def append_lambda_to_xml(file_name, eos_name, file_name_out=None):
param_list = lalsimutils.xml_to_ChooseWaveformParams_array(file_name)
from gwemlightcurves.KNModels import table
eos, eos_fam = table.get_lalsim_eos(eos_name)
print("Writing to xml:")
print("Event m1 m2 lambda1 lambda2")
for i in np.arange(len(param_list)):
m1 = param_list[i].m1
m2 = param_list[i].m2
param_list[i].lambda1 = calc_lambda_from_m(m1, eos_fam)
param_list[i].lambda2 = calc_lambda_from_m(m2, eos_fam)
print(i, "[", param_list[i].m1 / lal.MSUN_SI,
param_list[i].m2 / lal.MSUN_SI, param_list[i].lambda1,
param_list[i].lambda2, "]")
if file_name_out is None:
file_name_out = file_name.replace(".xml.gz", "") + "_" + eos_name
lalsimutils.ChooseWaveformParams_array_to_xml(param_list,
fname=file_name_out)
#
# USAGE
# python util_TestXMLParameterStorage.py ; ligolw_print -t sim_inspiral -c alpha5 output.xml.gz
import RIFT.lalsimutils as lalsimutils
import numpy as np
import lal
import argparse
parser = argparse.ArgumentParser()
parser.add_argument("--output-file", default="output")
opts = parser.parse_args()
P_list_in = []
for i in np.arange(5):
P_list_in.append(lalsimutils.ChooseWaveformParams(m1=lal.MSUN_SI *
(i + 1)))
P_list_in[-1].lambda1 = 3 * i + 1
P_list_in[-1].lambda2 = 3 * i + 2
P_list_in[-1].approx = 3 * i + 2
P_list_in[-1].print_params()
lalsimutils.ChooseWaveformParams_array_to_xml(P_list_in,
fname=opts.output_file)
print(" ------ ")
P_list_out = lalsimutils.xml_to_ChooseWaveformParams_array(opts.output_file +
".xml.gz")
for P in P_list_out:
P.print_params()
# Loop over everything, and find best point
best_fits_here = []
closest_so_far_d = 1e10
closest_so_far = ()
for group in glist:
# if opts.verbose:
# print " Checking ", group
res = nr_closest_to(P, dist_here, group) # pair of values returned
if res[1] < closest_so_far_d:
closest_so_far = (group, res[0])
closest_so_far_d = res[1]
if opts.verbose:
print(" NEW CLOSEST SIMULATION FOR THIS POSTERIOR POINT",
closest_so_far, closest_so_far_d)
if closest_so_far_d > opts.cut:
P_list_retain.append(P)
dist_list_retain.append(closest_so_far_d)
else:
print(
" Skipping point shown below as too close to simulations because the closest has distance",
closest_so_far_d, " which is less than ", opts.cut)
P.print_params()
lalsimutils.ChooseWaveformParams_array_to_xml(P_list_retain,
fname=opts.fname_output,
fref=P.fref)
np.savetxt(opts.fname_output + "_distances.dat",
np.array([dist_list_retain]).T)
val = val/ lal.MSUN_SI
if val < downselect_dict[param][0] or val > downselect_dict[param][1]:
include_item =False
if include_item:
P_out.append(P)
# Randomize parameters that have been requested to be randomized
# - note there is NO SANITY CHECKING if you do this
# - target: tidal parameters, more efficiently hammer on low-tide corner if necessary
if len(opts.random_parameter) >0:
random_ranges = {}
for indx in np.arange(len(opts.random_parameter)):
param = opts.random_parameter[indx]
random_ranges[param] = np.array(eval(opts.random_parameter_range[indx]))
for P in P_out:
for param in opts.random_parameter:
val = np.random.uniform( random_ranges[param][0], random_ranges[param][1])
if param in ['mc','m1','m2','mtot']:
val = val* lal.MSUN_SI
P.assign_param(param,val)
# Export
lalsimutils.ChooseWaveformParams_array_to_xml(P_out,fname=opts.inj_file_out,fref=P.fref)
