parts = [11, 12, 13, 14, 15]
# parts = [0,1,2,3,4,5]
# parts = [16,17,18,19,20]
sim_ns = [4, 8, 16, 32, 64, 128]
for p in parts:
for sim_n in sim_ns:
corr_path = f'{scorpy.DATADIR}/dbins/cxi/qcors/sim/{sim_n}/sim{sim_n}-alpha-p{p}-qcor.dbin'
padf_path = f'{scorpy.DATADIR}/dbins/cxi/padfs/sim/{sim_n}/sim{sim_n}-alpha-p{p}-padf.dbin'
print(f'!!!!#######{p=}, {sim_n=}')
print(f'!!!!#######')
print(f'!!!!#######')
corr = scorpy.CorrelationVol(path=corr_path)
padf = scorpy.PadfVol(nr=nr,
npsi=npsi,
rmax=rmax,
nl=nl,
wavelength=wavelength)
padf.fill_from_corr(corr,
theta0=False,
verbose=99,
tag=f'{part}_{sim_n}')
padf.save(padf_path)
print(f'!!!!#######')
print(f'!!!!#######')
# parts = [0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15]
# sim_ns = [256]
method = 'scat_pol'
r = scorpy.utils.convert_q2r(qmax, clen, wavelength * 1e10)
runs = [
108, 113, 109, 125, 110, 123, 118, 112, 119, 120, 102, 104, 105, 103, 121,
126
]
#### SCORPY CORRELATION
for run in runs:
print(f'######Run {run}')
pk = scorpy.PeakData(f'{scorpy.DATADIR}/cxi/run{run}_peaks.txt', qmax=qmax)
corr = scorpy.CorrelationVol(nq=nq, npsi=npsi, qmax=qmax, cos_sample=False)
corr.fill_from_peakdata(pk, method=method)
corr.save(f'{scorpy.DATADIR}/dbins/cxi/qcors/{run}/run{run}-qcor.dbin')
### Seeded Half runs
for seed in range(20):
print(f'Run: {run}')
print(f'Seed: {seed}.')
frames = pk.split_frames()
np.random.seed(seed)
np.random.shuffle(frames)
frames_a = frames[:int(len(frames) / 2)]
frames_b = frames[int(len(frames) / 2):]
corra = scorpy.CorrelationVol(nq, npsi, qmax, cos_sample=False)
#approx 40 mins for 100x100, 20 mins for 50x100, 7hrs for 2000x50
nbatches = 2000
nframes = 50
method = 'scat_sph'
integrated = True
corr2dfname = 'ortho-intenr-qmax15-2d-qcor.dbin'
######3D CORR
if run_3d:
cif = scorpy.CifData(path=f'{scorpy.DATADIR}/xtal/{ciffname}', qmax=qmax)
corr3d = scorpy.CorrelationVol(qmax=qmax,
nq=nq,
npsi=npsi,
cos_sample=cos_sample,
inc_self_corr=inc_self_corr)
corr3d.fill_from_cif(cif, verbose=2, method='scat_sph')
corr3d.save(f'{scorpy.DATADIR}/dbins/{corr3dfname}')
######2D CORR
if run_2d:
corr2d = scorpy.CorrelationVol(nq=nq,
npsi=npsi,
qmax=qmax,
cos_sample=cos_sample,
inc_self_corr=inc_self_corr)
mask_corr_config.write(f'input = {scorpy.DATADIR}cxi/mask/mask_im.dbin\n')
mask_corr_config.write(f'outpath = {scorpy.DATADIR}dbins/cxi/padfcorr/\n')
mask_corr_config.write(f'wavelength = {wavelength}\n')
mask_corr_config.write(f'pixel_width = {1/res}\n')
mask_corr_config.write(f'detector_z = {clen}\n')
mask_corr_config.write(f'nq = {nq}\n')
mask_corr_config.write(
f'nth = {npsi*2}\n'
) # padfcorr give theta from 0-360, so double ntheta so cut in half later
mask_corr_config.write(f'tag = padfcorr_mask\n')
mask_corr_config.write(f'qmax = {qmax*1e9}\n')
mask_corr_config.close()
os.system(
f'{scorpy.PADFCORRDIR}padfcorr {scorpy.DATADIR}dbins/cxi/padfcorr/padfcorr_mask_config.txt'
)
padfcorrvol = np.fromfile(
f'{scorpy.DATADIR}dbins/cxi/padfcorr/padfcorr_mask_correlation.dbin')
padfcorrvol = padfcorrvol.reshape((nq, nq, 2 * npsi))
corr = scorpy.CorrelationVol(nq, npsi, qmax, cos_sample=False)
