lalinference_pipe_utils.py 160 KB
Newer Older
1

2
#flow DAG Class definitions for LALInference Pipeline
3
# (C) 2012 John Veitch, Vivien Raymond, Kiersten Ruisard, Kan Wang
4

John Douglas Veitch's avatar
John Douglas Veitch committed
5
import itertools
6
import glue
7 8 9
from glue import pipeline
from ligo import segments
from ligo.segments import utils as segmentsUtils
10 11
from glue.ligolw import ligolw, lsctables
from glue.ligolw import utils as ligolw_utils
12
import os
13
import socket
14 15
import uuid
import ast
16
import pdb
John Douglas Veitch's avatar
John Douglas Veitch committed
17
import string
18
from math import floor,ceil,log,pow
19
import sys
20
import random
21
from itertools import permutations
22
import shutil
23
import numpy as np
24
from glob import glob
25
import math
26
from six.moves import range
27
from six import next
28
from functools import reduce
29 30 31 32 33

# We use the GLUE pipeline utilities to construct classes for each
# type of job. Each class has inputs and outputs, which are used to
# join together types of jobs into a DAG.

34 35 36 37 38
def findSegmentsToAnalyze(ifo, frametype, state_vector_channel, bits, gpsstart, gpsend):
    """Return list of segments whose data quality is good enough for PE. The data
    quality is examined with statevector in frame files. If frame files do not
    exist, return empty list.

39 40
    Parameters
    ----
41 42 43 44 45 46 47 48
    ifo: string
    frametype: string
    state_vector_channel: string
    bits: list of string
        List of bits. This function extracts the data taken when all of the
        bits in this list are "active" assuming such data is good enough for
        PE.
    gpsstart, gpsend: float
49 50 51
        GPS period to analyse
    """
    try:
52 53 54 55
        from glue.lal import Cache
        from gwdatafind import find_urls
        import gwpy
        from gwpy.timeseries import StateVector
56
    except ImportError:
57
        print('Unable to import necessary modules. Querying science segments not possible. Please try installing gwdatafind and gwpy')
58
        raise
59 60 61 62 63
    # search for frame file and read its statevector channel
    datacache = Cache.from_urls(find_urls(ifo[0], frametype, gpsstart, gpsend))
    if not datacache:
        return gwpy.segments.SegmentList([])
    flags = gwpy.timeseries.StateVector.read(
64 65
        datacache, state_vector_channel, start=gpsstart, end=gpsend,
        pad=0  # padding data so that errors are not raised even if found data are not continuous.
66 67 68 69 70 71
    ).to_dqflags()
    # extract segments all of whose bits are active
    segments = flags[bits[0]].active
    for bit in bits:
        segments -= ~flags[bit].active
    return segments
72

73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92
def guess_url(fslocation):
    """
    Try to work out the web address of a given path
    """
    SERVER="localhost"
    USER=os.environ['USER']
    HOST=socket.getfqdn()
    if 'public_html' in fslocation:
        k='public_html/'
    elif 'WWW' in fslocation:
        k='WWW/'
    elif 'www_html' in fslocation:
        k='www_html/'
    else:
        k=None
    if k is not None:
        (a,b)=fslocation.split(k)
        webpath=os.path.join('~%s'%(USER),b)
        onweb=True
    else:
93
        (c,d)=fslocation.split(USER,1)
94 95 96 97
        for k in ['public_html','WWW','www_html']:
            trypath=c+os.environ['USER']+'/'+k+d
            #Follow symlinks
            if os.path.realpath(trypath)==os.path.normpath(fslocation):
98 99
                #(a,b)=trypath.split(k)
                webpath=os.path.join('~%s'%(USER),d)
100 101 102 103 104 105 106 107 108 109 110
                onweb=True
                break
            else:
                webpath=fslocation
                onweb=False
    if 'atlas' in HOST:
        url="https://atlas1.atlas.aei.uni-hannover.de/"
    elif 'ligo-wa' in HOST:
        url="https://ldas-jobs.ligo-wa.caltech.edu/"
    elif 'ligo-la' in HOST:
        url="https://ldas-jobs.ligo-la.caltech.edu/"
111 112
    elif 'cit' in HOST or 'caltech' in HOST:
        url="https://ldas-jobs.ligo.caltech.edu/"
113 114 115 116 117 118 119 120 121 122
    elif 'uwm' in HOST or 'nemo' in HOST:
        url="https://ldas-jobs.phys.uwm.edu/"
    elif 'phy.syr.edu' in HOST:
        url="https://sugar-jobs.phy.syr.edu/"
    elif 'arcca.cf.ac.uk' in HOST:
        url="https://geo2.arcca.cf.ac.uk/"
    elif 'vulcan' in HOST:
        url="https://galahad.aei.mpg.de/"
    else:
        if onweb:
123
            url="http://%s/"%(HOST)
124
        else:
125
            url=HOST+':'
126 127 128
    url=url+webpath
    return(url)

John Douglas Veitch's avatar
John Douglas Veitch committed
129
class Event():
130
    """
131
    Represents a unique event to run on
132
    """
133
    new_id=itertools.count()
134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156
    def __init__(self,trig_time=None,SimInspiral=None,SimBurst=None,SnglInspiral=None,CoincInspiral=None,event_id=None,timeslide_dict=None,GID=None,ifos=None, duration=None,srate=None,trigSNR=None,fhigh=None,horizon_distance=None):
        self.trig_time=trig_time
        self.injection=SimInspiral
        self.burstinjection=SimBurst
        self.sngltrigger=SnglInspiral
        if timeslide_dict is None:
            self.timeslides={}
        else:
            self.timeslides=timeslide_dict
        self.GID=GID
        self.coinctrigger=CoincInspiral
        if ifos is None:
            self.ifos = []
        else:
            self.ifos = ifos
        self.duration = duration
        self.srate = srate
        self.trigSNR = trigSNR
        self.fhigh = fhigh
        self.horizon_distance = horizon_distance
        if event_id is not None:
            self.event_id=event_id
        else:
157
            self.event_id=next(Event.new_id)
158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177
        if self.injection is not None:
            self.trig_time=self.injection.get_end()
            if event_id is None: self.event_id=int(str(self.injection.simulation_id).split(':')[2])
        if self.burstinjection is not None:
            self.trig_time=self.burstinjection.get_end()
            if event_id is None: self.event_id=int(str(self.burstinjection.simulation_id).split(':')[2])
        if self.sngltrigger is not None:
            self.trig_time=self.sngltrigger.get_end()
            self.event_id=int(str(self.sngltrigger.event_id).split(':')[2])
        if self.coinctrigger is not None:
            self.trig_time=self.coinctrigger.end_time + 1.0e-9 * self.coinctrigger.end_time_ns
        if self.GID is not None:
            self.event_id=int(''.join(i for i in self.GID if i.isdigit()))
        self.engine_opts={}
    def set_engine_option(self,opt,val):
        """
        Can set event-specific options for the engine nodes
        using this option, e.g. ev.set_engine_option('time-min','1083759273')
        """
        self.engine_opts[opt]=val
John Douglas Veitch's avatar
John Douglas Veitch committed
178 179

dummyCacheNames=['LALLIGO','LALVirgo','LALAdLIGO','LALAdVirgo']
John Douglas Veitch's avatar
John Douglas Veitch committed
180

181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199
def create_events_from_coinc_and_psd(
    coinc_xml_obj, psd_dict, gid=None, threshold_snr=None, flow=20.0, roq=False
):
    """This function calculates seglen, fhigh, srate and horizon distance from
    coinc.xml and psd.xml.gz from GraceDB and create list of Events as input of
    pipeline. This function is based on Chris Pankow's script.

