def compute_exptime(band,trange,verbose=0,skypos=None,detsize=1.25,
retries=20,chunksz=10.e6,coadd=False):
"""Compute the effective exposure time."""
# FIXME: This skypos[] check appears to not work properly and leads
# to dramatic _underestimates_ of the exposure time.
if skypos and coadd:
tranges = fGetTimeRanges(band,skypos,verbose=verbose,trange=trange,
retries=retries)
else:
tranges=[trange]
if verbose>1:
print_inline('Computing exposure within {tr}'.format(tr=tranges))
if skypos and coadd:
print 'Based on skypos {sp}'.format(sp=skypos)
exptime = 0.
for trange in tranges:
# Traverse the exposure ranges in manageable chunks to hopefully keep
# the query response time below the HTTP timeout...
chunks = (np.linspace(trange[0],trange[1],
num=np.ceil((trange[1]-trange[0])/chunksz)) if
(trange[1]-trange[0])>chunksz else
np.array(trange))
for i,t in enumerate(chunks[:-1]):
exptime += exposure(band,[chunks[i],chunks[i+1]],verbose=verbose,
retries=retries)
return exptime
def netdead(band,t0,t1,tstep=1.,refrate=79.,verbose=0): print 'Time range: ['+str(t0)+', '+str(t1)+']' refrate = 79. # counts per second, nominal stim rate feeclkratio = 0.966 # not sure what detector property this adjusts for tec2fdead = 5.52e-6 # TEC to deadtime correction (Method 2) stimcount = gQuery.getValue(gQuery.stimcount(band,t0,t1)) totcount = (gQuery.getValue(gQuery.deadtime1(band,t0,t1)) + gQuery.getValue(gQuery.deadtime2(band,t0,t1))) exptime = t1-t0 # Method 0 # Empirical formula dead0 = tec2fdead*(totcount/exptime)/feeclkratio minrate,maxrate = refrate*.4,refrate+2. # Method 2 # Direct measurement of stims bins = np.linspace(0.,exptime-exptime%tstep,exptime//tstep+1)+t0 h = np.zeros(len(bins)) for i,t in enumerate(bins): print_inline(t1-t) h[i] = gQuery.getValue(gQuery.stimcount(band,t,t+tstep-0.0001)) #h,xh = np.histogram(stimt-trange[0],bins=bins) ix = ((h<=maxrate) & (h>=minrate)).nonzero()[0] dead2 = (1.-((h[ix]/tstep)/feeclkratio)/refrate).mean() # Method 1 # Empirical formula except with counts binned into 1s h = np.zeros(len(bins)) for i,t in enumerate(bins): print_inline(t1-t) h[i] = gQuery.getValue(gQuery.deadtime1(band,t,t+tstep-0.0001)) + gQuery.getValue(gQuery.deadtime2(band,t,t+tstep-0.0001)) dead1 = (tec2fdead*(h/tstep)/feeclkratio).mean() expt = compute_exptime(band,[t0,t1]) print dead0,dead1,dead2 return [t0, t1, dead0, dead1, dead2, expt, stimcount, totcount]
def fGetTimeRanges(band,skypos,trange=None,detsize=1.1,verbose=0,
maxgap=1.,minexp=1.,retries=100.,predicted=False,
skyrange=None,maxgap_override=False):
"""Find the contiguous time ranges within a time range at a
specific location.
minexp - Do not include exposure time less than this.
maxgap - Gaps in exposure longer than this initiate a new time range.
detsize - Fiddle with this if you want to exlude the edges of the
detector.
predicted - Use the aspect solutions to estimate what exposure will be
available once the database is fully populated.
"""
times = get_valid_times(band,skypos,trange=trange,detsize=detsize,
verbose=verbose,retries=retries,skyrange=skyrange)
if not len(times):
return np.array([],dtype='float64')
if verbose:
print_inline('Parsing ~'+str(len(times)-1)+' seconds of raw exposure.')
# NOTE: The minimum meaningful maxgap is 1 second.
if maxgap<1 and not maxgap_override:
raise 'maxgap must be >=1 second'
tranges = distinct_tranges(times,maxgap=maxgap)
ix = np.where(np.array(tranges)[:,1]-np.array(tranges)[:,0]>=minexp)
tranges = np.array(tranges)[ix].tolist()
return np.array(tranges,dtype='float64')
def compute_exptime(band,trange,verbose=0,skypos=None,detsize=1.25,
retries=20,coadd=False):
"""Compute the effective exposure time."""
