bool, optional
Calculate the mean g and subtract it
"""
from esutil.numpy_util import strmatch, combine_arrlist
psfnums=get_psfnums(psfnums)
shnums=get_shnums(shnums)
ntot=len(shnums)*len(psfnums)*62
itot=1
if progress:
from progressbar import ProgressBar
prog=ProgressBar(width=70, color='green')
#prog=ProgressBar(width=70, color='green', block='▣', empty='□')
#prog=ProgressBar(width=70, color='green', block='◧', empty='◫')
#prog=ProgressBar(width=70, color='green', block='■', empty='□')
#prog=ProgressBar(width=70, color='green', block='=', empty='-')
datalist=[]
for shnum in shnums:
shlist=[]
for psfnum in psfnums:
for ccd in xrange(1,62+1):
if progress:
#prog.update(frac=float(itot)/ntot,
# message='%s/%s' % (itot,ntot))
prog.update(frac=float(itot)/ntot)
itot += 1
fname=get_output_path(run=run, psfnum=psfnum, shnum=shnum,
ccd=ccd, ftype='shear')
if os.path.exists(fname):
data0=read_fits_output(run=run, psfnum=psfnum,
shnum=shnum, ccd=ccd,
ftype='shear',
columns=columns,
verbose=False)
logic=(data0['flags']==0) | (data0['flags']==65536)
if objtype:
logic=logic & strmatch(data0['model'],objtype)
if s2n_min is not None:
logic=logic & (data0[s2n_field] > s2n_min)
if s2n_max is not None:
logic=logic & (data0[s2n_field] < s2n_max)
if s2_max is not None:
logic=logic & (data0['s2'] < s2_max)
if gsens_min is not None:
logic=logic \
& (data0['gsens'][:,0] > gsens_min) \
& (data0['gsens'][:,1] > gsens_min)
if gerr_max is not None:
g1err=sqrt(data0['gcov'][:,0,0])
g2err=sqrt(data0['gcov'][:,1,1])
logic=logic \
& (g1err < gerr_max) & (g2err < gerr_max)
wkeep,=where(logic)
if wkeep.size==0:
print 'No objects passed cuts'
else:
data0=data0[wkeep]
shlist.append(data0)
shdata=combine_arrlist(shlist)
if subtract_mean:
g1mean = shdata['g'][:,0].mean()
g2mean = shdata['g'][:,1].mean()
shdata['g'][:,0] -= g1mean
shdata['g'][:,1] -= g2mean
datalist.append(shdata)
if len(datalist)==0:
raise RuntimeError("no outputs were found")
data=combine_arrlist(datalist)
return data
def read(fileobj, **keywords):
"""
Name:
io.read
Usage:
import esutil
data = esutil.io.read(
filename/fileobject,
typ=None,
ext=0,
rows=None, fields=None, columns=None,
header=False,
combine=False,
view=None,
lower=False, upper=False,
noroot=True, seproot=False,
verbose=False,
ensure_native=False)
Purpose:
Provide a single interface to read from a variety of file types.
Supports reading from a list of files.
Inputs:
filename/fileobject:
File name or an open file object. Can also be a sequence. If a
sequence is input, the return value will, by default, be a list of
results. If the return types are numpy arrays, one can send the
combine=True keyword to combine them into a single array as long
as the data types match.
Keywords:
type:
A string describing the file type, see below. If this is not sent,
then the file type is determined from the file extension.
ext:
The file extension. If multiple extensions are supported by the
file type, such as for FITS, then use this keyword to select which
is to be read. Default is the first extension with data.
rows:
For numpy record-type files such as FITS binary tables or simple
REC files, setting this keyword will return a subset of the rows.
For FITS, this requires reading the entire file and selecting a
subset. For REC files only the requested rows are read from disk
by using the recfile package. Default is all rows.
fields=, columns=:
For numpy record-type files such as FITS binary tables or simple
REC files, return a subset of the columns or fields. The keywords
"fields" and "columns" are synonyms. For FITS, this requires
reading the entire file and selecting a subset. For REC files only
the requested rows are read from disk by using the recfile package.