    Parameters
    ----------
    coinc_xml_obj: glue.ligolw.ligolw.Document
        file object of coinc.xml
    psd_dict: dictionary of REAL8FrequencySeries
        PSDs of all the ifos
    threshold_snr: float
        snr threshold for detection
    flow: float
        lower frequecy cutoff for overlap calculation
    roq: bool
        Whether the run uses ROQ or not
200 201 202 203 204 205
    """
    output=[]
    from lal import series as lalseries
    import lal
    from lalsimulation import SimInspiralChirpTimeBound, GetApproximantFromString, IMRPhenomDGetPeakFreq
    from ligo.gracedb.rest import GraceDb, HTTPError
206
    try:
207 208 209
        from gstlal import reference_psd
    except ImportError:
        reference_psd = None
210 211 212 213 214
    try:
        from gwpy.frequencyseries import FrequencySeries
        from gwpy.astro import inspiral_range
    except ImportError:
        inspiral_range = None
215 216
    coinc_events = lsctables.CoincInspiralTable.get_table(coinc_xml_obj)
    sngl_event_idx = dict((row.event_id, row) for row in lsctables.SnglInspiralTable.get_table(coinc_xml_obj))
217 218 219 220 221
    ifos = sorted(coinc_events[0].instruments)
    trigSNR = coinc_events[0].snr
    # Parse PSD
    srate_psdfile=16384
    fhigh=None
222 223 224 225 226 227
    if psd_dict is not None:
        psd = list(psd_dict.values())[0]
        srate_psdfile = pow(
            2.0, ceil(log(psd.f0 + psd.deltaF * (psd.data.length - 1), 2))
        ) * 2
    coinc_map = lsctables.CoincMapTable.get_table(coinc_xml_obj)
228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246
    for coinc in coinc_events:
        these_sngls = [sngl_event_idx[c.event_id] for c in coinc_map if c.coinc_event_id == coinc.coinc_event_id]
        dur=[]
        srate=[]
        horizon_distance=[]
        for e in these_sngls:
            if roq==False:
                chirplen = SimInspiralChirpTimeBound(flow, e.mass1 * lal.MSUN_SI, e.mass2 * lal.MSUN_SI, 0.0, 0.0)
                fstop = IMRPhenomDGetPeakFreq(e.mass1, e.mass2, 0.0, 0.0)
                dur.append(pow(2.0, ceil( log(max(8.0, chirplen + 2.0), 2) ) ) )
                srate.append(pow(2.0, ceil( log(fstop, 2) ) ) * 2)
            # determine horizon distance
            if threshold_snr is not None:
                if e.eff_distance is not None and not math.isnan(e.eff_distance):
                    if e.snr > threshold_snr:
                        horizon_distance.append(e.eff_distance * e.snr / threshold_snr)
                    else:
                        horizon_distance.append(2 * e.eff_distance)
                else:
247 248 249 250 251 252
                    if psd_dict is not None:
                        # Calculate horizon distance from psd to determine
                        # upper limit of distance prior.
                        psd = psd_dict[e.ifo]
                        # If roq is not used, fstop has not been calculated up
                        # to this point.
253 254
                        if not roq==False:
                            fstop = IMRPhenomDGetPeakFreq(e.mass1, e.mass2, 0.0, 0.0)
255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283
                        # reference_psd.HorizonDistance is more precise
                        # calculator of horizon distance than
                        # gwpy.astro.inspiral_range.
                        if reference_psd is not None:
                            HorizonDistanceObj = reference_psd.HorizonDistance(
                                f_min = flow, f_max = fstop, delta_f = 1.0 / 32.0,
                                m1 = e.mass1, m2 = e.mass2
                            )
                            horizon_distance.append(
                                HorizonDistanceObj(psd, snr = threshold_snr)[0]
                            )
                        # If reference_psd is not available, use
                        # gwpy.astro.inspiral_range.
                        elif inspiral_range is not None:
                            gwpy_psd = FrequencySeries(
                                psd.data.data, f0 = psd.f0, df = psd.deltaF
                            )
                            try:
                                horizon_distance.append(
                                    inspiral_range(
                                        gwpy_psd, threshold_snr, e.mass1, e.mass2,
                                        flow, fstop, True
                                    ).value
                                )
                            # If flow of psd is lower than f_ISCO, inspiral_range
                            # raises IndexError. In this case, nothing is
                            # appended to horizon_distance.
                            except IndexError:
                                pass
284 285 286 287 288
        if srate:
            if max(srate)<srate_psdfile:
                srate = max(srate)
            else:
                srate = srate_psdfile
289
                if psd_dict is not None:
290 291 292 293 294 295 296 297 298 299 300
                    fhigh = srate_psdfile/2.0 * 0.95 # Because of the drop-off near Nyquist of the PSD from gstlal
        else:
            srate = None
        if dur:
            duration = max(dur)
        else:
            duration = None
        horizon_distance = max(horizon_distance) if len(horizon_distance) > 0 else None
        ev=Event(CoincInspiral=coinc, GID=gid, ifos = ifos, duration = duration, srate = srate,
                 trigSNR = trigSNR, fhigh = fhigh, horizon_distance=horizon_distance)
        output.append(ev)
301

302 303
    print("Found %d coinc events in table." % len(coinc_events))
    return output
304

305 306 307 308 309
def open_pipedown_database(database_filename,tmp_space):
    """
    Open the connection to the pipedown database
    """
    if not os.access(database_filename,os.R_OK):
310
        raise Exception('Unable to open input file: %s'%(database_filename))
311
    from glue.ligolw import dbtables
312
    import sqlite3
313 314 315
    working_filename=dbtables.get_connection_filename(database_filename,tmp_path=tmp_space)
    connection = sqlite3.connect(working_filename)
    if tmp_space:
316
        dbtables.set_temp_store_directory(connection,tmp_space)
317
    #dbtables.DBTable_set_connection(connection)
318
    return (connection,working_filename)
319

320
def get_zerolag_lloid(database_connection, dumpfile=None, gpsstart=None, gpsend=None, max_cfar=-1, min_cfar=-1):
321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361
    """
    Returns a list of Event objects
    from pipedown data base. Can dump some stats to dumpfile if given,
    and filter by gpsstart and gpsend to reduce the nunmber or specify
    max_cfar to select by combined FAR
    """
    output={}
    if gpsstart is not None: gpsstart=float(gpsstart)
    if gpsend is not None: gpsend=float(gpsend)
    # Get coincs
    get_coincs = "SELECT sngl_inspiral.end_time+sngl_inspiral.end_time_ns*1e-9,sngl_inspiral.ifo,coinc_event.coinc_event_id,sngl_inspiral.snr,sngl_inspiral.chisq,coinc_inspiral.combined_far \
            FROM sngl_inspiral join coinc_event_map on (coinc_event_map.table_name=='sngl_inspiral' and coinc_event_map.event_id ==\
            sngl_inspiral.event_id) join coinc_event on (coinc_event.coinc_event_id==coinc_event_map.coinc_event_id) \
            join coinc_inspiral on (coinc_event.coinc_event_id==coinc_inspiral.coinc_event_id) \
    WHERE coinc_event.time_slide_id=='time_slide:time_slide_id:1'\
            "
    if gpsstart is not None:
        get_coincs=get_coincs+' and coinc_inspiral.end_time+coinc_inspiral.end_time_ns*1.0e-9 > %f'%(gpsstart)
    if gpsend is not None:
        get_coincs=get_coincs+' and coinc_inspiral.end_time+coinc_inspiral.end_time_ns*1.0e-9 < %f'%(gpsend)
    if max_cfar !=-1:
        get_coincs=get_coincs+' and coinc_inspiral.combined_far < %f'%(max_cfar)
    if min_cfar != -1:
        get_coincs=get_coincs+' and coinc_inspiral.combined_far > %f'%(min_cfar)
    db_out=database_connection.cursor().execute(get_coincs)
    extra={}
    for (sngl_time, ifo, coinc_id, snr, chisq, cfar) in db_out:
        coinc_id=int(coinc_id.split(":")[-1])
        if not coinc_id in output.keys():
            output[coinc_id]=Event(trig_time=sngl_time,timeslide_dict={},event_id=int(coinc_id))
            extra[coinc_id]={}
        output[coinc_id].timeslides[ifo]=0
        output[coinc_id].ifos.append(ifo)
        extra[coinc_id][ifo]={'snr':snr,'chisq':chisq,'cfar':cfar}
    if dumpfile is not None:
        fh=open(dumpfile,'w')
        for co in output.keys():
            for ifo in output[co].ifos:
                fh.write('%s %s %s %s %s %s %s\n'%(str(co),ifo,str(output[co].trig_time),str(output[co].timeslides[ifo]),str(extra[co][ifo]['snr']),str(extra[co][ifo]['chisq']),str(extra[co][ifo]['cfar'])))
        fh.close()
    return output.values()
362

