core.py

""" The core tools used in pyfstat """
from __future__ import division, absolute_import, print_function

import os
import logging
import copy

import glob
import numpy as np
import scipy.special
import scipy.optimize

import lal
import lalpulsar
import pyfstat.helper_functions as helper_functions
import pyfstat.tcw_fstat_map_funcs as tcw

# workaround for matplotlib on X-less remote logins
if 'DISPLAY' in os.environ:
    import matplotlib.pyplot as plt
else:
    logging.info('No $DISPLAY environment variable found, so importing \
                  matplotlib.pyplot with non-interactive "Agg" backend.')
    import matplotlib
    matplotlib.use('Agg')
    import matplotlib.pyplot as plt

helper_functions.set_up_matplotlib_defaults()
args, tqdm = helper_functions.set_up_command_line_arguments()
detector_colors = {'h1': 'C0', 'l1': 'C1'}


class Bunch(object):
    """ Turns dictionary into object with attribute-style access

    Parameters
    ----------
    dict
        Input dictionary

    Examples
    --------
    >>> data = Bunch(dict(x=1, y=[1, 2, 3], z=True))
    >>> print(data.x)
    1
    >>> print(data.y)
    [1, 2, 3]
    >>> print(data.z)
    True

    """
    def __init__(self, dictionary):
        self.__dict__.update(dictionary)


def read_par(filename=None, label=None, outdir=None, suffix='par',
             return_type='dict', comments=['%', '#'], raise_error=False):
    """ Read in a .par or .loudest file, returns a dict or Bunch of the data

    Parameters
    ----------
    filename : str
        Filename (path) containing rows of `key=val` data to read in.
    label, outdir, suffix : str, optional
        If filename is None, form the file to read as `outdir/label.suffix`.
    return_type : {'dict', 'bunch'}, optional
        If `dict`, return a dictionary, if 'bunch' return a Bunch
    comments : str or list of strings, optional
        Characters denoting that a row is a comment.
    raise_error : bool, optional
        If True, raise an error for lines which are not comments, but cannot
        be read.

    Notes
    -----
    This can also be used to read in .loudest files, or any file which has
    rows of `key=val` data (in which the val can be understood using eval(val)

    Returns
    -------
    d: Bunch or dict
        The par values as either a `Bunch` or dict type

    """
    if filename is None:
        filename = '{}/{}.{}'.format(outdir, label, suffix)
    if os.path.isfile(filename) is False:
        raise ValueError("No file {} found".format(filename))
    d = {}
    with open(filename, 'r') as f:
        d = _get_dictionary_from_lines(f, comments, raise_error)
    if return_type in ['bunch', 'Bunch']:
        return Bunch(d)
    elif return_type in ['dict', 'dictionary']:
        return d
    else:
        raise ValueError('return_type {} not understood'.format(return_type))


def _get_dictionary_from_lines(lines, comments, raise_error):
    """ Return dictionary of key=val pairs for each line in lines

    Parameters
    ----------
    comments : str or list of strings
        Characters denoting that a row is a comment.
    raise_error : bool
        If True, raise an error for lines which are not comments, but cannot
        be read.

    Returns
    -------
    d: Bunch or dict
        The par values as either a `Bunch` or dict type

    """
    d = {}
    for line in lines:
        if line[0] not in comments and len(line.split('=')) == 2:
            try:
                key, val = line.rstrip('\n').split('=')
                key = key.strip()
                try:
                    d[key] = np.float64(eval(val.rstrip('; ')))
                except NameError:
                    d[key] = val.rstrip('; ')
            except SyntaxError:
                if raise_error:
                    raise IOError('Line {} not understood'.format(line))
                pass
    return d


def predict_fstat(h0, cosi, psi, Alpha, Delta, Freq, sftfilepattern,
                  minStartTime, maxStartTime, IFO=None, assumeSqrtSX=None,
                  **kwargs):
    """ Wrapper to lalapps_PredictFstat

    Parameters
    ----------
    h0, cosi, psi, Alpha, Delta, Freq : float
        Signal properties, see `lalapps_PredictFstat --help` for more info.
    sftfilepattern : str
        Pattern matching the sftfiles to use.
    minStartTime, maxStartTime : int
    IFO : str
        See `lalapps_PredictFstat --help`
    assumeSqrtSX : float or None
        See `lalapps_PredictFstat --help`, if None this option is not used