# FIXME: This skypos[] check appears to not work properly and leads
# to dramatic _underestimates_ of the exposure time.
if skypos and coadd:
tranges = fGetTimeRanges(band,skypos,verbose=verbose,trange=trange,
retries=retries)
else:
tranges=[trange]
if verbose>1:
print_inline('Computing exposure within {tr}'.format(tr=tranges))
if skypos and coadd:
print 'Based on skypos {sp}'.format(sp=skypos)
exptime = 0.
for trange in tranges:
exptime += exposure(band,[trange[0],trange[1]],verbose=verbose,
retries=retries)
return exptime
def netdead(band, t0, t1, tstep=1., refrate=79., verbose=0):
print 'Time range: [' + str(t0) + ', ' + str(t1) + ']'
refrate = 79. # counts per second, nominal stim rate
feeclkratio = 0.966 # not sure what detector property this adjusts for
tec2fdead = 5.52e-6 # TEC to deadtime correction (Method 2)
stimcount = gQuery.getValue(gQuery.stimcount(band, t0, t1))
totcount = (gQuery.getValue(gQuery.deadtime1(band, t0, t1)) +
gQuery.getValue(gQuery.deadtime2(band, t0, t1)))
exptime = t1 - t0
# Method 0
# Empirical formula
dead0 = tec2fdead * (totcount / exptime) / feeclkratio
minrate, maxrate = refrate * .4, refrate + 2.
# Method 2
# Direct measurement of stims
bins = np.linspace(0., exptime - exptime % tstep,
exptime // tstep + 1) + t0
h = np.zeros(len(bins))
for i, t in enumerate(bins):
print_inline(t1 - t)
h[i] = gQuery.getValue(gQuery.stimcount(band, t, t + tstep - 0.0001))
#h,xh = np.histogram(stimt-trange[0],bins=bins)
ix = ((h <= maxrate) & (h >= minrate)).nonzero()[0]
dead2 = (1. - ((h[ix] / tstep) / feeclkratio) / refrate).mean()
# Method 1
# Empirical formula except with counts binned into 1s
h = np.zeros(len(bins))
for i, t in enumerate(bins):
print_inline(t1 - t)
h[i] = gQuery.getValue(gQuery.deadtime1(
band, t, t + tstep - 0.0001)) + gQuery.getValue(
gQuery.deadtime2(band, t, t + tstep - 0.0001))
dead1 = (tec2fdead * (h / tstep) / feeclkratio).mean()
expt = compute_exptime(band, [t0, t1])
print dead0, dead1, dead2
return [t0, t1, dead0, dead1, dead2, expt, stimcount, totcount]
def fGetTimeRanges(band,skypos,trange=None,tscale=1000.,detsize=1.25,verbose=0,
maxgap=1.,minexp=1.,retries=100.,predicted=False):
"""Find the contiguous time ranges within a time range at a
specific location.
minexp - Do not include exposure time less than this.
maxgap - Gaps in exposure longer than this initiate a new time range.
detsize - Fiddle with this if you want to exlude the edges of the
detector.
predicted - Use the aspect solutions to estimate what exposure will be
available once the database is fully populated.
"""
try:
if not np.array(trange).tolist():
trange = [1,1000000000000]
if len(np.shape(trange))==2:
trange=trange[0]
times = (np.array(gQuery.getArray(gQuery.exposure_ranges(band,
skypos[0],skypos[1],t0=trange[0],t1=trange[1],detsize=detsize,
tscale=tscale),verbose=verbose,retries=retries),
dtype='float64')[:,0]/tscale
if not predicted else get_aspect(band,skypos,trange,
tscale=tscale,verbose=verbose)['t'])
except IndexError:
if verbose:
print "No exposure time available at {pos}".format(pos=skypos)
return np.array([],dtype='float64')
except TypeError:
print "Is one of the inputs malformed?"
raise
except:
raise
if verbose:
print_inline('Parsing '+str(len(times)-1)+' seconds of exposure.: ['+str(trange[0])+', '+str(trange[1])+']')
blah = []
for i in xrange(len(times[0:-1])):
blah.append(times[i+1]-times[i])
# A drop in data with duration greater than maxgap initiates a
# new exposure
gaps = np.where(np.array(blah)>maxgap)
ngaps = len(gaps[0])
chunks = []
for i in range(ngaps):
if not i:
chunk = [times[0],times[gaps[0][i]]]
elif i==ngaps-1:
chunk = [times[gaps[0][i]+1],times[-1]]
else:
chunk = [times[gaps[0][i]+1],times[gaps[0][i+1]]]
# If the duration of this slice is less than minexp, do not
# count it as valid exposure.
if chunk[1]-chunk[0]<minexp:
continue
else:
chunks.append(chunk)
if not ngaps:
if times.min()==times.max():
chunks.append([times.min(),times.min()+1])
else:
chunks.append([times.min(),times.max()])
return np.array(chunks,dtype='float64')