Default is all columns.
header:
If True, and the file type supports header+data, return a tuple
(data, header). Can also be 'only' in which case only the header
is read and returned (rec and fits only for now). Default is
False.
combine: If a list of filenames/fileobjects is sent, the default
behavior is to return a list of data. If combine=True and the
data are numpy arrays, attempt to combine them into a single
array. Only works if the data types match.
view: If the result is derived from a numpy array, set this to
pick the view. E.g. pyfits returns a special pyfits type for
binary table. You can request a simple numpy array with fields
by setting view=numpy.ndarray, or a numpy recarray type with
view=numpy.recarray
lower,upper: For FITS files, if true convert the case of the
fields to all lower or all upper. Certain FITS writers
tend to write all fields names as capitals which can result
in annoyance.
noroot: For XML files, do not return the root name as the base
name in the dictionary. Default is True
seproot: For XML files, return a tuple (data, rootname) instead of
just the data under the root.
ensure_native: For numpy arrays, make sure data is in native
byte ordering.
Currently Supported File Types:
fits
Flexible Image Transport System
rec
Simple ascii header followed by data in binary or text form. These
files can be written/read using the esutil.sfile module. REC files
support appending rows. Also supports reading sub-selections of
rows and columns.
xml
Extensible Markup Language
json
JavaScript Object Notation. Less flexible than XML but more useful
in most practical situations such as storing inhomogeneous data in
a portable way.
yaml
A nice, human readable markup language, especially useful
for configuration files. YAML stands for
YAML Ain't Markup Language
pyobj
A straight dump of an object to disk using it's repr(). Files are
written using pprint, read simply using eval(open(file).read()).
This is not secure so use with caution.
Revision History:
Use **keywords for input and for sending to all called methods. Much
more flexible when adding new keywords and file types.
2010
"""
verbose = keywords.get('verbose', False)
# If input is a sequence, read them all.
if isinstance(fileobj, (list,tuple)):
combine = keywords.get('combine', False)
# a list was given
alldata = []
for f in fileobj:
# note, only fields/columns is begin passed on but not rows
# also note seproot is not being passed on
data = read(f, **keywords)
alldata.append(data)
if combine:
if len(fileobj) == 1:
alldata = alldata[0]
else:
fn,fobj,type,fs = _get_fname_ftype_from_inputs(fileobj[0], **keywords)
if type == 'fits' or type == 'rec':
# this will only work if the all data has the
# same structure
if verbose:
stderr.write("Combining arrays\n")
alldata = numpy_util.combine_arrlist(alldata)
return alldata
# a scalar was input
fname,fobj,type,fs = _get_fname_ftype_from_inputs(fileobj, **keywords)
if fs == 'hdfs':
with hdfs.HDFSFile(fname, verbose=verbose) as hdfs_file:
data = hdfs_file.read(read, **keywords)
return data
else:
if verbose:
stderr.write("Reading: %s\n" % fname)
# pick the right reader based on type
try:
if type == 'fits':
data = read_fits(fobj, **keywords)
elif type == 'json':