P.m1 = event_dict["m1"]*lal.MSUN_SI; P.m2=event_dict["m2"]*lal.MSUN_SI; P.s1z = event_dict["s1z"]; P.s2z = event_dict["s2z"]
# Load in ini file to select relevant fmin, fref [latter usually unused]
config = ConfigParser.ConfigParser()
config.read(opts.use_ini)
fmin_vals ={}
fmin_fiducial = -1
ifo_list = eval(config.get('analysis','ifos'))
for ifo in ifo_list:
fmin_vals[ifo] = unsafe_config_get(config,['lalinference','flow'])[ifo]
fmin_fiducial = fmin_vals[ifo]
event_dict["IFOs"] = ifo_list
print( "IFO list from ini ", ifo_list)
P.fmin = fmin_fiducial
P.fref = unsafe_config_get(config,['engine','fref'])
# Write 'target_params.xml.gz' file
lalsimutils.ChooseWaveformParams_array_to_xml([P], "target_params")
helper_psd_args = ''
srate=4096 # default, built into helper, unwise to go lower, LI will almost never do higher
if opts.make_bw_psds:
helper_psd_args += " --assume-fiducial-psd-files --fmax " + str(srate/2-1)
# Create provenance info : we want run to be reproducible
if True:
os.mkdir("reproducibility")
# Write this script and its arguments
import shutil, json
# thisfile = os.path.realpath(__file__)
if otps.verbose:
print(downselect_dict)
# downselection procedure: fix units so I can downselect on mass parameters
for p in ['mc', 'm1', 'm2', 'mtot']:
if p in downselect_dict.keys():
downselect_dict[p] = np.array(downselect_dict[p], dtype=np.float64)
downselect_dict[p] *= lal.MSUN_SI # add back units
P_list_in = lalsimutils.xml_to_ChooseWaveformParams_array(opts.fname)
P_inj = lalsimutils.xml_to_ChooseWaveformParams_array(opts.fname)[opts.event]
param_ref_vals = {}
for param in downselect_dict.keys():
param_ref_vals[param] = P_inj.extract_param(param)
P_list = []
for P in P_list_in:
include_item = True
for param in downselect_dict.keys():
param_here = P.extract_param(param)
if not (param_here > downselect_dict[param][0]
and param_here < downselect_dict[param][1]):
include_item = False
continue
if include_item:
P_list.append(P)
lalsimutils.ChooseWaveformParams_array_to_xml(
P_list, opts.fname_out.replace(".xml.gz",
"")) # don't write the postfix twice
if fstart_NR < P.fmin or not opts.require_fmin_above_NR_start:
# only add mass point if a valid mass choice, given NR starting frequency.
grid = grid + [newline]
print(" ---- DONE WITH GRID SETUP --- ")
print(" grid points # = ", len(grid))
grid_out, P_list = evaluate_overlap_on_grid(hfBase, param_names, grid)
if len(grid_out) == 0:
print(" No points survive....")
###
### Optional: Write grid to XML file (ONLY if using cutoff option)
###
lalsimutils.ChooseWaveformParams_array_to_xml(P_list,
fname=opts.fname,
fref=P.fref)
###
### Write output to text file: p1 p2 p3 ... overlap, only including named params
###
headline = ' '.join(param_names + ['ip'])
if int(
np.version.short_version.split('.')[1]
) >= 7: # http://docs.scipy.org/doc/numpy/reference/generated/numpy.savetxt.html
np.savetxt(opts.fname + ".dat", grid_out, header=headline)
else:
np.savetxt(opts.fname + ".dat", grid_out) #
###
### Optional: Scatterplot
else:
full_spoke[key] = [[mtot, lnLhere, sigma_here]]
###
### Create ILE job to follow it up
###
# Create xml file with correct mass and NR parameters
m1, m2 = best_matches_masses[(opts.nr_group, opts.nr_param)]
wfP = nrwf.WaveformModeCatalog(opts.nr_group,
opts.nr_param,
metadata_only=True)
wfP.P.assign_param('mtot', (m1 + m2) * lal.MSUN_SI)
wfP.P.print_params()
P_list = [wfP.P]
lalsimutils.ChooseWaveformParams_array_to_xml(