363
def get_zerolag_pipedown(database_connection, dumpfile=None, gpsstart=None, gpsend=None, max_cfar=-1, min_cfar=-1):
364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404
    """
    Returns a list of Event objects
    from pipedown data base. Can dump some stats to dumpfile if given,
    and filter by gpsstart and gpsend to reduce the nunmber or specify
    max_cfar to select by combined FAR
    """
    output={}
    if gpsstart is not None: gpsstart=float(gpsstart)
    if gpsend is not None: gpsend=float(gpsend)
    # Get coincs
    get_coincs = "SELECT sngl_inspiral.end_time+sngl_inspiral.end_time_ns*1e-9,sngl_inspiral.ifo,coinc_event.coinc_event_id,sngl_inspiral.snr,sngl_inspiral.chisq,coinc_inspiral.combined_far \
            FROM sngl_inspiral join coinc_event_map on (coinc_event_map.table_name=='sngl_inspiral' and coinc_event_map.event_id ==\
            sngl_inspiral.event_id) join coinc_event on (coinc_event.coinc_event_id==coinc_event_map.coinc_event_id) \
            join coinc_inspiral on (coinc_event.coinc_event_id==coinc_inspiral.coinc_event_id) \
            WHERE coinc_event.time_slide_id=='time_slide:time_slide_id:10049'\
            "
    if gpsstart is not None:
        get_coincs=get_coincs+' and coinc_inspiral.end_time+coinc_inspiral.end_time_ns*1.0e-9 > %f'%(gpsstart)
    if gpsend is not None:
        get_coincs=get_coincs+' and coinc_inspiral.end_time+coinc_inspiral.end_time_ns*1.0e-9 < %f'%(gpsend)
    if max_cfar !=-1:
        get_coincs=get_coincs+' and coinc_inspiral.combined_far < %f'%(max_cfar)
    if min_cfar != -1:
        get_coincs=get_coincs+' and coinc_inspiral.combined_far > %f'%(min_cfar)
    db_out=database_connection.cursor().execute(get_coincs)
    extra={}
    for (sngl_time, ifo, coinc_id, snr, chisq, cfar) in db_out:
        coinc_id=int(coinc_id.split(":")[-1])
        if not coinc_id in output.keys():
            output[coinc_id]=Event(trig_time=sngl_time,timeslide_dict={},event_id=int(coinc_id))
            extra[coinc_id]={}
        output[coinc_id].timeslides[ifo]=0
        output[coinc_id].ifos.append(ifo)
        extra[coinc_id][ifo]={'snr':snr,'chisq':chisq,'cfar':cfar}
    if dumpfile is not None:
        fh=open(dumpfile,'w')
        for co in output.keys():
            for ifo in output[co].ifos:
                fh.write('%s %s %s %s %s %s %s\n'%(str(co),ifo,str(output[co].trig_time),str(output[co].timeslides[ifo]),str(extra[co][ifo]['snr']),str(extra[co][ifo]['chisq']),str(extra[co][ifo]['cfar'])))
        fh.close()
    return output.values()
405

406
def get_timeslides_pipedown(database_connection, dumpfile=None, gpsstart=None, gpsend=None, max_cfar=-1):
407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441
    """
    Returns a list of Event objects
    with times and timeslide offsets
    """
    output={}
    if gpsstart is not None: gpsstart=float(gpsstart)
    if gpsend is not None: gpsend=float(gpsend)
    db_segments=[]
    sql_seg_query="SELECT search_summary.out_start_time, search_summary.out_end_time from search_summary join process on process.process_id==search_summary.process_id where process.program=='thinca'"
    db_out = database_connection.cursor().execute(sql_seg_query)
    for d in db_out:
        if d not in db_segments:
            db_segments.append(d)
    seglist=segments.segmentlist([segments.segment(d[0],d[1]) for d in db_segments])
    db_out_saved=[]
    # Get coincidences
    get_coincs="SELECT sngl_inspiral.end_time+sngl_inspiral.end_time_ns*1e-9,time_slide.offset,sngl_inspiral.ifo,coinc_event.coinc_event_id,sngl_inspiral.snr,sngl_inspiral.chisq,coinc_inspiral.combined_far \
                FROM sngl_inspiral join coinc_event_map on (coinc_event_map.table_name == 'sngl_inspiral' and coinc_event_map.event_id \
                == sngl_inspiral.event_id) join coinc_event on (coinc_event.coinc_event_id==coinc_event_map.coinc_event_id) join time_slide\
                on (time_slide.time_slide_id == coinc_event.time_slide_id and time_slide.instrument==sngl_inspiral.ifo)\
                join coinc_inspiral on (coinc_inspiral.coinc_event_id==coinc_event.coinc_event_id) where coinc_event.time_slide_id!='time_slide:time_slide_id:10049'"
    joinstr = ' and '
    if gpsstart is not None:
        get_coincs=get_coincs+ joinstr + ' coinc_inspiral.end_time+coinc_inspiral.end_time_ns*1e-9 > %f'%(gpsstart)
    if gpsend is not None:
        get_coincs=get_coincs+ joinstr+' coinc_inspiral.end_time+coinc_inspiral.end_time_ns*1e-9 <%f'%(gpsend)
    if max_cfar!=-1:
        get_coincs=get_coincs+joinstr+' coinc_inspiral.combined_far < %f'%(max_cfar)
    db_out=database_connection.cursor().execute(get_coincs)
    # Timeslide functionality requires obsolete pylal - will be removed
    import pylal
    from pylal import SnglInspiralUtils
    extra={}
    for (sngl_time, slide, ifo, coinc_id, snr, chisq, cfar) in db_out:
        coinc_id=int(coinc_id.split(":")[-1])
442
        seg=list(filter(lambda seg:sngl_time in seg,seglist))[0]
443 444 445 446 447 448 449 450 451 452 453 454 455 456
        slid_time = SnglInspiralUtils.slideTimeOnRing(sngl_time,slide,seg)
        if not coinc_id in output.keys():
            output[coinc_id]=Event(trig_time=slid_time,timeslide_dict={},event_id=int(coinc_id))
            extra[coinc_id]={}
        output[coinc_id].timeslides[ifo]=slid_time-sngl_time
        output[coinc_id].ifos.append(ifo)
        extra[coinc_id][ifo]={'snr':snr,'chisq':chisq,'cfar':cfar}
    if dumpfile is not None:
        fh=open(dumpfile,'w')
        for co in output.keys():
            for ifo in output[co].ifos:
                fh.write('%s %s %s %s %s %s %s\n'%(str(co),ifo,str(output[co].trig_time),str(output[co].timeslides[ifo]),str(extra[co][ifo]['snr']),str(extra[co][ifo]['chisq']),str(extra[co][ifo]['cfar'])))
        fh.close()
    return output.values()
457

458
def mkdirs(path):
459 460 461 462 463 464
    """
    Helper function. Make the given directory, creating intermediate
    dirs if necessary, and don't complain about it already existing.
    """
    if os.access(path,os.W_OK) and os.path.isdir(path): return
    else: os.makedirs(path)
465

John Douglas Veitch's avatar
John Douglas Veitch committed
466
def chooseEngineNode(name):
467 468 469 470 471 472 473 474 475 476
    if name=='lalinferencenest':
        return LALInferenceNestNode
    if name=='lalinferenceburst':
        return LALInferenceBurstNode
    if name=='lalinferencemcmc':
        return LALInferenceMCMCNode
    if name=='lalinferencedatadump':
        return LALInferenceDataDumpNode
    if name=='bayeswavepsd':
        return BayesWavePSDNode
477 478
    if name=='bayeswavepost':
        return BayesWavePostNode
479
    return EngineNode
John Douglas Veitch's avatar
John Douglas Veitch committed
480

481 482 483
def get_engine_name(cp):
    name=cp.get('analysis','engine')
    if name=='random':
484
        engine_list=['lalinferencenest','lalinferencemcmc']
485 486 487 488 489 490 491 492 493
        if cp.has_option('input','gid'):
            gid=cp.get('input','gid')
            engine_number=int(''.join(i for i in gid if i.isdigit())) % 2
        else:
            engine_number=random.randint(0,1)
        return engine_list[engine_number]
    else:
        return name

494
def scan_timefile(timefile):
495 496 497 498 499
    import re
    p=re.compile('[\d.]+')
    times=[]
    timefilehandle=open(timefile,'r')
    for time in timefilehandle:
500 501 502 503 504 505 506
        if not p.match(time):
            continue
        if float(time) in times:
            print('Skipping duplicate time %s'%(time))
            continue
        print('Read time %s'%(time))
        times.append(float(time))
507 508
    timefilehandle.close()
    return times
509

510
def get_xml_psds(psdxml,ifos,outpath,end_time=None):
511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532
    """
    Get a psd.xml.gz file and:
    1) Reads it
    2) Checks the psd file contains all the IFO we want to analyze
    3) Writes down the PSDs into an ascii file for each IFO in psd.xml.gz. The name of the file contains the trigtime (if given) and the IFO name.
    Input:
      psdxml: psd.xml.gz file
      ifos: list of ifos used for the analysis
      outpath: path where the ascii PSD will be written to
      (end_time): trigtime for this event. Will be used a part of the PSD file name
    """
    try:
        from lal import series as lalseries
    except ImportError:
        print("ERROR, cannot import lal.series in bppu/get_xml_psds()\n")
        raise

    out={}
    if not os.path.isdir(outpath):
        os.makedirs(outpath)
    if end_time is not None:
        time=repr(float(end_time))
533
    else:
534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555
        time=''
    #check we don't already have ALL the psd files #
    got_all=1
    for ifo in ifos:
        path_to_ascii_psd=os.path.join(outpath,ifo+'_psd_'+time+'.txt')
        # Check we don't already have that ascii (e.g. because we are running parallel runs of the save event
        if os.path.isfile(path_to_ascii_psd):
            got_all*=1
        else:
            got_all*=0
    if got_all==1:
        #print "Already have PSD files. Nothing to do...\n"
        for ifo in ifos:
            out[ifo]=os.path.join(outpath,ifo+'_psd_'+time+'.txt')
        return out