data = json_util.read(fobj, **keywords)
elif type == 'yaml':
data = read_yaml(fobj, **keywords)
elif type == 'rec':
data = read_rec(fobj, **keywords)
elif type == 'xml':
data = read_xml(fobj, **keywords)
elif type == 'pyobj':
data = read_pyobj(fobj, **keywords)
else:
raise ValueError("Don't know about file type '%s'" % type)
finally:
pass
return data
def doplot(serun, exposurename, example_wcs_byccd, nx=3,ny=3,diff=False,
imageformat='png', ptypes=['unbinned','comparebinned']):
stdout.write("serun: %s\n" % serun)
stdout.write("exposurename: %s\n" % exposurename)
stdout.write("\tReading checkpsf data\n")
alldata = []
allstats = []
for ccd in range(1,62+1):
try:
data = deswl.files.wlse_read(exposurename,ccd,'checkpsf',
serun=serun)
fields=['e1','e2','e1interp','e2interp']
stats = du.stats_xy(data, fields, nx=nx, ny=ny, typ='median')
ccd_wcs = example_wcs_byccd[ccd]
fx,fy = ccd_wcs.image2sky(data['x'], data['y'])
mfx,mfy = ccd_wcs.image2sky(stats['mx'],stats['my'])
newdata = numpy_util.add_fields(data, [('fx','f4'),('fy','f4')])
newdata['fx'] = fx - 337.3
newdata['fy'] = fy + 15.0
fadd=[('mfx','f4'),('mfy','f4'),('me1diff','f4'),('me2diff','f4')]
newstats = numpy_util.add_fields(stats,fadd)
newstats['mfx'] = mfx - 337.3
newstats['mfy'] = mfy + 15.0
newstats['me1diff'] = newstats['me1interp']-newstats['me1']
newstats['me2diff'] = newstats['me2interp']-newstats['me2']
alldata.append(newdata)
allstats.append(newstats)
except:
stdout.write("Failed ccd=%s\n" % ccd)
print sys.exc_info()
data = numpy_util.combine_arrlist(alldata)
stats = numpy_util.combine_arrlist(allstats)
stdout.write("\tPlotting whiskers\n")
plt = du.setuplot('Agg')
plt.clf()
nplots = len(ptypes)
xsize=8*nplots
fig=plt.figure(figsize=(xsize,7))
iplot = 1
u, v = polar2whisker(data['e1'], data['e2'])
uinterp, vinterp = polar2whisker(data['e1interp'], data['e2interp'])
mu, mv = polar2whisker(stats['me1'],stats['me2'])
muinterp, mvinterp = polar2whisker(stats['me1interp'],stats['me2interp'])
mudiff, mvdiff = polar2whisker(stats['me1diff'],stats['me2diff'])
# example size to plot
psize=0.01
if 'unbinned' in ptypes or 'compareunbinned' in ptypes:
stdout.write("\t\tDoing plot type 'unbinned'\n")
ax = fig.add_subplot(1,nplots,iplot)
iplot += 1
# plot individual stars
scale=3.
whiskers(ax, data['fx'], data['fy'], u, v, scale=scale,
linewidth=0.25)
# a measure of scale
xtext=-1.2
ax.text(xtext, 0.9, str(psize), verticalalignment='center')
whiskers(ax, xtext+0.18, 0.9, psize, 0.0, color='blue', scale=scale)
if 'compareunbinned' in ptypes:
stdout.write("\t\tOverplotting 'compareunbinned'\n")
whiskers(ax, data['fx'], data['fy'], uinterp, vinterp,
scale=scale, color='red', linewidth=0.25)
# add a legend
ypos = 1.1
ystep = 0.07
xtext = -1.1
ax.text(xtext, ypos-ystep, "data", verticalalignment='center')
whiskers(ax, xtext+0.23, ypos-ystep, psize, 0.0,
color='black', scale=scale)
ax.text(xtext, ypos-2*ystep, "interp", verticalalignment='center')
whiskers(ax, xtext+0.23, ypos-2*ystep, psize, 0.0,
color='red', scale=scale)
ax.set_xlim(-1.3,1.3)
ax.set_ylim(-1.3,1.3)
set_minor_ticks(ax)
if 'comparebinned' in ptypes:
# comparison plot for the binned data
ax = fig.add_subplot(1,nplots,iplot)
iplot += 1
scale=10.
