# bolton@utah@iac 2014mayo
# Set this:
# export CAP_DATA_DIR=/data/CAP/bad
# Set the filename:
# Need to loop over 1--5 somehow...
fname = os.getenv('CAP_DATA_DIR') + '/ndArch-nsc5-bad00.fits'
# Get the data:
data, baselines, infodict = io.read_ndArch(fname)
# Build the log-lambda baseline and pixel boundaries:
hires_loglam_air = infodict['coeff0'] + infodict['coeff1'] * n.arange(infodict['nwave'])
hires_logbound_air = pxs.cen2bound(hires_loglam_air)
hires_wave_air = 10.**hires_loglam_air
hires_wavebound_air = 10.**hires_logbound_air
hires_dwave_air = hires_wavebound_air[1:] - hires_wavebound_air[:-1]
# Transform the baselines to vacuum wavelengths:
hires_wave = a2v.a2v(hires_wave_air)
hires_wavebound = a2v.a2v(hires_wavebound_air)
hires_dwave = hires_wavebound[1:] - hires_wavebound[:-1]
# Not sure what the interpretation is at wavelengths below 2000ang...
# We will probably cut that out for now anyway, since these are the
# first-pass redshift templates that we are building.
# Now, we want a flattened view (not copy) of the hi-res data:
npars = data.size // infodict['nwave']
# Exponent to divide out: exp_div = 12 ssp_flam /= 10.**exp_div # Get a subset of the templates: izero = n.argmin(n.abs(n.log10(ssp_agegyr) - (-2.5))) idx = 10 * n.arange(15) + izero n.log10(ssp_agegyr[idx]) nsub_age = len(idx) # Work out the pixel boundaries and widths # (we go to log-lambda because the # spacing is uniform in that): ssp_loglam = n.log10(ssp_wave) ssp_logbound = pxs.cen2bound(ssp_loglam) ssp_wavebound = 10.**ssp_logbound #ssp_dwave = ssp_wavebound[1:] - ssp_wavebound[:-1] # Note that there is some slight unsmoothness in the # width of the sampling intervals. I think this is fine, # but it will imply a slightly variable resolution # implicit in the data as it currently stands if we dial # it in straight from the pixel widths. So instead, let's # try setting it based on the average loglam spacing, # since that is essentially uniform. # p.plot(ssp_logbound[1:] - ssp_logbound[:-1], hold=False) dloglam = n.mean(ssp_logbound[1:] - ssp_logbound[:-1]) # This will work: #ssp_dwave = 10.**(ssp_loglam + 0.5 * dloglam) - \ # 10.**(ssp_loglam - 0.5 * dloglam)
# Exponent to divide out: exp_div = 12 ssp_flam /= 10.**exp_div # Get a subset of the templates: izero = n.argmin(n.abs(n.log10(ssp_agegyr) - (-2.5))) idx = 10 * n.arange(15) + izero n.log10(ssp_agegyr[idx]) nsub_age = len(idx) # Work out the pixel boundaries and widths # (we go to log-lambda because the # spacing is uniform in that): ssp_loglam = n.log10(ssp_wave) ssp_logbound = pxs.cen2bound(ssp_loglam) ssp_wavebound = 10.**ssp_logbound #ssp_dwave = ssp_wavebound[1:] - ssp_wavebound[:-1] # Note that there is some slight unsmoothness in the # width of the sampling intervals. I think this is fine, # but it will imply a slightly variable resolution # implicit in the data as it currently stands if we dial # it in straight from the pixel widths. So instead, let's # try setting it based on the average loglam spacing, # since that is essentially uniform. # p.plot(ssp_logbound[1:] - ssp_logbound[:-1], hold=False) dloglam = n.mean(ssp_logbound[1:] - ssp_logbound[:-1]) # This will work: #ssp_dwave = 10.**(ssp_loglam + 0.5 * dloglam) - \ # 10.**(ssp_loglam - 0.5 * dloglam)
# Exponent to divide out: exp_div = 12 ssp_flam /= 10.**exp_div # Get a subset of the templates: izero = n.argmin(n.abs(n.log10(ssp_agegyr) - (-2.5))) idx = 10 * n.arange(15) + izero n.log10(ssp_agegyr[idx]) nsub_age = len(idx) # Work out the pixel boundaries and widths # (we go to log-lambda because the # spacing is uniform in that): ssp_loglam = n.log10(ssp_wave) ssp_logbound = pxs.cen2bound(ssp_loglam) ssp_wavebound = 10.**ssp_logbound #ssp_dwave = ssp_wavebound[1:] - ssp_wavebound[:-1] # Note that there is some slight unsmoothness in the # width of the sampling intervals. I think this is fine, # but it will imply a slightly variable resolution # implicit in the data as it currently stands if we dial # it in straight from the pixel widths. So instead, let's # try setting it based on the average loglam spacing, # since that is essentially uniform. # p.plot(ssp_logbound[1:] - ssp_logbound[:-1], hold=False) dloglam = n.mean(ssp_logbound[1:] - ssp_logbound[:-1]) # This will work: #ssp_dwave = 10.**(ssp_loglam + 0.5 * dloglam) - \ # 10.**(ssp_loglam - 0.5 * dloglam)
# bolton@utah@iac 2014mayo
# Set this:
# export CAP_DATA_DIR=/data/CAP/bad
# Set the filename:
# Need to loop over 1--5 somehow...
fname = os.getenv('CAP_DATA_DIR') + '/ndArch-nsc5-bad00.fits'
# Get the data:
data, baselines, infodict = io.read_ndArch(fname)
# Build the log-lambda baseline and pixel boundaries:
hires_loglam_air = infodict['coeff0'] + infodict['coeff1'] * n.arange(
infodict['nwave'])
hires_logbound_air = pxs.cen2bound(hires_loglam_air)
hires_wave_air = 10.**hires_loglam_air
hires_wavebound_air = 10.**hires_logbound_air
hires_dwave_air = hires_wavebound_air[1:] - hires_wavebound_air[:-1]
# Transform the baselines to vacuum wavelengths:
hires_wave = a2v.a2v(hires_wave_air)
hires_wavebound = a2v.a2v(hires_wavebound_air)
hires_dwave = hires_wavebound[1:] - hires_wavebound[:-1]
# Not sure what the interpretation is at wavelengths below 2000ang...
# We will probably cut that out for now anyway, since these are the
# first-pass redshift templates that we are building.
# Now, we want a flattened view (not copy) of the hi-res data:
npars = data.size // infodict['nwave']