    Returns
    -------
    twoF_expected, twoF_sigma : float
        The expectation and standard deviation of 2F

    """
    tempory_filename = 'fs.tmp'

    cl_pfs = []
    cl_pfs.append("lalapps_PredictFstat")
    cl_pfs.append("--h0={}".format(h0))
    cl_pfs.append("--cosi={}".format(cosi))
    cl_pfs.append("--psi={}".format(psi))
    cl_pfs.append("--Alpha={}".format(Alpha))
    cl_pfs.append("--Delta={}".format(Delta))
    cl_pfs.append("--Freq={}".format(Freq))

    cl_pfs.append("--DataFiles='{}'".format(sftfilepattern))
    if assumeSqrtSX:
        cl_pfs.append("--assumeSqrtSX={}".format(assumeSqrtSX))
    if IFO:
        if ',' in IFO:
            logging.warning('Multiple detector selection not available, using'
                            ' all available data')
        else:
            cl_pfs.append("--IFO={}".format(IFO))

    cl_pfs.append("--minStartTime={}".format(int(minStartTime)))
    cl_pfs.append("--maxStartTime={}".format(int(maxStartTime)))
    cl_pfs.append("--outputFstat={}".format(tempory_filename))

    cl_pfs = " ".join(cl_pfs)
    helper_functions.run_commandline(cl_pfs)
    d = read_par(filename=tempory_filename)
    os.remove(tempory_filename)
    return float(d['twoF_expected']), float(d['twoF_sigma'])


class BaseSearchClass(object):
    """ The base search class providing parent methods to other searches """

    def _add_log_file(self):
        """ Log output to a file, requires class to have outdir and label """
        logfilename = '{}/{}.log'.format(self.outdir, self.label)
        fh = logging.FileHandler(logfilename)
        fh.setLevel(logging.INFO)
        fh.setFormatter(logging.Formatter(
            '%(asctime)s %(levelname)-8s: %(message)s',
            datefmt='%y-%m-%d %H:%M'))
        logging.getLogger().addHandler(fh)

    def _shift_matrix(self, n, dT):
        """ Generate the shift matrix

        Parameters
        ----------
        n : int
            The dimension of the shift-matrix to generate
        dT : float
            The time delta of the shift matrix

        Returns
        -------
        m : ndarray, shape (n,)
            The shift matrix.

        """
        m = np.zeros((n, n))
        factorial = np.math.factorial
        for i in range(n):
            for j in range(n):
                if i == j:
                    m[i, j] = 1.0
                elif i > j:
                    m[i, j] = 0.0
                else:
                    if i == 0:
                        m[i, j] = 2*np.pi*float(dT)**(j-i) / factorial(j-i)
                    else:
                        m[i, j] = float(dT)**(j-i) / factorial(j-i)
        return m

    def _shift_coefficients(self, theta, dT):
        """ Shift a set of coefficients by dT

        Parameters
        ----------
        theta : array-like, shape (n,)
            Vector of the expansion coefficients to transform starting from the
            lowest degree e.g [phi, F0, F1,...].
        dT : float
            Difference between the two reference times as tref_new - tref_old.

        Returns
        -------
        theta_new : ndarray, shape (n,)
            Vector of the coefficients as evaluated as the new reference time.
        """
        n = len(theta)
        m = self._shift_matrix(n, dT)
        return np.dot(m, theta)

    def _calculate_thetas(self, theta, delta_thetas, tbounds, theta0_idx=0):
        """ Calculates the set of thetas given delta_thetas, the jumps

        This is used when generating data containing glitches or timing noise.
        Specifically, the source parameters of the signal are not constant in
        time, but jump by `delta_theta` at `tbounds`.

        Parameters
        ----------
        theta : array_like
            The source parameters of size (n,).
        delta_thetas : array_like
            The jumps in the source parameters of size (m, n) where m is the
            number of jumps.
        tbounds : array_like
            Time boundaries of the jumps of size (m+2,).
        theta0_idx : int
            Index of the segment for which the theta are defined.