    # We need to convert the PSD for one or more IFOS. Open the file
    if not os.path.isfile(psdxml):
        print("ERROR: impossible to open the psd file %s. Exiting...\n"%psdxml)
        sys.exit(1)
    xmlpsd =  lalseries.read_psd_xmldoc(ligolw_utils.load_filename(psdxml,contenthandler = lalseries.PSDContentHandler))
    # Check the psd file contains all the IFOs we want to analize
556
    for ifo in ifos:
557
        if not ifo in xmlpsd:
558 559 560 561 562
            print("ERROR. The PSD for the ifo %s does not seem to be contained in %s\n"%(ifo,psdxml))
            sys.exit(1)
    #loop over ifos in psd xml file
    for instrument in xmlpsd.keys():
        #name of the ascii file we are going to write the PSD into
563
        path_to_ascii_psd=os.path.join(outpath,instrument+'_psd_'+time+'.txt')
564 565 566 567 568 569 570 571
        # Check we don't already have that ascii (e.g. because we are running parallel runs of the save event
        if os.path.isfile(path_to_ascii_psd):
            continue
        # get data for the IFO
        ifodata=xmlpsd[instrument]
        #check data is not empty
        if ifodata is None:
            continue
572 573
        # write down PSD into an ascii file
        combine = np.c_[ifodata.f0 + np.arange(ifodata.data.length) * ifodata.deltaF, ifodata.data.data]
574 575
        np.savetxt(path_to_ascii_psd,combine)
        # set node.psds dictionary with the path to the ascii files
576
        ifo=instrument
577
        out[ifo]=os.path.join(outpath,ifo+'_psd_'+time+'.txt')
578
    return out
579

580 581 582 583
def get_trigger_chirpmass(coinc_xml_obj):
    coinc_events = lsctables.CoincInspiralTable.get_table(coinc_xml_obj)
    sngl_event_idx = dict((row.event_id, row) for row in lsctables.SnglInspiralTable.get_table(coinc_xml_obj))
    coinc_map = lsctables.CoincMapTable.get_table(coinc_xml_obj)
584 585 586 587 588 589 590 591 592 593 594 595 596 597
    mass1 = []
    mass2 = []
    for coinc in coinc_events:
        these_sngls = [sngl_event_idx[c.event_id] for c in coinc_map if c.coinc_event_id == coinc.coinc_event_id]
        for e in these_sngls:
            mass1.append(e.mass1)
            mass2.append(e.mass2)
    # check that trigger masses are identical in each IFO
    assert len(set(mass1)) == 1
    assert len(set(mass2)) == 1

    mchirp = (mass1[0]*mass2[0])**(3./5.) / ( (mass1[0] + mass2[0])**(1./5.) )

    return mchirp
598

599
def get_roq_mchirp_priors(path, roq_paths, roq_params, key, coinc_xml_obj=None, sim_inspiral=None):
600

601 602 603 604
    ## XML and GID cannot be given at the same time
    ## sim_inspiral must already point at the right row
    mc_priors = {}

605 606
    if coinc_xml_obj is not None and sim_inspiral is not None:
        print("Error in get_roq_mchirp_priors, cannot use both coinc.xml and sim_inspiral\n")
607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623
        sys.exit(1)

    for roq in roq_paths:
        params=os.path.join(path,roq,'params.dat')
        roq_params[roq]=np.genfromtxt(params,names=True)
        mc_priors[roq]=[float(roq_params[roq]['chirpmassmin']),float(roq_params[roq]['chirpmassmax'])]
    ordered_roq_paths=[item[0] for item in sorted(roq_params.items(), key=key)][::-1]
    # below is to construct non-overlapping mc priors for multiple roq mass-bin runs
    '''i=0
    for roq in ordered_roq_paths:
      if i>0:
        # change min, just set to the max of the previous one since we have already aligned it in the previous iteration of this loop
        #mc_priors[roq][0]+= (mc_priors[roq_lengths[i-1]][1]-mc_priors[roq][0])/2.
        mc_priors[roq][0]=mc_priors[ordered_roq_paths[i-1]][1]
      if i<len(roq_paths)-1:
        mc_priors[roq][1]-= (mc_priors[roq][1]- mc_priors[ordered_roq_paths[i+1]][0])/2.
      i+=1'''
624 625
    if coinc_xml_obj is not None:
        trigger_mchirp = get_trigger_chirpmass(coinc_xml_obj)
626
    elif sim_inspiral is not None:
627
        trigger_mchirp = sim_inspiral.mchirp
628 629
    else:
        trigger_mchirp = None
630

631
    return mc_priors, trigger_mchirp
632

633
def get_roq_component_mass_priors(path, roq_paths, roq_params, key, coinc_xml_obj=None, sim_inspiral=None):
634

635
    ## coinc_xml_obj and sim_inspiral cannot be given at the same time
636 637 638
    ## sim_inspiral must already point at the right row
    m1_priors = {}
    m2_priors = {}
639

640 641
    if coinc_xml_obj is not None and sim_inspiral is not None:
        print("Error in get_roq_mchirp_priors, cannot use both coinc.xml and sim_inspiral\n")
642
        sys.exit(1)
643

644 645 646 647 648
    for roq in roq_paths:
        params=os.path.join(path,roq,'params.dat')
        roq_params[roq]=np.genfromtxt(params,names=True)
        m1_priors[roq]=[float(roq_params[roq]['mass1min']),float(roq_params[roq]['mass1max'])]
        m2_priors[roq]=[float(roq_params[roq]['mass2min']),float(roq_params[roq]['mass2max'])]
649

650 651
    if coinc_xml_obj is not None:
        trigger_mchirp = get_trigger_chirpmass(coinc_xml_obj)
652 653 654 655
    elif sim_inspiral is not None:
        trigger_mchirp = sim_inspiral.mchirp
    else:
        trigger_mchirp = None
656

657
    return m1_priors, m2_priors, trigger_mchirp
658

659
def get_roq_mass_freq_scale_factor(mc_priors, trigger_mchirp, force_flow=None):
660 661
    mc_min = min([prange[0] for prange in mc_priors.values()])
    mc_max = max([prange[1] for prange in mc_priors.values()])
662 663 664 665 666 667 668 669 670
    scale_factor = 1.
    if force_flow == None and trigger_mchirp != None:
        if trigger_mchirp >= mc_max:
            scale_factor = 2.**(floor(trigger_mchirp/mc_max))
        if trigger_mchirp <= mc_min:
            scale_factor = (2./3.2)**(ceil(trigger_mchirp/mc_min))
    elif force_flow != None:
        scale_factor = 20./force_flow
    return scale_factor
671

672 673
def create_pfn_tuple(filename,protocol='file://',site='local'):
    return( (os.path.basename(filename),protocol+os.path.abspath(filename),site) )
674

675
def mchirp_from_components(m1, m2):
676
    return (m1*m2)**(3.0/5.0) / (m1+m2)**(1.0/5.0)
677 678

def Query_ROQ_Bounds_Type(path, roq_paths):
679 680 681 682 683 684 685 686 687 688 689 690
    # Assume that parametrization of ROQ bounds is independent of seglen; just look at first one
    import numpy as np
    roq = roq_paths[0]
    params = os.path.join(path,roq,'params.dat')
    roq_params0 = np.genfromtxt(params,names=True)
    roq_names_set = set(roq_params0.dtype.names)
    component_mass_bounds_set = set(['mass1min', 'mass1max', 'mass2min', 'mass2max'])
    chirp_mass_q_bounds_set = set(['chirpmassmin', 'chirpmassmax', 'qmin', 'qmax'])
    if roq_names_set.issuperset(component_mass_bounds_set):
        roq_bounds = 'component_mass'
    elif roq_names_set.issuperset(chirp_mass_q_bounds_set):
        roq_bounds = 'chirp_mass_q'
691
    else:
692 693 694
        print('Invalid bounds for ROQ. Ether (m1,m2) or (mc,q) bounds are supported.')
        sys.exit(1)
    return roq_bounds
695