whiskers(ax, stats['mfx'], stats['mfy'], mu, mv, scale=scale)
whiskers(ax, stats['mfx'], stats['mfy'], muinterp, mvinterp,
scale=scale, color='red')
# legend
ypos = 1.1
ystep = 0.07
xtext = -1.1
ax.text(xtext, ypos, str(psize), verticalalignment='center')
whiskers(ax, xtext+0.23, ypos, psize, 0.0, color='blue', scale=scale)
ax.text(xtext, ypos-ystep, "data", verticalalignment='center')
whiskers(ax, xtext+0.23, ypos-ystep, psize, 0.0,
color='black', scale=scale)
ax.text(xtext, ypos-2*ystep, "interp", verticalalignment='center')
whiskers(ax, xtext+0.23, ypos-2*ystep, psize, 0.0,
color='red', scale=scale)
ax.set_xlim(-1.3,1.3)
ax.set_ylim(-1.3,1.3)
set_minor_ticks(ax)
if 'diffbinned' in ptypes:
# comparison plot for the binned data
ax = fig.add_subplot(1,nplots,iplot)
iplot += 1
scale=50.
psize=0.001
whiskers(ax, stats['mfx'], stats['mfy'], mudiff, mvdiff, scale=scale)
ax.text(xtext, ypos, str(psize), verticalalignment='center')
whiskers(ax, xtext+0.23, ypos, psize, 0.0, color='blue', scale=scale)
ax.set_xlim(-1.3,1.3)
ax.set_ylim(-1.3,1.3)
set_minor_ticks(ax)
outdir='/home/users/esheldon/www/tmp/plots'
outdir=os.path.join(outdir, serun)
if not os.path.exists(outdir):
os.makedirs(outdir)
ptypes_string = '-'.join(ptypes)
outfile = "%s-checkpsf-%s.%s" % (exposurename,ptypes_string, imageformat)
outfile=os.path.join(outdir,outfile)
stdout.write("\tWriting plot file: %s\n" % outfile)
plt.savefig(outfile, bbox_inches='tight', pad_inches=0.1)
def plot_shearxy_byccd(serun, region,
example_wcs_byccd=None, outfile=None, typ='pdf'):
"""
Take the input data, split by ccd, and plot each separately. The x,y
are converted to a ra,dec type system that preserved the layout of
the camera. We don't use the actual ra/dec here because we want to
be able to combine multiple exposures on the same plot.
"""
from numpy import arctan2, pi
fpath=deswl.files.wlse_collated_path(serun,'gal',ftype='rec',region=region)
data = esutil.io.read(fpath, verbose=True)
if data.size == 0:
stdout.write("No objects in region: %s\n" % region)
return
outdir=os.path.dirname(fpath)
outdir=os.path.join(outdir, 'plots')
if not os.path.exists(outdir):
os.makedirs(outdir)
outfile=os.path.basename(fpath).replace('.rec', '-shear-byccd.'+typ)
outfile=os.path.join(outdir, outfile)
stdout.write("Will write image: %s\n" % outfile)
plt=setuplot('Agg')
plt.clf()
if example_wcs_byccd is None:
example_wcs_byccd = get_exposure_wcs_example()
h,rev = esutil.stat.histogram(data['ccd'], rev=True, min=1,max=62)
allstats = []
#for ccd in range(1,63):
running_shear1=0.0
running_shear2=0.0
running_shear_weight=0.0
for iccd in range(len(h)):
if rev[iccd] != rev[iccd+1]:
w = rev[ rev[iccd]:rev[iccd+1] ]
ccd = data['ccd'][w[0]]
stdout.write('\tccd=%s\n' % ccd)