        Returns
        -------
        ndarray
            The set of thetas, shape (m+1, n).

        """
        thetas = [theta]
        for i, dt in enumerate(delta_thetas):
            if i < theta0_idx:
                pre_theta_at_ith_glitch = self._shift_coefficients(
                    thetas[0], tbounds[i+1] - self.tref)
                post_theta_at_ith_glitch = pre_theta_at_ith_glitch - dt
                thetas.insert(0, self._shift_coefficients(
                    post_theta_at_ith_glitch, self.tref - tbounds[i+1]))

            elif i >= theta0_idx:
                pre_theta_at_ith_glitch = self._shift_coefficients(
                    thetas[i], tbounds[i+1] - self.tref)
                post_theta_at_ith_glitch = pre_theta_at_ith_glitch + dt
                thetas.append(self._shift_coefficients(
                    post_theta_at_ith_glitch, self.tref - tbounds[i+1]))
        self.thetas_at_tref = thetas
        return thetas

    def _get_list_of_matching_sfts(self):
        """ Returns a list of sfts matching the attribute sftfilepattern """
        sftfilepatternlist = np.atleast_1d(self.sftfilepattern.split(';'))
        matches = [glob.glob(p) for p in sftfilepatternlist]
        matches = [item for sublist in matches for item in sublist]
        if len(matches) > 0:
            return matches
        else:
            raise IOError('No sfts found matching {}'.format(
                self.sftfilepattern))

    def set_ephemeris_files(self, earth_ephem=None, sun_ephem=None):
        """ Set the ephemeris files to use for the Earth and Sun

        Parameters
        ----------
        earth_ephem, sun_ephem: str
            Paths of the two files containing positions of Earth and Sun,
            respectively at evenly spaced times, as passed to CreateFstatInput

        Note: If not manually set, default values in ~/.pyfstat are used

        """

        earth_ephem_default, sun_ephem_default = (
                helper_functions.get_ephemeris_files())

        if earth_ephem is None:
            self.earth_ephem = earth_ephem_default
        if sun_ephem is None:
            self.sun_ephem = sun_ephem_default


class ComputeFstat(BaseSearchClass):
    """ Base class providing interface to `lalpulsar.ComputeFstat` """

    @helper_functions.initializer
    def __init__(self, tref, sftfilepattern=None, minStartTime=None,
                 maxStartTime=None, binary=False, BSGL=False,
                 transientWindowType=None, t0Band=None, tauBand=None,
                 tauMin=None,
                 dt0=None, dtau=None,
                 detectors=None, minCoverFreq=None, maxCoverFreq=None,
                 injectSources=None, injectSqrtSX=None, assumeSqrtSX=None,
                 SSBprec=None,
                 tCWFstatMapVersion='lal', cudaDeviceName=None):
        """
        Parameters
        ----------
        tref : int
            GPS seconds of the reference time.
        sftfilepattern : str
            Pattern to match SFTs using wildcards (*?) and ranges [0-9];
            mutiple patterns can be given separated by colons.
        minStartTime, maxStartTime : float GPStime
            Only use SFTs with timestemps starting from (including, excluding)
            this epoch
        binary : bool
            If true, search of binary parameters.
        BSGL : bool
            If true, compute the BSGL rather than the twoF value.
        transientWindowType: str
            If 'rect' or 'exp',
            allow for the Fstat to be computed over a transient range.
            ('none' instead of None explicitly calls the transient-window
            function, but with the full range, for debugging)
        t0Band, tauBand: int
            if >0, search t0 in (minStartTime,minStartTime+t0Band)
                   and tau in (tauMin,2*Tsft+tauBand).
            if =0, only compute CW Fstat with t0=minStartTime,
                   tau=maxStartTime-minStartTime.
        tauMin: int
            defaults to 2*Tsft
        dt0, dtau: int
            grid resolutions in transient start-time and duration,
            both default to Tsft
        detectors : str
            Two character reference to the data to use, specify None for no
            contraint. If multiple-separate by comma.
        minCoverFreq, maxCoverFreq : float
            The min and max cover frequency passed to CreateFstatInput, if
            either is None the range of frequencies in the SFT less 1Hz is
            used.
        injectSources : dict or str
            Either a dictionary of the values to inject, or a string pointing
            to the .cff file to inject
        injectSqrtSX :
            Not yet implemented
        assumeSqrtSX : float
            Don't estimate noise-floors but assume (stationary) per-IFO
            sqrt{SX} (if single value: use for all IFOs). If signal only,
            set sqrtSX=1
        SSBprec : int
            Flag to set the SSB calculation: 0=Newtonian, 1=relativistic,
            2=relativisitic optimised, 3=DMoff, 4=NO_SPIN
        tCWFstatMapVersion: str
            Choose between standard 'lal' implementation,
            'pycuda' for gpu, and some others for devel/debug.
        cudaDeviceName: str
            GPU name to be matched against drv.Device output.