696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740
class LALInferencePipelineDAG(pipeline.CondorDAG):
    def __init__(self,cp,dax=False,site='local'):
        self.subfiles=[]
        self.config=cp
        self.engine=get_engine_name(cp)
        self.EngineNode=chooseEngineNode(self.engine)
        self.site=site
        if cp.has_option('paths','basedir'):
            self.basepath=cp.get('paths','basedir')
        else:
            self.basepath=os.getcwd()
            print('No basepath specified, using current directory: %s'%(self.basepath))
        mkdirs(self.basepath)
        print("Generating LALInference DAG in {0}".format(self.basepath))
        if dax:
            os.chdir(self.basepath)
        self.posteriorpath=os.path.join(self.basepath,'posterior_samples')
        mkdirs(self.posteriorpath)
        daglogdir=cp.get('paths','daglogdir')
        mkdirs(daglogdir)
        self.daglogfile=os.path.join(daglogdir,'lalinference_pipeline-'+str(uuid.uuid1())+'.log')
        super(LALInferencePipelineDAG,self).__init__(self.daglogfile,dax=dax)
        if cp.has_option('paths','cachedir'):
            self.cachepath=cp.get('paths','cachedir')
        else:
            self.cachepath=os.path.join(self.basepath,'caches')
        mkdirs(self.cachepath)
        if cp.has_option('paths','logdir'):
            self.logpath=cp.get('paths','logdir')
        else:
            self.logpath=os.path.join(self.basepath,'log')
        mkdirs(self.logpath)
        if cp.has_option('analysis','ifos'):
            self.ifos=ast.literal_eval(cp.get('analysis','ifos'))
        else:
            self.ifos=['H1','L1','V1']
        self.segments={}
        if cp.has_option('datafind','veto-categories'):
            self.veto_categories=cp.get('datafind','veto-categories')
        else: self.veto_categories=[]
        for ifo in self.ifos:
            self.segments[ifo]=[]
        self.computeroqweightsnode={}
        self.bayeslinenode={}
        self.bayeswavepsdnode={}
741
        self.bayeswavepostnode={}
742 743 744 745
        self.dq={}
        self.frtypes=ast.literal_eval(cp.get('datafind','types'))
        self.channels=ast.literal_eval(cp.get('data','channels'))
        self.use_available_data=False
746 747 748 749
        if cp.has_option("paths","webdir"):
            self.webdir=cp.get('paths','webdir')
        if cp.has_option("paths","existing_webdir"):
            self.webdir=cp.get('paths','existing_webdir')
750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775
        if cp.has_option('analysis','dataseed'):
            self.dataseed=cp.getint('analysis','dataseed')
        else:
            self.dataseed=None
        # Set up necessary job files.
        self.prenodes={}
        self.datafind_job = pipeline.LSCDataFindJob(self.cachepath,self.logpath,self.config,dax=self.is_dax())
        self.datafind_job.add_opt('url-type','file')
        # If running on OSG use its datafind server
        if cp.has_option('analysis','osg') and cp.getboolean('analysis','osg'):
            self.datafind_job.add_opt('server','datafind.ligo.org')
        if cp.has_option('condor','accounting_group'):
            self.datafind_job.add_condor_cmd('accounting_group',cp.get('condor','accounting_group'))
        if cp.has_option('condor','accounting_group_user'):
            self.datafind_job.add_condor_cmd('accounting_group_user',cp.get('condor','accounting_group_user'))
        self.datafind_job.set_sub_file(os.path.abspath(os.path.join(self.basepath,'datafind.sub')))
        self.preengine_job = EngineJob(self.config, os.path.join(self.basepath,'prelalinference.sub'),self.logpath,engine='lalinferencedatadump',ispreengine=True,dax=self.is_dax())
        self.preengine_job.set_grid_site('local')
        self.preengine_job.set_universe('vanilla')
        if self.config.getboolean('analysis','roq'):
            self.computeroqweights_job = ROMJob(self.config,os.path.join(self.basepath,'computeroqweights.sub'),self.logpath,dax=self.is_dax())
            self.computeroqweights_job.set_grid_site('local')
        if self.config.has_option('condor','bayesline'):
            self.bayesline_job = BayesLineJob(self.config,os.path.join(self.basepath,'bayesline.sub'),self.logpath,dax=self.is_dax())
            self.bayesline_job.set_grid_site('local')
        self.bayeswavepsd_job={}
776
        self.bayeswavepost_job={}
777
        if self.config.has_option('condor','bayeswave'):
778 779 780
            for ifo in self.ifos:
                self.bayeswavepsd_job[ifo] = BayesWavePSDJob(self.config,os.path.join(self.basepath,'bayeswavepsd_%s.sub'%(ifo)),self.logpath,dax=self.is_dax())
                self.bayeswavepsd_job[ifo].set_grid_site('local')
781 782
                self.bayeswavepost_job[ifo] = BayesWavePostJob(self.config,os.path.join(self.basepath,'bayeswavepost_%s.sub'%(ifo)),self.logpath,dax=self.is_dax())
                self.bayeswavepost_job[ifo].set_grid_site('local')
783 784 785 786 787 788 789 790
        # Need to create a job file for each IFO combination
        self.engine_jobs={}
        ifocombos=[]
        for N in range(1,len(self.ifos)+1):
            for a in permutations(self.ifos,N):
                ifocombos.append(a)
        for ifos in ifocombos:
            self.engine_jobs[ifos] = EngineJob(self.config, os.path.join(self.basepath,'engine_%s.sub'%(reduce(lambda x,y:x+y, map(str,ifos)))),self.logpath,engine=self.engine,dax=self.is_dax(), site=site)
791 792 793 794 795 796 797 798
        if "summarypages" in self.config.get('condor','resultspage'):
            self.results_page_job = PESummaryResultsPageJob(self.config, os.path.join(self.basepath,'resultspage.sub'),self.logpath,dax=self.is_dax())
            self.results_page_job.set_grid_site('local')
        else:
            self.results_page_job = ResultsPageJob(self.config,os.path.join(self.basepath,'resultspage.sub'),self.logpath,dax=self.is_dax())
            self.results_page_job.set_grid_site('local')
            self.cotest_results_page_job = ResultsPageJob(self.config,os.path.join(self.basepath,'resultspagecoherent.sub'),self.logpath,dax=self.is_dax())
            self.cotest_results_page_job.set_grid_site('local')
799 800 801 802 803
        if self.engine=='lalinferencemcmc':
            self.combine_job = CombineMCMCJob(self.config,os.path.join(self.basepath,'combine_files.sub'),self.logpath,dax=self.is_dax())
            self.combine_job.set_grid_site('local')
            self.merge_job = MergeJob(self.config,os.path.join(self.basepath,'merge_runs.sub'),self.logpath,dax=self.is_dax(),engine='mcmc')
            self.merge_job.set_grid_site('local')
804
        else:
805 806 807 808 809 810 811 812
            self.merge_job = MergeJob(self.config,os.path.join(self.basepath,'merge_runs.sub'),self.logpath,dax=self.is_dax(),engine='nest')
            self.merge_job.set_grid_site('local')
        self.coherence_test_job = CoherenceTestJob(self.config,os.path.join(self.basepath,'coherence_test.sub'),self.logpath,dax=self.is_dax())
        self.coherence_test_job.set_grid_site('local')
        self.gracedbjob = GraceDBJob(self.config,os.path.join(self.basepath,'gracedb.sub'),self.logpath,dax=self.is_dax())
        self.gracedbjob.set_grid_site('local')
        self.mapjob = SkyMapJob(cp, os.path.join(self.basepath,'skymap.sub'), self.logpath)
        self.plotmapjob = PlotSkyMapJob(cp, os.path.join(self.basepath,'plotskymap.sub'),self.logpath)
Salvatore Vitale's avatar
Salvatore Vitale committed
813
        self.postruninfojob=PostRunInfoJob(self.config,os.path.join(self.basepath,'postrungdbinfo.sub'),self.logpath,dax=self.is_dax())
814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833
        # Process the input to build list of analyses to do
        self.events=self.setup_from_inputs()

        # Sanity checking
        if len(self.events)==0:
            print('No input events found, please check your config if you expect some events')
        self.times=[e.trig_time for e in self.events]

        # Set up the segments
        if not (self.config.has_option('input','gps-start-time') and self.config.has_option('input','gps-end-time')) and len(self.times)>0:
            (mintime,maxtime)=self.get_required_data(self.times)
            if not self.config.has_option('input','gps-start-time'):
                self.config.set('input','gps-start-time',str(int(floor(mintime))))
            if not self.config.has_option('input','gps-end-time'):
                self.config.set('input','gps-end-time',str(int(ceil(maxtime))))
        self.add_science_segments()

        # Save the final configuration that is being used
        # first to the run dir
        conffilename=os.path.join(self.basepath,'config.ini')
834
        with open(conffilename,'w') as conffile:
835 836 837
            self.config.write(conffile)
        if self.config.has_option('paths','webdir'):
            mkdirs(self.config.get('paths','webdir'))
838
            with open(os.path.join(self.config.get('paths','webdir'),'config.ini'),'w') as conffile:
839 840 841 842 843 844 845
                self.config.write(conffile)

        # Generate the DAG according to the config given
        for event in self.events: self.add_full_analysis(event)
        if self.config.has_option('analysis','upload-to-gracedb'):
            if self.config.getboolean('analysis','upload-to-gracedb'):
                self.add_gracedb_FITSskymap_upload(self.events[0],engine=self.engine)
Salvatore Vitale's avatar
Salvatore Vitale committed
846
            if self.config.has_option('condor','gdbinfo') and self.config.has_option('analysis','ugid') and self.config.getboolean('analysis','upload-to-gracedb'):
847
                if self.config.has_option('gracedbinfo','server'):
848
                    gdb_srv=self.config.get('gracedbinfo','server')
849
                else:
850
                    gdb_srv=None
851 852

                self.add_gracedb_info_node(None,event.GID,analysis='LIB',issky=True,server=gdb_srv)
Salvatore Vitale's avatar
Salvatore Vitale committed
853

