teff = cat['Teff'][inds]
logg = cat['logg'][inds]
mh = cat['FeH'][inds]
alpham = cat['alphaM'][inds]
ak = 0.05*np.ones(len(inds))
lab = np.vstack((teff,logg,mh,alpham,ak))
ds = dataset.Dataset(
wl, ids, norm_flux, norm_ivar, lab, ids, norm_flux, norm_ivar)
ds.test_label_vals = lab.T
# generate model test spectra
m.infer_spectra(ds)
Cinv = ds.test_ivar / (1 + ds.test_ivar*m.scatters**2)
#res = Cinv*(ds.test_flux - m.model_spectra)**2
res = (ds.test_flux - m.model_spectra)
# get height above the plane
c = SkyCoord(ra, dec, unit='deg')
lat = np.abs(c.icrs.galactic.b)
for ii in range(0, len(ids)):
prefix = ids[ii].split(".")[0]
plot(
ii, wl, norm_flux, norm_ivar, m.model_spectra, m.coeffs,
m.scatters, m.chisq, m.pivots, 6200, 6700,
[5577, 5890, 5896, 6384, 6300, 6363, 6562, 6584],
"%s.png" %prefix)
def run_one_date(date):
# load a spectrum
ds = load_dataset(date)
names = np.array([get_name(val) for val in ds.test_ID])
nobj = len(ds.test_ID)
print("%s obj" %nobj)
inds = np.arange(nobj)
m = load_model()
model_spec = get_model_spectra(ds, m)
resid = get_residuals(ds, m)
print("get data to fit")
x,y,yerr = get_data_to_fit(ds,m,resid)
print("fit Gaussian")
fits = np.array([fit_li(x,yval,yerrval) for yval,yerrval in zip(y,yerr)])
popt = fits[:,0]
pcov = fits[:,1]
# get rid of the failed fits
print("purge failed tests")
lens = np.array([len(val) for val in popt])
bad = np.where(lens==1)[0]
names_keep = np.delete(names, bad)
inds_keep = np.delete(inds, bad)
x_keep = np.delete(x, bad, axis=0)
y_keep = np.delete(y, bad, axis=0)
yerr_keep = np.delete(yerr, bad, axis=0)
med_err = np.median(yerr_keep, axis=1)
popt_keep = np.array([x for x in popt if np.array(x).any() != [0]])
pcov_keep = np.array([x for x in pcov if np.array(x).any() != [0]])
amps = popt_keep[:,0]
amp_err = np.sqrt(pcov_keep[:,0,0])
center = popt_keep[:,1]
width = popt_keep[:,2]
print("find candidates")
keep = select(med_err, amps, amp_err, width)
cands = names_keep[keep]
outf = open('%s_candidates.txt' %date, 'w')
outf.write("%s Candidates Total\n" %len(ds.test_ID))
for val in cands: outf.write("%s.fits\n" %val)
outf.close()
inds_cands = inds_keep[keep]
popt_cands = popt_keep[keep]
pcov_cands = pcov_keep[keep]
x_cands = x_keep[keep]
y_cands = y_keep[keep]
yerr_cands = yerr_keep[keep]
print("first plot")
out = [plot_fit(
popt_val, pcov_val, x_val, y_val, yerr_val,
figname="%s_%s_fit.png" %(date, name_val)) for
popt_val, pcov_val, x_val, y_val, yerr_val, name_val in
zip(popt_cands, pcov_cands, x_cands, y_cands, yerr_cands, cands)]
print("second plot")
out = [plot(
ii, ds.wl, ds.test_flux, ds.test_ivar, model_spec,
m.coeffs, m.scatters, m.chisqs, m.pivots,
figname="%s_%s_spec.png" %(date,name)) for ii,name in zip(inds_cands, cands)]
m.scatters = scat
m.pivots = pivot
m.scales = np.ones(len(pivot))
# labels
labeldir = "/Users/annaho/Github_Repositories/TheCannon/data/LAMOST/Label_Transfer"
inputf = pyfits.open("%s/Ho_et_all_catalog_v2.fits" % labeldir)
cat = inputf[1].data
inputf.close()
inds = np.array([np.where(cat['id'] == val)[0][0] for val in ids])
teff = cat['cannon_teff'][inds]
logg = cat['cannon_logg'][inds]
mh = cat['cannon_m_h'][inds]
alpham = cat['cannon_alpha_m'][inds]