        """

        self.set_ephemeris_files()
        self.init_computefstatistic_single_point()

    def _get_SFTCatalog(self):
        """ Load the SFTCatalog

        If sftfilepattern is specified, load the data. If not, attempt to
        create data on the fly.

        Returns
        -------
        SFTCatalog: lalpulsar.SFTCatalog

        """
        if hasattr(self, 'SFTCatalog'):
            return
        if self.sftfilepattern is None:
            for k in ['minStartTime', 'maxStartTime', 'detectors']:
                if getattr(self, k) is None:
                    raise ValueError('You must provide "{}" to injectSources'
                                     .format(k))
            C1 = getattr(self, 'injectSources', None) is None
            C2 = getattr(self, 'injectSqrtSX', None) is None
            if C1 and C2:
                raise ValueError('You must specify either one of injectSources'
                                 ' or injectSqrtSX')
            SFTCatalog = lalpulsar.SFTCatalog()
            Tsft = 1800
            Toverlap = 0
            Tspan = self.maxStartTime - self.minStartTime
            detNames = lal.CreateStringVector(
                *[d for d in self.detectors.split(',')])
            multiTimestamps = lalpulsar.MakeMultiTimestamps(
                self.minStartTime, Tspan, Tsft, Toverlap, detNames.length)
            SFTCatalog = lalpulsar.MultiAddToFakeSFTCatalog(
                SFTCatalog, detNames, multiTimestamps)
            return SFTCatalog

        logging.info('Initialising SFTCatalog')
        constraints = lalpulsar.SFTConstraints()
        if self.detectors:
            if ',' in self.detectors:
                logging.warning('Multiple detector selection not available,'
                                ' using all available data')
            else:
                constraints.detector = self.detectors
        if self.minStartTime:
            constraints.minStartTime = lal.LIGOTimeGPS(self.minStartTime)
        if self.maxStartTime:
            constraints.maxStartTime = lal.LIGOTimeGPS(self.maxStartTime)
        logging.info('Loading data matching pattern {}'.format(
                     self.sftfilepattern))
        SFTCatalog = lalpulsar.SFTdataFind(self.sftfilepattern, constraints)

        SFT_timestamps = [d.header.epoch for d in SFTCatalog.data]
        self.SFT_timestamps = [float(s) for s in SFT_timestamps]
        if len(SFT_timestamps) == 0:
            raise ValueError('Failed to load any data')
        if args.quite is False and args.no_interactive is False:
            try:
                from bashplotlib.histogram import plot_hist
                print('Data timestamps histogram:')
                plot_hist(SFT_timestamps, height=5, bincount=50)
            except ImportError:
                pass

        cl_tconv1 = 'lalapps_tconvert {}'.format(int(SFT_timestamps[0]))
        output = helper_functions.run_commandline(cl_tconv1,
                                                  log_level=logging.DEBUG)
        tconvert1 = output.rstrip('\n')
        cl_tconv2 = 'lalapps_tconvert {}'.format(int(SFT_timestamps[-1]))
        output = helper_functions.run_commandline(cl_tconv2,
                                                  log_level=logging.DEBUG)
        tconvert2 = output.rstrip('\n')
        logging.info('Data spans from {} ({}) to {} ({})'.format(
            int(SFT_timestamps[0]),
            tconvert1,
            int(SFT_timestamps[-1]),
            tconvert2))

