def add_meas_record(self, bl1, pol1, conj1, bl2, pol2, conj2):
#amp_line = n.zeros((self.nprms,), dtype=n.double)
phs_line = n.zeros((self.nprms,), dtype=n.double)
i1,j1 = bl2ij(bl1)
i2,j2 = bl2ij(bl2)
if conj1: i1,j1 = j1,i1
if conj2: i2,j2 = j2,i2
i1,j1 = self.antpol_index(i1,pol1), self.antpol_index(j1,pol1)
i2,j2 = self.antpol_index(i2,pol2), self.antpol_index(j2,pol2)
#amp_line[j2], amp_line[i2], amp_line[j1], amp_line[i1] = 1, 1, -1, -1
#self.M_amp.append(amp_line)
#phs_line[j2], phs_line[i2], phs_line[j1], phs_line[i1] = 1, -1, -1, 1
phs_line[j2] += 1; phs_line[i2] += -1; phs_line[j1] += -1; phs_line[i1] += 1
self.M_phs.append(phs_line)
self.meas_order.append((bl1,pol1,conj1,bl2,pol2,conj2))
def estimate_xtalk(ant_sol, vis, meas_order):
dsum,dwgt = {},{}
for (bl,pol,sep,cnj),d in zip(meas_order,vis):
i,j = bl2ij(bl)
gi,gj = ant_sol[i][pol], ant_sol[j][pol]
if cnj: gij = gi * n.conj(gj)
else: gij = gj * n.conj(gi)
d /= gij
dsum[sep] = dsum.get(sep,0) + d
dwgt[sep] = dwgt.get(sep,0) + 1
xtalk = []
for (bl,pol,sep,cnj),d in zip(meas_order,vis):
i,j = bl2ij(bl)
gi,gj = ant_sol[i][pol], ant_sol[j][pol]
if cnj: gij = gi * n.conj(gj)
else: gij = gj * n.conj(gi)
davg = dsum[sep] / dwgt[sep]
davg *= gij
xtalk.append(d - davg)
return n.array(xtalk)
def add_meas_record(self, bl, pol, sep, conj):
amp_line = n.zeros((self.nprms,), dtype=n.double)
phs_line = n.zeros((self.nprms,), dtype=n.double)
i,j = bl2ij(bl)
if conj: i,j = j,i
s = self.sep_index(sep)
i,j = self.antpol_index(i,pol), self.antpol_index(j,pol)
amp_line[i], amp_line[j], amp_line[s] = 1, 1, 1
self.M_amp.append(amp_line)
phs_line[i], phs_line[j], phs_line[s] = -1, 1, 1
self.M_phs.append(phs_line)
self.meas_order.append((bl,pol,sep,conj))
def apply_gains(data, gains):
rms = data['rms_array']
a1 = np.ndarray(rms.shape)
a2 = np.ndarray(rms.shape)
for bl in trange(rms.shape[0]):
for freq in range(rms.shape[3]):
for pol in range(rms.shape[4]):
(i1, i2) = bl2ij(bl)
i1 = np.argmin(gains['HERA_list'] - i1)
i2 = np.argmin(gains['HERA_list'] - i2)
g1 = np.mean(gains['hera_gains'][pol, i1, np.argmin(np.abs(
gains['freqs'] - data['freq_array'][freq] / 1e6)), np.logical_and(gains['lsts'] < 11, gains['lsts'] > 6)])
g2 = np.mean(gains['hera_gains'][pol, i2, np.argmin(np.abs(
gains['freqs'] - data['freq_array'][freq] / 1e6)), np.logical_and(gains['lsts'] < 11, gains['lsts'] > 6)])
a1[bl, :, 0, freq, pol] = np.tile(g1, (rms.shape[1],))
a2[bl, :, 0, freq, pol] = np.tile(g2, (rms.shape[1],))
tsys = np.divide(rms, np.sqrt(np.multiply(a1, a2)))
tsys[tsys <= 0] = 0
return tsys
def find_peaks(d):
p = n.zeros(d.size)
np = n.argsort(d)[-NDLY:]
while n.any(p != np):
p = np
np = n.unique(n.where(d[p-1] > d[p], p-1, p))
np = n.unique(n.where(d[(np+1)%d.size] > d[np], (np+1)%d.size, np))
return np
bl01 = ij2bl(0,1)
bins = find_peaks(data[bl01])
ok_bls = {bl01: n.where(bins > NCHAN/2, bins-NCHAN, bins)}
NDLY = len(ok_bls[bl01])
if True:
print bl2ij(bl01), bins, dlys_noshift[bins]
p.plot(dlys_noshift, data[bl01])
p.plot(dlys_noshift[bins], data[bl01][bins], '^')
p.show()
# inds = n.argsort(ans,axis=None)[-NDLY**2:]
# best_inds = {}
# for ind in inds:
# x = ind / ans.shape[0]
# y = ind % ans.shape[0]
# _x = None
# for cnt in range(100):
# _y = n.argmax(ans[x])
# _x = n.argmax(ans[:,_y])
# if _x == x: break
def blp2ijp(blp):
bl,pol = int(blp) / 16, (blp % 16) - 9
i,j = miriad.bl2ij(bl)
return i,j,pol
def blp2ijp(blp):
bl, pol = int(blp) / 16, (blp % 16) - 9
i, j = miriad.bl2ij(bl)
return i, j, pol
p = n.zeros(d.size)
np = n.argsort(d)[-NDLY:]
while n.any(p != np):
p = np
np = n.unique(n.where(d[p - 1] > d[p], p - 1, p))
np = n.unique(
n.where(d[(np + 1) % d.size] > d[np], (np + 1) % d.size, np))
return np
bl01 = ij2bl(0, 1)
bins = find_peaks(data[bl01])
ok_bls = {bl01: n.where(bins > NCHAN / 2, bins - NCHAN, bins)}
NDLY = len(ok_bls[bl01])
if True:
print bl2ij(bl01), bins, dlys_noshift[bins]
p.plot(dlys_noshift, data[bl01])
p.plot(dlys_noshift[bins], data[bl01][bins], '^')
p.show()
# inds = n.argsort(ans,axis=None)[-NDLY**2:]
# best_inds = {}
# for ind in inds:
# x = ind / ans.shape[0]
# y = ind % ans.shape[0]
# _x = None
# for cnt in range(100):
# _y = n.argmax(ans[x])
# _x = n.argmax(ans[:,_y])
# if _x == x: break
# else: x = _x