# no copy is made here!
stats = stats_shearxy(data[w], nx=3, ny=3)
ccd_wcs = example_wcs_byccd[ccd]
xx,yy = ccd_wcs.image2sky(stats['mx'],stats['my'])
xx -= 337.3
yy += 15.0
stats['mx'] = xx
stats['my'] = yy
allstats.append(stats)
weights = 1.0/(stats['mshear1_err']**2 + stats['mshear2_err']**2)
s1 = stats['mshear1']*weights
s2 = stats['mshear2']*weights
s1sum=s1.sum()
s2sum=s2.sum()
wsum = weights.sum()
running_shear1 += s1sum
running_shear2 += s2sum
running_shear_weight += wsum
thise1 = s1sum/wsum
thise2 = s2sum/wsum
thisangle = 0.5*arctan2(thise2, thise1)*180./pi
stdout.write("\t\t<e1>=%s\n" % thise1 )
stdout.write("\t\t<e2>=%s\n" % thise2 )
stdout.write("\t\t<angle>=%s\n" % thisangle )
#plt.plot(xx,yy,'.',markersize=1)
#plt.show()
stats = numpy_util.combine_arrlist(allstats)
mshear1 = running_shear1/running_shear_weight
mshear2 = running_shear2/running_shear_weight
mangle = 0.5*arctan2(mshear2, mshear1)*180.0/pi
mshear = numpy.sqrt( mshear1**2 + mshear2**2 )
mangle_err = stats['mangle'].std()/numpy.sqrt(stats['mangle'].size)
print "mins:",stats['mshear1'].min(), stats['mshear2'].min(), \
stats['mshear'].min()
print "maxs:",stats['mshear1'].max(), stats['mshear2'].max(), \
stats['mshear'].max()
stdout.write("overall averages: \n\tshear1: "
"%s\n\tshear2: %s\n\tangle: %s\n" % (mshear1, mshear2,mangle))
stdout.write('\tangle error approximately %s\n' % mangle_err)
# x component of the "vector" version
u = stats['mshear']*numpy.cos(stats['mangle'])#*10000
# y component of the "vector" version
v = stats['mshear']*numpy.sin(stats['mangle'])#*10000
# scale=1 means a vector of length 1 will cover essentially
# all the plot. I want them slightly smaller, so I'm using
# scale=1.5
scale=1.5
ax=plt.axes()
from matplotlib.ticker import MultipleLocator as ml
ax.xaxis.set_minor_locator(ml(0.1))
ax.yaxis.set_minor_locator(ml(0.1))
#plt.quiver(stats['mx'], stats['my'], u, v,headwidth=0,
# scale=scale, pivot='middle')
whiskers(plt, stats['mx'], stats['my'], u, v)
xtext=336.0 - 337.3
whiskers(plt, 336.18-337.3, -14.1+15.0, 0.05, 0.0, color='blue')
plt.text(xtext, -14.1+15, "0.05", verticalalignment='center')
ystart=-15.6 + 15
ystep=-0.08
istep = 0
ytext=ystart+istep*ystep
plt.text(xtext, ytext, r"$\langle \gamma_1 \rangle=%0.3f$" % mshear1,
verticalalignment='center')
istep+=1
ytext=ystart+istep*ystep
plt.text(xtext, ytext, r"$\langle \gamma_2 \rangle=%0.3f$" % mshear2,
verticalalignment='center')
istep+=1
ytext=ystart+istep*ystep
plt.text(xtext,ytext,r"$\langle \theta \rangle=%0.2f^{\circ}$" % mangle,
verticalalignment='center')
# plot a whisker representing the average
istep+=1
ytext=ystart+istep*ystep
svec1 = mshear*numpy.cos( mangle*pi/180. )
svec2 = mshear*numpy.sin( mangle*pi/180. )
plt.text(xtext, ytext, r"$\langle \gamma \rangle=%0.3f$" % mshear,
verticalalignment='center')
istep+=1
ytext=ystart+istep*ystep
whiskers(plt, 336.225-337.3, ytext, svec1, svec2, color='red')
label = 'region%s' % region
ax=plt.axes()
plt.text(0.90, 0.9, label,
horizontalalignment='center',
verticalalignment='center',
transform=ax.transAxes,
fontsize=18)
# this is so the arrows have the right angle
plt.axis('equal')
plt.ylim(-16.1+15.0,-13.9+15.0)
if outfile is not None:
stdout.write("Writing file: %s\n" % outfile)
plt.savefig(outfile, bbox_inches='tight', pad_inches=0.2)
else:
plt.show()
#print 'mean shear: ',stats['mshear']
return example_wcs_byccd