        if self.minStartTime is None:
            self.minStartTime = int(SFT_timestamps[0])
        if self.maxStartTime is None:
            self.maxStartTime = int(SFT_timestamps[-1])

        detector_names = list(set([d.header.name for d in SFTCatalog.data]))
        self.detector_names = detector_names
        if len(detector_names) == 0:
            raise ValueError('No data loaded.')
        logging.info('Loaded {} data files from detectors {}'.format(
            len(SFT_timestamps), detector_names))

        return SFTCatalog

    def init_computefstatistic_single_point(self):
        """ Initilisation step of run_computefstatistic for a single point """

        SFTCatalog = self._get_SFTCatalog()

        logging.info('Initialising ephems')
        ephems = lalpulsar.InitBarycenter(self.earth_ephem, self.sun_ephem)

        logging.info('Initialising FstatInput')
        dFreq = 0
        if self.transientWindowType:
            self.whatToCompute = lalpulsar.FSTATQ_ATOMS_PER_DET
        else:
            self.whatToCompute = lalpulsar.FSTATQ_2F

        FstatOAs = lalpulsar.FstatOptionalArgs()
        FstatOAs.randSeed = lalpulsar.FstatOptionalArgsDefaults.randSeed
        if self.SSBprec:
            logging.info('Using SSBprec={}'.format(self.SSBprec))
            FstatOAs.SSBprec = self.SSBprec
        else:
            FstatOAs.SSBprec = lalpulsar.FstatOptionalArgsDefaults.SSBprec
        FstatOAs.Dterms = lalpulsar.FstatOptionalArgsDefaults.Dterms
        FstatOAs.runningMedianWindow = lalpulsar.FstatOptionalArgsDefaults.runningMedianWindow
        FstatOAs.FstatMethod = lalpulsar.FstatOptionalArgsDefaults.FstatMethod
        if self.assumeSqrtSX is None:
            FstatOAs.assumeSqrtSX = lalpulsar.FstatOptionalArgsDefaults.assumeSqrtSX
        else:
            mnf = lalpulsar.MultiNoiseFloor()
            assumeSqrtSX = np.atleast_1d(self.assumeSqrtSX)
            mnf.sqrtSn[:len(assumeSqrtSX)] = assumeSqrtSX
            mnf.length = len(assumeSqrtSX)
            FstatOAs.assumeSqrtSX = mnf
        FstatOAs.prevInput = lalpulsar.FstatOptionalArgsDefaults.prevInput
        FstatOAs.collectTiming = lalpulsar.FstatOptionalArgsDefaults.collectTiming

        if hasattr(self, 'injectSources') and type(self.injectSources) == dict:
            logging.info('Injecting source with params: {}'.format(
                self.injectSources))
            PPV = lalpulsar.CreatePulsarParamsVector(1)
            PP = PPV.data[0]
            PP.Amp.h0 = self.injectSources['h0']
            PP.Amp.cosi = self.injectSources['cosi']
            PP.Amp.phi0 = self.injectSources['phi0']
            PP.Amp.psi = self.injectSources['psi']
            PP.Doppler.Alpha = self.injectSources['Alpha']
            PP.Doppler.Delta = self.injectSources['Delta']
            if 'fkdot' in self.injectSources:
                PP.Doppler.fkdot = np.array(self.injectSources['fkdot'])
            else:
                PP.Doppler.fkdot = np.zeros(lalpulsar.PULSAR_MAX_SPINS)
                for i, key in enumerate(['F0', 'F1', 'F2']):
                    PP.Doppler.fkdot[i] = self.injectSources[key]
            PP.Doppler.refTime = self.tref
            if 't0' not in self.injectSources:
                PP.Transient.type = lalpulsar.TRANSIENT_NONE
            FstatOAs.injectSources = PPV
        elif hasattr(self, 'injectSources') and type(self.injectSources) == str:
            logging.info('Injecting source from param file: {}'.format(
                self.injectSources))
            PPV = lalpulsar.PulsarParamsFromFile(self.injectSources, self.tref)
            FstatOAs.injectSources = PPV
        else:
            FstatOAs.injectSources = lalpulsar.FstatOptionalArgsDefaults.injectSources
        if hasattr(self, 'injectSqrtSX') and self.injectSqrtSX is not None:
            raise ValueError('injectSqrtSX not implemented')
        else:
            FstatOAs.InjectSqrtSX = lalpulsar.FstatOptionalArgsDefaults.injectSqrtSX
        if self.minCoverFreq is None or self.maxCoverFreq is None:
            fAs = [d.header.f0 for d in SFTCatalog.data]
            fBs = [d.header.f0 + (d.numBins-1)*d.header.deltaF
                   for d in SFTCatalog.data]
            self.minCoverFreq = np.min(fAs) + 0.5
            self.maxCoverFreq = np.max(fBs) - 0.5
            logging.info('Min/max cover freqs not provided, using '
                         '{} and {}, est. from SFTs'.format(
                             self.minCoverFreq, self.maxCoverFreq))