corr.vol = padfcorrvol[:, :, :npsi]
corr.save(f'{scorpy.DATADIR}dbins/cxi/mask_qcor.dbin')
# corr = scorpy.CorrelationVol(path=f'{scorpy.DATADIR}/cxi/mask/mask_qcor.dbin')
corr.plot_q1q2(xlabel='$\\Delta\\Psi [rad]$', ylabel='$q$ [$\u212b^{-1}$]')
plt.show()
# npixforthispeak = len(np.where(corr2d.vol[loc]>0)[0])
# corr2d_inte.vol[int(inds[0]), int(inds[1]), int(inds[2])] = s
# s, loc = corr3d.integrate_region(point[0], point[1], point[2], inte_rq, inte_rq, inte_rt)
# npixforthispeak = len(np.where(corr3d.vol[loc]>0)[0])
# corr3d_inte.vol[int(inds[0]), int(inds[1]), int(inds[2])] = s
# corr3d_inte.save(f'{scorpy.DATADIR}/dbins/intenr/intenr-qmax0264-3d-inte-qcor.dbin')
# corr2d_inte.save(f'{scorpy.DATADIR}/dbins/intenr/intenr-qmax0264-2d-inte-qcor.dbin')
corr2d_inte = scorpy.CorrelationVol(path=f'{scorpy.DATADIR}/dbins/intenr/intenr-qmax0264-2d-inte-qcor.dbin')
corr2d_inte.qpsi_correction()
corr3d_inte = scorpy.CorrelationVol(path=f'{scorpy.DATADIR}/dbins/intenr/intenr-qmax0264-3d-inte-qcor.dbin')
# corr2d_inte = scorpy.CorrelationVol(path=f'{scorpy.DATADIR}/dbins/inten1/inten1-qmax0264-2d-inte-qcor.dbin')
# corr2d_inte.qpsi_correction()
# corr3d_inte = scorpy.CorrelationVol(path=f'{scorpy.DATADIR}/dbins/inten1/inten1-qmax0264-3d-inte-qcor.dbin')
loc = np.unique(np.where(corr3d_inte.vol >0)[0])
print(loc)
import scorpy
import numpy as np
import matplotlib.pyplot as plt
plt.close('all')
v1 = scorpy.CorrelationVol(
50,
90,
2,
)
v1.vol = np.random.random(v1.vol.shape)
v1.plot_q1q2()
v1.save(scorpy.DATADIR / 'test.dbin')
v2 = scorpy.CorrelationVol(path=scorpy.DATADIR / 'test.dbin')
v2.plot_q1q2()
plt.show()
# corr2.fill_from_cif(cif, method=method)
# corr1.plot_q1q2()
# corr2.plot_q1q2()
method = 'scat_sph'
geom = scorpy.ExpGeom(f'{scorpy.DATADIR}/geoms/plot-test.geom')
pk = scorpy.PeakData(f'{scorpy.DATADIR}/patternsim/plot-test.h5',
geo=geom,
qmax=1)
pk.plot_peaks()
corr1 = scorpy.CorrelationVol(100,
180,
1,
cos_sample=False,
inc_self_corr=True)
corr1.fill_from_peakdata(pk, method=method)
corr2 = scorpy.CorrelationVol(100,
180,
1,
cos_sample=False,
inc_self_corr=False)
corr2.fill_from_peakdata(pk, method=method)
corr1.plot_q1q2()
corr2.plot_q1q2()
plt.show()
# print('Padfcorr')
# os.system(f'{scorpy.PADFCORRDIR}padfcorr /tmp/padfcorr_tmp_config.txt')
# tmppadfcorrvol = np.fromfile(f'/tmp/padfcorr_tmp_frame_correlation.dbin')
# tmppadfcorrvol = tmppadfcorrvol.reshape((nq, nq, 2*npsi))
# pc_corr = scorpy.CorrelationVol(nq, npsi, qmax, cos_sample=False)
# pc_corr.vol = tmppadfcorrvol[:,:, :npsi]
# pc_corr.plot_sumax(0, log=True)
# sc_corr.plot_sumax(0, log=True)
# pc_corr.plot_q1q2( log=True)
# sc_corr.plot_q1q2( log=True)
pc_corr = scorpy.CorrelationVol(
path=f'{scorpy.DATADIR}/dbins/cxi/run118_padfcorr_qcor.dbin')
sc_corr = scorpy.CorrelationVol(
path=f'{scorpy.DATADIR}/dbins/cxi/run118_qcor.dbin')
pc_corr.plot_q1q2(log=True,
title='Run 118 - PADFCORR - log10',
xlabel='$\\Delta \\Psi$',
ylabel='$q_1=q_2$')
sc_corr.plot_q1q2(log=True,
title='Run 118 - SCORPY - log10',
xlabel='$\\Delta \\Psi$',
ylabel='$q_1=q_2$')
# pc_corr.plot_q1q2()
# sc_corr.plot_q1q2()
# pc_corr.plot_sumax(0, log=True)