854 855 856 857 858 859 860 861 862 863 864 865 866
        self.dagfilename="lalinference_%s-%s"%(self.config.get('input','gps-start-time'),self.config.get('input','gps-end-time'))
        self.set_dag_file(os.path.join(self.basepath,self.dagfilename))
        if self.is_dax():
            self.set_dax_file(self.dagfilename)

    def add_full_analysis(self,event):
        if self.engine=='lalinferencenest' or  self.engine=='lalinferenceburst':
            result=self.add_full_analysis_lalinferencenest(event)
        elif self.engine=='lalinferencemcmc':
            result=self.add_full_analysis_lalinferencemcmc(event)
        else:
            raise Exception('Unknown engine {0}'.format(self.engine))
        return result
867

868 869 870 871
    def create_frame_pfn_file(self):
        """
        Create a pegasus cache file name, uses inspiralutils
        """
872
        from lalapps import inspiralutils as iu
873 874 875 876
        gpsstart=self.config.get('input','gps-start-time')
        gpsend=self.config.get('input','gps-end-time')
        pfnfile=iu.create_frame_pfn_file(self.frtypes,gpsstart,gpsend)
        return pfnfile
877

878 879 880 881 882 883 884 885 886 887 888
    def get_required_data(self,times):
        """
        Calculate the data that will be needed to process all events
        """
        #psdlength = self.config.getint('input','max-psd-length')
        padding=self.config.getint('input','padding')
        if self.config.has_option('engine','seglen') or self.config.has_option('lalinference','seglen'):
            if self.config.has_option('engine','seglen'):
                seglen = int(np.ceil(self.config.getfloat('engine','seglen')))
            if self.config.has_option('lalinference','seglen'):
                seglen = self.config.getint('lalinference','seglen')
889 890 891 892 893
            try:
                use_gracedbpsd = (not self.config.getboolean('input','ignore-gracedb-psd'))
            except:
                use_gracedbpsd = True
            if (use_gracedbpsd and os.path.isfile(os.path.join(self.basepath,'psd.xml.gz'))) or self.config.has_option('condor','bayesline') or self.config.has_option('condor','bayeswave'):
894 895 896 897 898 899 900 901
                psdlength = 0
                padding = 0
                self.config.set('input','padding',str(padding))
                if self.config.has_option('condor','bayeswave'):
                    if (np.log2(seglen)%1):
                        seglen = np.power(2., np.ceil(np.log2(seglen)))
            else:
                psdlength = 32*seglen
902
        else:
903
            seglen = max(e.duration for e in self.events)
904 905 906 907 908
            try:
                use_gracedbpsd = (not self.config.getboolean('input','ignore-gracedb-psd'))
            except:
                use_gracedbpsd = True
            if (use_gracedbpsd and os.path.isfile(os.path.join(self.basepath,'psd.xml.gz'))) or self.config.has_option('condor','bayesline') or self.config.has_option('condor','bayeswave'):
909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933
                psdlength = 0
                padding = 0
                self.config.set('input','padding',str(padding))
                if self.config.has_option('condor','bayeswave'):
                    if (np.log2(seglen)%1):
                        seglen = np.power(2., np.ceil(np.log2(seglen)))
            else:
                psdlength = 32*seglen
        # Assume that the data interval is (end_time - seglen -padding , end_time + psdlength +padding )
        # -> change to (trig_time - seglen - padding - psdlength + 2 , trig_time + padding + 2) to estimate the psd before the trigger for online follow-up.
        # Also require padding before start time
        return (min(times)-padding-seglen-psdlength+2,max(times)+padding+2)

    def setup_from_times(self,times):
        """
        Generate a DAG from a list of times
        """
        for time in self.times:
            self.add_full_analysis(Event(trig_time=time))

    def select_events(self):
        """
        Read events from the config parser. Understands both ranges and comma separated events, or combinations
        eg. events=[0,1,5:10,21] adds to the analysis the events: 0,1,5,6,7,8,9,10 and 21
        """
934
        events=[]
935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950
        times=[]
        raw_events=self.config.get('input','events').replace('[','').replace(']','').split(',')
        for raw_event in raw_events:
            if ':' in raw_event:
                limits=raw_event.split(':')
                if len(limits) != 2:
                    print("Error: in event config option; ':' must separate two numbers.")
                    exit(0)
                low=int(limits[0])
                high=int(limits[1])
                if low>high:
                    events.extend(range(int(high),int(low)+1))
                elif high>low:
                    events.extend(range(int(low),int(high)+1))
            else:
                events.append(int(raw_event))
951
        return events
952

953 954 955 956 957 958 959 960 961 962 963 964 965 966
    def setup_from_inputs(self):
        """
        Scan the list of inputs, i.e.
        gps-time-file, injection-file, sngl-inspiral-file, coinc-inspiral-file, pipedown-database
        in the [input] section of the ini file.
        And process the events found therein
        """
        events=[]
        gpsstart=None
        gpsend=None
        if self.config.has_option('input','gps-start-time'):
            gpsstart=self.config.getfloat('input','gps-start-time')
        if self.config.has_option('input','gps-end-time'):
            gpsend=self.config.getfloat('input','gps-end-time')
967
        inputnames=['gps-time-file','burst-injection-file','injection-file','coinc-xml','pipedown-db','gid','gstlal-db']
968 969 970 971 972 973 974 975 976 977 978 979 980
        ReadInputFromList=sum([ 1 if self.config.has_option('input',name) else 0 for name in inputnames])
        # If no input events given, just return an empty list (e.g. for PP pipeline)
        if ReadInputFromList!=1 and (gpsstart is None or gpsend is None):
            return []
        # Review: Clean up this section
        if self.config.has_option('input','events'):
            selected_events=self.config.get('input','events')
            print('Selected events %s'%(str(selected_events)))

            if selected_events=='all':
                selected_events=None
            else:
                selected_events=self.select_events()
981
        else:
982 983 984 985 986 987
            selected_events=None

        if(self.config.has_option('engine','correlatedGaussianLikelihood') or
           self.config.has_option('engine','bimodalGaussianLikelihood') or
           self.config.has_option('engine','rosenbrockLikelihood')):
            analytic_test = True
988
        else:
989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055
            analytic_test = False

        # No input file given, analyse the entire time stretch between gpsstart and gpsend
        if self.config.has_option('input','analyse-all-time') and self.config.getboolean('input','analyse-all-time')==True:
            print('Setting up for analysis of continuous time stretch %f - %f'%(gpsstart,gpsend))
            if self.config.has_option('engine','seglen'):
                seglen=self.config.getfloat('engine','seglen')
            else:
                print('ERROR: seglen must be specified in [engine] section when running without input file')
                sys.exit(1)
            if(self.config.has_option('input','segment-overlap')):
                overlap=self.config.getfloat('input','segment-overlap')
            else:
                overlap=32.;
            if(overlap>seglen):
                print('ERROR: segment-overlap is greater than seglen')
                sys.exit(1)
            # Now divide gpsstart - gpsend into jobs of seglen - overlap length
            t=gpsstart
            events=[]
            while(t<gpsend):
                ev=Event(trig_time=t+seglen-2)
                ev.set_engine_option('segment-start',str(t-overlap))
                if not analytic_test:
                    ev.set_engine_option('time-min',str(t))
                tMax=t + seglen - overlap
                if tMax>=gpsend:
                    tMax=gpsend
                if not analytic_test:
                    ev.set_engine_option('time-max',str(tMax))
                events.append(ev)
                t=tMax
            return events