        self.FstatInput = lalpulsar.CreateFstatInput(SFTCatalog,
                                                     self.minCoverFreq,
                                                     self.maxCoverFreq,
                                                     dFreq,
                                                     ephems,
                                                     FstatOAs
                                                     )

        logging.info('Initialising PulsarDoplerParams')
        PulsarDopplerParams = lalpulsar.PulsarDopplerParams()
        PulsarDopplerParams.refTime = self.tref
        PulsarDopplerParams.Alpha = 1
        PulsarDopplerParams.Delta = 1
        PulsarDopplerParams.fkdot = np.array([0, 0, 0, 0, 0, 0, 0])
        self.PulsarDopplerParams = PulsarDopplerParams

        logging.info('Initialising FstatResults')
        self.FstatResults = lalpulsar.FstatResults()

        if self.BSGL:
            if len(self.detector_names) < 2:
                raise ValueError("Can't use BSGL with single detectors data")
            else:
                logging.info('Initialising BSGL')

            # Tuning parameters - to be reviewed
            numDetectors = 2
            if hasattr(self, 'nsegs'):
                p_val_threshold = 1e-6
                Fstar0s = np.linspace(0, 1000, 10000)
                p_vals = scipy.special.gammaincc(2*self.nsegs, Fstar0s)
                Fstar0 = Fstar0s[np.argmin(np.abs(p_vals - p_val_threshold))]
                if Fstar0 == Fstar0s[-1]:
                    raise ValueError('Max Fstar0 exceeded')
            else:
                Fstar0 = 15.
            logging.info('Using Fstar0 of {:1.2f}'.format(Fstar0))
            oLGX = np.zeros(10)
            oLGX[:numDetectors] = 1./numDetectors
            self.BSGLSetup = lalpulsar.CreateBSGLSetup(numDetectors,
                                                       Fstar0,
                                                       oLGX,
                                                       True,
                                                       1)
            self.twoFX = np.zeros(10)
            self.whatToCompute = (self.whatToCompute +
                                  lalpulsar.FSTATQ_2F_PER_DET)

        if self.transientWindowType:
            logging.info('Initialising transient parameters')
            self.windowRange = lalpulsar.transientWindowRange_t()
            transientWindowTypes = {'none': lalpulsar.TRANSIENT_NONE,
                                    'rect': lalpulsar.TRANSIENT_RECTANGULAR,
                                    'exp':  lalpulsar.TRANSIENT_EXPONENTIAL}
            if self.transientWindowType in transientWindowTypes:
                self.windowRange.type = transientWindowTypes[self.transientWindowType]
            else:
                raise ValueError(
                    'Unknown window-type ({}) passed as input, [{}] allows.'
                    .format(self.transientWindowType,
                            ', '.join(transientWindowTypes)))

            # default spacing
            self.Tsft = int(1.0/SFTCatalog.data[0].header.deltaF)
            self.windowRange.dt0 = self.Tsft
            self.windowRange.dtau = self.Tsft