ak = 0.05 * np.ones(len(inds))
lab = np.vstack((teff, logg, mh, alpham, ak))
ds = dataset.Dataset(wl, ids, norm_flux, norm_ivar, lab, ids, norm_flux,
norm_ivar)
ds.test_label_vals = lab.T
# generate model test spectra
m.infer_spectra(ds)
for ii in range(0, len(ids)):
prefix = ids[ii].split(".")[0]
plot(ii, wl, norm_flux, norm_ivar, m.model_spectra, m.coeffs, m.scatters,
m.chisq, m.pivots, 7200, 8000, "%s.png" % prefix)
def run_one_date(date):
# load a spectrum
ds = load_dataset(date)
names = np.array([get_name(val) for val in ds.test_ID])
nobj = len(ds.test_ID)
print("%s obj" % nobj)
inds = np.arange(nobj)
m = load_model()
model_spec = get_model_spectra(ds, m)
resid = get_residuals(ds, m)
print("get data to fit")
x, y, yerr = get_data_to_fit(ds, m, resid)
print("fit Gaussian")
fits = np.array(
[fit_li(x, yval, yerrval) for yval, yerrval in zip(y, yerr)])
popt = fits[:, 0]
pcov = fits[:, 1]
# get rid of the failed fits
print("purge failed tests")
lens = np.array([len(val) for val in popt])
bad = np.where(lens == 1)[0]
names_keep = np.delete(names, bad)
inds_keep = np.delete(inds, bad)
x_keep = np.delete(x, bad, axis=0)
y_keep = np.delete(y, bad, axis=0)
yerr_keep = np.delete(yerr, bad, axis=0)
med_err = np.median(yerr_keep, axis=1)
popt_keep = np.array([x for x in popt if np.array(x).any() != [0]])
pcov_keep = np.array([x for x in pcov if np.array(x).any() != [0]])
amps = popt_keep[:, 0]
amp_err = np.sqrt(pcov_keep[:, 0, 0])
center = popt_keep[:, 1]
width = popt_keep[:, 2]
print("find candidates")
keep = select(med_err, amps, amp_err, width)
cands = names_keep[keep]
outf = open('%s_candidates.txt' % date, 'w')
outf.write("%s Candidates Total\n" % len(ds.test_ID))
for val in cands:
outf.write("%s.fits\n" % val)
outf.close()
inds_cands = inds_keep[keep]
popt_cands = popt_keep[keep]
pcov_cands = pcov_keep[keep]
x_cands = x_keep[keep]
y_cands = y_keep[keep]
yerr_cands = yerr_keep[keep]
print("first plot")
out = [
plot_fit(popt_val,
pcov_val,
x_val,
y_val,
yerr_val,
figname="%s_%s_fit.png" % (date, name_val))
for popt_val, pcov_val, x_val, y_val, yerr_val, name_val in zip(
popt_cands, pcov_cands, x_cands, y_cands, yerr_cands, cands)
]
print("second plot")
out = [
plot(ii,
ds.wl,
ds.test_flux,
ds.test_ivar,
model_spec,
m.coeffs,
m.scatters,
m.chisqs,
m.pivots,
figname="%s_%s_spec.png" % (date, name))
for ii, name in zip(inds_cands, cands)
]
m.scales = np.ones(len(pivot))
# labels
labeldir = "/Users/annaho/Github_Repositories/TheCannon/data/LAMOST/Label_Transfer"
inputf = pyfits.open("%s/Ho_et_all_catalog_v2.fits" %labeldir)
cat = inputf[1].data
inputf.close()
inds = np.array([np.where(cat['id']==val)[0][0] for val in ids])
teff = cat['cannon_teff'][inds]
logg = cat['cannon_logg'][inds]
mh = cat['cannon_m_h'][inds]
alpham = cat['cannon_alpha_m'][inds]
ak = 0.05*np.ones(len(inds))
lab = np.vstack((teff,logg,mh,alpham,ak))
ds = dataset.Dataset(
wl, ids, norm_flux, norm_ivar, lab, ids, norm_flux, norm_ivar)
ds.test_label_vals = lab.T
# generate model test spectra
m.infer_spectra(ds)
for ii in range(0,len(ids)):
prefix = ids[ii].split(".")[0]
plot(
ii, wl, norm_flux, norm_ivar,
m.model_spectra, m.coeffs, m.scatters, m.chisq, m.pivots,
7200, 8000, "%s.png" %prefix)