        # ASCII list of GPS times
        if self.config.has_option('input','gps-time-file'):
            times=scan_timefile(self.config.get('input','gps-time-file'))
            if self.config.has_option('input','timeslides-ascii'):
            # The timeslides-ascii files contains one row per trigtime, and a column per IFO
            # Note: the IFO order is the same given in the ifos section of the [analysis] tag
                print("Reading timeslides from ascii file. Columns order is understood as follow:")
                for this_ifo,ifo in enumerate(self.ifos):
                    print("Column %d"%this_ifo + "= %s "%(ifo))
                dest=self.config.get('input','timeslides-ascii')
                if not os.path.isfile(dest):
                    print("ERROR the ascii file %s containing the timeslides does not exist\n"%dest)
                    exit(1)
                else:
                    from numpy import loadtxt
                    data=loadtxt(dest).reshape(-1,len(self.ifos))
                    if len(self.ifos)!= len(data[0,:]):
                        print("ERROR: ascii timeslide file must contain a column for each IFO used in the analysis!\n")
                        exit(1)
                    if len(times)!=len(data[:,0]):
                        print('ERROR: ascii timeslide must contain a row for each trigtime. Exiting...\n')
                        exit(1)
                    timeslides={}
                    for this_time,time in enumerate(times):
                        timeslides[this_time]={}
                        for this_ifo,ifo in enumerate(self.ifos):
                            timeslides[this_time][ifo]=data[this_time,this_ifo]
                events=[Event(trig_time=time,timeslide_dict=timeslides[i_time]) for i_time,time in enumerate(times)]
            else:
                events=[Event(trig_time=time) for time in times]
        # Siminspiral Table
        if self.config.has_option('input','injection-file'):
            injTable = lsctables.SimInspiralTable.get_table(
1056
                              ligolw_utils.load_filename(self.config.get('input','injection-file'),
1057 1058 1059 1060 1061 1062
                                                  contenthandler=lsctables.use_in(ligolw.LIGOLWContentHandler)) )
            events=[Event(SimInspiral=inj) for inj in injTable]
            self.add_pfn_cache([create_pfn_tuple(self.config.get('input','injection-file'))])
        # SimBurst Table
        if self.config.has_option('input','burst-injection-file'):
            injfile=self.config.get('input','burst-injection-file')
1063
            injTable=lsctables.SimBurstTable.get_table(ligolw_utils.load_filename(injfile,contenthandler = lsctables.use_in(LIGOLWContentHandler)))
1064 1065 1066
            events=[Event(SimBurst=inj) for inj in injTable]
            self.add_pfn_cache([create_pfn_tuple(self.config.get('input','burst-injection-file'))])
        # LVAlert CoincInspiral Table
1067 1068
        gid = None
        if self.config.has_option('input','gid') or self.config.has_option('input', 'coinc-xml'):
1069 1070 1071 1072 1073 1074
            flow=20.0
            if self.config.has_option('lalinference','flow'):
                flow=min(ast.literal_eval(self.config.get('lalinference','flow')).values())
            threshold_snr = None
            if not self.config.has_option('engine','distance-max') and self.config.has_option('input','threshold-snr'):
                threshold_snr=self.config.getfloat('input','threshold-snr')
1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089

            # get coinc object and psd object
            from lal import series as lalseries
            psd_file_obj = None
            if self.config.has_option('input', 'gid'):
                from ligo.gracedb.rest import GraceDb, HTTPError
                gid = self.config.get('input', 'gid')
                if self.config.has_option('analysis','service-url'):
                    client = GraceDb(
                        service_url=self.config.get('analysis', 'service-url')
                    )
                else:
                    client = GraceDb()
                print("Download %s coinc.xml" % gid)
                coinc_file_obj = client.files(gid, "coinc.xml")
1090
                print("Download %s psd.xml.gz" % gid)
1091
                try:
1092 1093 1094
                    psd_file_obj = client.files(gid, "psd.xml.gz")
                except HTTPError:
                    print("Failed to download %s psd.xml.gz. lalinference will estimate the psd itself." % gid)
1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131
            else:
                coinc_file_obj = open(self.config.get('input', 'coinc-xml'), "rb")
                try:
                    psd_file_obj =  open(self.config.get('input', 'psd-xml-gz'), "rb")
                except:
                    print("lalinference will estimate the psd itself.")

            # write down the objects to files
            coinc_xml_obj = ligolw_utils.load_fileobj(
                coinc_file_obj,
                contenthandler = lsctables.use_in(ligolw.LIGOLWContentHandler)
            )[0]
            ligolw_utils.write_filename(
                coinc_xml_obj, os.path.join(self.basepath, "coinc.xml")
            )
            if psd_file_obj is not None:
                path_to_psd = os.path.join(self.basepath, "psd.xml.gz")
                psd_xml_obj = ligolw_utils.load_fileobj(
                    psd_file_obj,
                    contenthandler = lalseries.PSDContentHandler
                )[0]
                psd_dict = lalseries.read_psd_xmldoc(psd_xml_obj)
                ligolw_utils.write_filename(psd_xml_obj, path_to_psd, gz = True)
                ifos = sorted(
                    lsctables.CoincInspiralTable.get_table(
                        coinc_xml_obj
                    )[0].instruments
                )
                get_xml_psds(
                    os.path.realpath(path_to_psd), ifos,
                    os.path.realpath(os.path.join(self.basepath, "PSDs")),
                    end_time=None
                )
            else:
                psd_dict = None

            events = create_events_from_coinc_and_psd(
1132
                         coinc_xml_obj, psd_dict, gid, threshold_snr=threshold_snr, flow=flow,
1133 1134 1135
                         roq=self.config.getboolean('analysis','roq')
                     )

1136 1137
        # pipedown-database
        if self.config.has_option('input','gstlal-db'):
1138 1139
            queryfunc=get_zerolag_lloid
            dbname=self.config.get('input','gstlal-db')
1140
        elif self.config.has_option('input','pipedown-db'):
1141 1142
            queryfunc=get_zerolag_pipedown
            dbname=self.config.get('input','pipedown-db')
1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170
        else: dbname=None
        if dbname:
            db_connection = open_pipedown_database(dbname,None)[0]
            # Timeslides
            if self.config.has_option('input','time-slide-dump'):
                timeslidedump=self.config.get('input','time-slide-dump')
            else:
                timeslidedump=None
            if self.config.has_option('input','min-cfar'):
                mincfar=self.config.getfloat('input','min-cfar')
            else:
                mincfar=-1
            if self.config.has_option('input','max-cfar'):
                maxcfar=self.config.getfloat('input','max-cfar')
            else:
                maxcfar=-1
            if self.config.get('input','timeslides').lower()=='true':
                events=get_timeslides_pipedown(db_connection, gpsstart=gpsstart, gpsend=gpsend,dumpfile=timeslidedump,max_cfar=maxcfar)
            else:
                events=queryfunc(db_connection, gpsstart=gpsstart, gpsend=gpsend, dumpfile=timeslidedump,max_cfar=maxcfar,min_cfar=mincfar)
        if(selected_events is not None):
            used_events=[]
            for i in selected_events:
                e=events[i]
                e.event_id=i
                used_events.append(e)
            events=used_events
        if gpsstart is not None:
1171
            events = list(filter(lambda e: not e.trig_time<gpsstart, events))
1172
        if gpsend is not None:
1173
            events = list(filter(lambda e: not e.trig_time>gpsend, events))
1174
        return events
John Douglas Veitch's avatar
John Douglas Veitch committed
1175

1176 1177 1178 1179 1180 1181 1182 1183 1184 1185
    def add_full_analysis_lalinferencenest(self,event):
        """
        Generate an end-to-end analysis of a given event (Event class)
        For LALinferenceNest code. Uses parallel runs if specified
        """
        evstring=str(event.event_id)
        if event.trig_time is not None:
            evstring=str(event.trig_time)+'-'+str(event.event_id)
        if self.config.has_option('analysis','nparallel'):
            Npar=self.config.getint('analysis','nparallel')
1186
        else:
1187 1188 1189 1190
            Npar=4
        # Set up the parallel engine nodes
        enginenodes=[]
        bwpsdnodes={}
1191
        bwpostnodes={}
1192
        for i in range(Npar):
1193
            n,bwpsdnodes,bwpostnodes=self.add_engine_node(event,bwpsd=bwpsdnodes,bwpost=bwpostnodes)
1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212
            if n is not None:
                if i>0:
                    n.add_var_arg('--dont-dump-extras')
                enginenodes.append(n)
        if len(enginenodes)==0:
            return False
        myifos=enginenodes[0].get_ifos()
        # Merge the results together
        pagedir=os.path.join(self.webdir,evstring,myifos)
        #pagedir=os.path.join(self.basepath,evstring,myifos)
        mkdirs(pagedir)
        mergenode=MergeNode(self.merge_job,parents=enginenodes,engine='nest')
        mergenode.set_pos_output_file(os.path.join(self.posteriorpath,'posterior_%s_%s.hdf5'%(myifos,evstring)))
        self.add_node(mergenode)
        # Call finalize to build final list of available data
        enginenodes[0].finalize()
        enginenodes[0].set_psd_files()
        enginenodes[0].set_snr_file()
        if self.config.getboolean('analysis','coherence-test') and len(enginenodes[0].ifos)>1:
1213
            if self.site!='local':
1214
                zipfilename='postproc_'+evstring+'.tar.gz'
1215
            else:
1216
                zipfilename=None
1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256
            if "summarypages" not in self.config.get('condor','resultspage'):
                respagenode=self.add_results_page_node(resjob=self.cotest_results_page_job,outdir=pagedir,parent=mergenode,gzip_output=zipfilename,ifos=enginenodes[0].ifos)
                respagenode.set_psd_files(enginenodes[0].get_psd_files())
                respagenode.set_snr_file(enginenodes[0].get_snr_file())
                if os.path.exists(self.basepath+'/coinc.xml'):
                    try:
                        gid = self.config.get('input','gid')
                    except:
                        gid = None
                    respagenode.set_coinc_file(os.path.join(self.basepath, 'coinc.xml'), gid)