            # special treatment of window_type = none
            # ==> replace by rectangular window spanning all the data
            if self.windowRange.type == lalpulsar.TRANSIENT_NONE:
                self.windowRange.t0 = int(self.minStartTime)
                self.windowRange.t0Band = 0
                self.windowRange.tau = int(self.maxStartTime-self.minStartTime)
                self.windowRange.tauBand = 0
            else:  # user-set bands and spacings
                if self.t0Band is None:
                    self.windowRange.t0Band = 0
                else:
                    if not isinstance(self.t0Band, int):
                        logging.warn('Casting non-integer t0Band={} to int...'
                                     .format(self.t0Band))
                        self.t0Band = int(self.t0Band)
                    self.windowRange.t0Band = self.t0Band
                    if self.dt0:
                        self.windowRange.dt0 = self.dt0
                if self.tauBand is None:
                    self.windowRange.tauBand = 0
                else:
                    if not isinstance(self.tauBand, int):
                        logging.warn('Casting non-integer tauBand={} to int...'
                                     .format(self.tauBand))
                        self.tauBand = int(self.tauBand)
                    self.windowRange.tauBand = self.tauBand
                    if self.dtau:
                        self.windowRange.dtau = self.dtau
                    if self.tauMin is None:
                        self.windowRange.tau = int(2*self.Tsft)
                    else:
                        if not isinstance(self.tauMin, int):
                            logging.warn('Casting non-integer tauMin={} to int...'
                                         .format(self.tauMin))
                            self.tauMin = int(self.tauMin)
                        self.windowRange.tau = self.tauMin

            logging.info('Initialising transient FstatMap features...')
            self.tCWFstatMapFeatures, self.gpu_context = (
                tcw.init_transient_fstat_map_features(
                    self.tCWFstatMapVersion == 'pycuda', self.cudaDeviceName))

    def get_fullycoherent_twoF(self, tstart, tend, F0, F1, F2, Alpha, Delta,
                               asini=None, period=None, ecc=None, tp=None,
                               argp=None):
        """ Returns twoF or ln(BSGL) fully-coherently at a single point """
        self.PulsarDopplerParams.fkdot = np.array([F0, F1, F2, 0, 0, 0, 0])
        self.PulsarDopplerParams.Alpha = Alpha
        self.PulsarDopplerParams.Delta = Delta
        if self.binary:
            self.PulsarDopplerParams.asini = asini
            self.PulsarDopplerParams.period = period
            self.PulsarDopplerParams.ecc = ecc
            self.PulsarDopplerParams.tp = tp
            self.PulsarDopplerParams.argp = argp

        lalpulsar.ComputeFstat(self.FstatResults,
                               self.FstatInput,
                               self.PulsarDopplerParams,
                               1,
                               self.whatToCompute
                               )

        if not self.transientWindowType:
            if self.BSGL is False:
                return self.FstatResults.twoF[0]

            twoF = np.float(self.FstatResults.twoF[0])
            self.twoFX[0] = self.FstatResults.twoFPerDet(0)
            self.twoFX[1] = self.FstatResults.twoFPerDet(1)
            log10_BSGL = lalpulsar.ComputeBSGL(twoF, self.twoFX,
                                               self.BSGLSetup)
            return log10_BSGL/np.log10(np.exp(1))

        self.windowRange.t0 = int(tstart)  # TYPE UINT4
        if self.windowRange.tauBand == 0:
            # true single-template search also in transient params:
            # actual (t0,tau) window was set with tstart, tend before
            self.windowRange.tau = int(tend - tstart)  # TYPE UINT4

        self.FstatMap = tcw.call_compute_transient_fstat_map(
            self.tCWFstatMapVersion, self.tCWFstatMapFeatures,
            self.FstatResults.multiFatoms[0], self.windowRange)
        if self.tCWFstatMapVersion == 'lal':
            F_mn = self.FstatMap.F_mn.data
        else:
            F_mn = self.FstatMap.F_mn

        twoF = 2*np.max(F_mn)
        if self.BSGL is False:
            if np.isnan(twoF):
                return 0
            else:
                return twoF

        FstatResults_single = copy.copy(self.FstatResults)
        FstatResults_single.lenth = 1
        FstatResults_single.data = self.FstatResults.multiFatoms[0].data[0]
        FS0 = lalpulsar.ComputeTransientFstatMap(
            FstatResults_single.multiFatoms[0], self.windowRange, False)
        FstatResults_single.data = self.FstatResults.multiFatoms[0].data[1]
        FS1 = lalpulsar.ComputeTransientFstatMap(
            FstatResults_single.multiFatoms[0], self.windowRange, False)