P_list, fname="single-pt") # choice of fref irrelevant
with open(str(opts.run_dir) + "/command-single.sh", 'r') as runfile:
rf = str(runfile.readlines()[1])
rf = rf.replace('create_event_dag_via_grid',
'integrate_likelihood_extrinsic')
rf += " --maximize-only"
rf = rf.split()
rf[rf.index("--sim-xml") + 1] = "single-pt.xml.gz"
rf[rf.index("--output-file") + 1] = "ILE-single.xml.gz"
if "--n-copies" in rf:
rf[rf.index("--n-copies") + 1] = ""
if "--n-max" in rf:
rf[rf.index("--n-max") + 1] = " 10000 "
rf_submit = ' '.join(rf)
if "--n-copies" in rf:
# Loop, removing duplicates sequentially
P_list_out = []
n = len(P_list)
indxList = np.arange(n - 1)
print(indxList)
param_names = opts.compare_parameter
for indx in indxList:
print(" Testing ", indx, "out of ", n)
do_we_add_it = True
for indx2 in np.arange(indx + 1, n):
are_all_parameters_the_same = True
for param in param_names:
param_val = P_list[indx].extract_param(param)
param_val2 = P_list[indx2].extract_param(param)
if rosDebug:
print(param, param_val, param_val2)
if not (param_val == param_val2):
are_all_parameters_the_same = False
# conclude test for this index
if are_all_parameters_the_same:
do_we_add_it = False
if rosDebug:
print(" Match between ", indx, " and ", indx2)
if do_we_add_it:
P_list_out = P_list_out + [P_list[indx]]
# Write output
print(" Saving ", len(P_list_out))
lalsimutils.ChooseWaveformParams_array_to_xml(
P_list_out, fname=opts.output_file) # note xml.gz is appended
nCount +=1
# Cross-look-up
try:
P_sample = sd_P[spoke_id][0] # if this fails
P_sample.waveFlags =None
P_sample.nonGRparams = None
P_sample.print_params()
except:
nFailures +=1
print(" Failed cross lookup for ", spoke_id, nCount, " failure count = ", nFailures)
continue
# Clean
sd_here =spokes.CleanSpokeEntries(sd_dat[spoke_id])
# Refine: find mass values
code, mvals_new = spokes.Refine(sd_here[:,0], sd_here[:,1])
if mvals_new is None:
continue # Failed
mvals_new = np.array(mvals_new)
mvals_new = mvals_new[ mvals_new > 0] # eliminate negtive masses!
print(key, len(sd_here), code, mvals_new)
if code == 'refined' or code =='extended':
for m in mvals_new:
print(m)
P = P_sample.manual_copy()
P.tref = P_sample.tref # BE VERY CAREFUL: The structure swig-bound to store time is NOT a float, and requires careful memory management
P.assign_param('mtot',m*lal.MSUN_SI)
P_list.append(P)
if P_list:
lalsimutils.ChooseWaveformParams_array_to_xml(P_list, opts.save_refinement_fname)
param = parameters_to_center[pIndex]
val = P_ref_list[indx].extract_param(param)
dat_ref[indx, pIndex] = val
P_out = []
dx = np.mean(dat, axis=0) - np.mean(dat_ref, axis=0)
dat_out = []
for indx in np.arange(len(P_list)):
bInclude = True
vec = np.zeros(len(parameters_to_center))
for pIndex in np.arange(len(parameters_to_center)):
param = parameters_to_center[pIndex]
vec[pIndex] = valNew = dat[indx, pIndex] - dx[pIndex]
if param == 'eta':
if valNew > 0.25 or valNew < 0.001:
bInclude = False
continue
if param == 's1z' or param == 's2z':
if valNew > 1 or valNew < -1:
bInclude = False
continue
P_list[indx].assign_param(param, valNew)
P_list[indx].tref = float(P_list[indx].tref)
P_out.append(P_list[indx])
dat_out.append(vec)
print(" Recentering report: last two should be equal : ", np.mean(dat, axis=0),
np.mean(dat_ref, axis=0), np.mean(dat_out, axis=0))
lalsimutils.ChooseWaveformParams_array_to_xml(P_out, "shifted.xml.gz")