                par_mergenodes=[]
                for ifo in enginenodes[0].ifos:
                    co_merge_job = MergeJob(self.config,os.path.join(self.basepath,'merge_runs_%s.sub'%(ifo)),self.logpath,dax=self.is_dax(),engine='nest')
                    co_merge_job.set_grid_site('local')
                    cotest_nodes=[]
                    for i in range(Npar):
                        cot_node,bwpsdnodes,bwpostnodes=self.add_engine_node(event,bwpsd=bwpsdnodes,bwpost=bwpostnodes,ifos=[ifo],co_test=True)
                        if cot_node is not None:
                            if i>0:
                                cot_node.add_var_arg('--dont-dump-extras')
                            cotest_nodes.append(cot_node)
                    if len(cotest_nodes)==0:
                        return False
                    for co in cotest_nodes:
                        co.set_psdstart(enginenodes[0].GPSstart)
                        co.set_psdlength(enginenodes[0].psdlength)
                        if co!=cotest_nodes[0]:
                            co.add_var_arg('--dont-dump-extras')
                        else:
                            co.set_psd_files()
                            co.set_snr_file()
                    pmergenode=MergeNode(co_merge_job,parents=cotest_nodes,engine='nest')
                    pmergenode.set_pos_output_file(os.path.join(self.posteriorpath,'posterior_%s_%s.hdf5'%(ifo,evstring)))
                    self.add_node(pmergenode)
                    par_mergenodes.append(pmergenode)
                    presultsdir=os.path.join(pagedir,ifo)
                    mkdirs(presultsdir)
                    if self.site!='local':
                        pzipfilename='postproc_'+evstring+'_'+ifo+'.tar.gz'
1257
                    else:
1258
                        pzipfilename=None
1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273

                    mkdirs(os.path.join(self.basepath,'coherence_test'))
                    subresnode=self.add_results_page_node(outdir=presultsdir,parent=pmergenode, gzip_output=pzipfilename,ifos=ifo)
                    subresnode.set_psd_files(cotest_nodes[0].get_psd_files())
                    subresnode.set_snr_file(cotest_nodes[0].get_snr_file())
                    if os.path.exists(self.basepath+'/coinc.xml'):
                        try:
                            gid = self.config.get('input','gid')
                        except:
                            gid = None
                        subresnode.set_coinc_file(os.path.join(self.basepath, 'coinc.xml'), gid)
                    if self.config.has_option('input','injection-file') and event.event_id is not None:
                        subresnode.set_injection(self.config.get('input','injection-file'),event.event_id)
                    elif self.config.has_option('input','burst-injection-file') and event.event_id is not None:
                        subresnode.set_injection(self.config.get('input','burst-injection-file'),event.event_id)
1274 1275 1276 1277 1278 1279
                coherence_node=CoherenceTestNode(self.coherence_test_job,outfile=os.path.join(self.basepath,'coherence_test','coherence_test_%s_%s.dat'%(myifos,evstring)))
                coherence_node.add_coherent_parent(mergenode)
                for parmergenode in par_mergenodes:
                    coherence_node.add_incoherent_parent(parmergenode)
                self.add_node(coherence_node)
                respagenode.add_parent(coherence_node)
John Douglas Veitch's avatar
John Douglas Veitch committed
1280
                respagenode.set_bayes_coherent_incoherent(coherence_node.get_output_files()[0])
1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313

            elif "summarypages" in self.config.get('condor','resultspage'):
                par_mergenodes=[]
                for ifo in enginenodes[0].ifos:
                    co_merge_job = MergeJob(self.config,os.path.join(self.basepath,'merge_runs_%s.sub'%(ifo)),self.logpath,dax=self.is_dax(),engine='nest')
                    co_merge_job.set_grid_site('local')
                    cotest_nodes=[]
                    for i in range(Npar):
                        cot_node,bwpsdnodes,bwpostnodes=self.add_engine_node(event,bwpsd=bwpsdnodes,bwpost=bwpostnodes,ifos=[ifo],co_test=True)
                        if cot_node is not None:
                            if i>0:
                                cot_node.add_var_arg('--dont-dump-extras')
                            cotest_nodes.append(cot_node)
                    if len(cotest_nodes)==0:
                        return False
                    for co in cotest_nodes:
                        co.set_psdstart(enginenodes[0].GPSstart)
                        co.set_psdlength(enginenodes[0].psdlength)
                        if co!=cotest_nodes[0]:
                            co.add_var_arg('--dont-dump-extras')
                        else:
                            co.set_psd_files()
                            co.set_snr_file()
                    pmergenode=MergeNode(co_merge_job,parents=cotest_nodes,engine='nest')
                    pmergenode.set_pos_output_file(os.path.join(self.posteriorpath,'posterior_%s_%s.hdf5'%(ifo,evstring)))
                    self.add_node(pmergenode)
                    par_mergenodes.append(pmergenode)
                    presultsdir=os.path.join(pagedir,ifo)
                    mkdirs(presultsdir)
                    if self.site!='local':
                        pzipfilename='postproc_'+evstring+'_'+ifo+'.tar.gz'
                    else:
                        pzipfilename=None
1314

1315 1316
                respagenode=self.add_results_page_node_pesummary(outdir=pagedir,parent=mergenode,gzip_output=None,ifos=enginenodes[0].ifos,
                    evstring=evstring, coherence=True)
1317 1318 1319 1320 1321 1322 1323 1324
                respagenode.set_psd_files(enginenodes[0].ifos, enginenodes[0].get_psd_files())
                if os.path.exists(self.basepath+'/coinc.xml'):
                    try:
                        gid = self.config.get('input','gid')
                    except:
                        gid = None
                    respagenode.set_coinc_file(os.path.join(self.basepath, 'coinc.xml'), gid)

1325
        else:
1326 1327 1328 1329 1330
            if self.site!='local':
                zipfilename='postproc_'+evstring+'.tar.gz'
            else:
                zipfilename=None
            # Note: Disabled gzip_output for now. Possibly need it for future Pegasus use
1331 1332

            if "summarypages" in self.config.get('condor','resultspage'):
1333 1334
                respagenode=self.add_results_page_node_pesummary(outdir=pagedir,parent=mergenode,gzip_output=None,ifos=enginenodes[0].ifos,
                    evstring=evstring, coherence=False)
1335 1336 1337 1338 1339 1340 1341
                respagenode.set_psd_files(enginenodes[0].ifos, enginenodes[0].get_psd_files())
                if os.path.exists(self.basepath+'/coinc.xml'):
                    try:
                        gid = self.config.get('input','gid')
                    except:
                        gid = None
                    respagenode.set_coinc_file(os.path.join(self.basepath, 'coinc.xml'), gid)
1342 1343 1344 1345 1346 1347 1348 1349
            else:
                respagenode=self.add_results_page_node(outdir=pagedir,parent=mergenode,gzip_output=None,ifos=enginenodes[0].ifos)
                respagenode.set_psd_files(enginenodes[0].get_psd_files())
                respagenode.set_snr_file(enginenodes[0].get_snr_file())
                if os.path.exists(self.basepath+'/coinc.xml'):
                    try:
                        gid = self.config.get('input','gid')
                    except:
1350
                        gid = none
1351
                    respagenode.set_coinc_file(os.path.join(self.basepath, 'coinc.xml'), gid)
1352 1353 1354 1355
                if self.config.has_option('input','injection-file') and event.event_id is not None:
                    respagenode.set_injection(self.config.get('input','injection-file'),event.event_id)
                elif self.config.has_option('input','burst-injection-file') and event.event_id is not None:
                    respagenode.set_injection(self.config.get('input','burst-injection-file'),event.event_id)
1356 1357

        if self.config.has_option('analysis','upload-to-gracedb'):
1358
            if self.config.has_option('gracedbinfo','server'):
1359
                gdb_srv=self.config.get('gracedbinfo','server')
1360
            else:
1361
                gdb_srv=None
1362

1363
            if self.config.getboolean('analysis','upload-to-gracedb') and event.GID is not None:
1364 1365
                self.add_gracedb_start_node(event.GID,'LALInference',[sciseg.get_df_node() for sciseg in enginenodes[0].scisegs.values()],server=gdb_srv)
                self.add_gracedb_log_node(respagenode,event.GID,server=grb_srv)
1366 1367 1368 1369 1370 1371
            elif self.config.has_option('analysis','ugid'):
                # LIB will want to upload info to gracedb but if we pass the gid in the usual way the pipeline
                # will try to pull inspiral-only XML tables from the gdb page, failing.
                # To avoid that, LIB will read the gracedDB id to upload info to as an ugid=ID option
                # in the analysis section.
                ugid=self.config.get('analysis','ugid')
Salvatore Vitale's avatar
Salvatore Vitale committed
1372
                event.GID=ugid