        # for now, use the Doppler parameter with
        # multi-detector F maximised over t0,tau
        # to return BSGL
        # FIXME: should we instead compute BSGL over the whole F_mn
        # and return the maximum of that?
        idx_maxTwoF = np.argmax(F_mn)

        self.twoFX[0] = 2*FS0.F_mn.data[idx_maxTwoF]
        self.twoFX[1] = 2*FS1.F_mn.data[idx_maxTwoF]
        log10_BSGL = lalpulsar.ComputeBSGL(
                twoF, self.twoFX, self.BSGLSetup)

        return log10_BSGL/np.log10(np.exp(1))

    def calculate_twoF_cumulative(self, F0, F1, F2, Alpha, Delta, asini=None,
                                  period=None, ecc=None, tp=None, argp=None,
                                  tstart=None, tend=None, npoints=1000,
                                  ):
        """ Calculate the cumulative twoF along the obseration span

        Parameters
        ----------
        F0, F1, F2, Alpha, Delta: float
            Parameters at which to compute the cumulative twoF
        asini, period, ecc, tp, argp: float, optional
            Binary parameters at which to compute the cumulative 2F
        tstart, tend: int
            GPS times to restrict the range of data used - automatically
            truncated to the span of data available
        npoints: int
            Number of points to compute twoF along the span

        Notes
        -----
        The minimum cumulatibe twoF is hard-coded to be computed over
        the first 6 hours from either the first timestampe in the data (if
        tstart is smaller than it) or tstart.

        """
        SFTminStartTime = self.SFT_timestamps[0]
        SFTmaxStartTime = self.SFT_timestamps[-1]
        tstart = np.max([SFTminStartTime, tstart])
        min_tau = np.max([SFTminStartTime - tstart, 0]) + 3600*6
        max_tau = SFTmaxStartTime - tstart
        taus = np.linspace(min_tau, max_tau, npoints)
        twoFs = []
        if not self.transientWindowType:
            # still call the transient-Fstat-map function, but using the full range
            self.transientWindowType = 'none'
            self.init_computefstatistic_single_point()
        for tau in taus:
            detstat = self.get_fullycoherent_twoF(
                tstart=tstart, tend=tstart+tau, F0=F0, F1=F1, F2=F2,
                Alpha=Alpha, Delta=Delta, asini=asini, period=period, ecc=ecc,
                tp=tp, argp=argp)
            twoFs.append(detstat)

        return taus, np.array(twoFs)

    def _calculate_predict_fstat_cumulative(self, N, label=None, outdir=None,
                                            IFO=None, pfs_input=None):
        """ Calculates the predicted 2F and standard deviation cumulatively

        Parameters
        ----------
        N : int
            Number of timesteps to use between minStartTime and maxStartTime.
        label, outdir : str, optional
            The label and directory to read in the .loudest file from
        IFO : str
        pfs_input : dict, optional
            Input kwargs to predict_fstat (alternative to giving label and
            outdir).

        Returns
        -------
        times, pfs, pfs_sigma : ndarray, size (N,)

        """

        if pfs_input is None:
            if os.path.isfile('{}/{}.loudest'.format(outdir, label)) is False:
                raise ValueError(
                    'Need a loudest file to add the predicted Fstat')
            loudest = read_par(label=label, outdir=outdir, suffix='loudest')
            pfs_input = {key: loudest[key] for key in
                         ['h0', 'cosi', 'psi', 'Alpha', 'Delta', 'Freq']}
        times = np.linspace(self.minStartTime, self.maxStartTime, N+1)[1:]
        times = np.insert(times, 0, self.minStartTime + 86400/2.)
        out = [predict_fstat(minStartTime=self.minStartTime, maxStartTime=t,
                             sftfilepattern=self.sftfilepattern, IFO=IFO,
                             **pfs_input) for t in times]
        pfs, pfs_sigma = np.array(out).T
        return times, pfs, pfs_sigma

    def plot_twoF_cumulative(self, label, outdir, add_pfs=False, N=15,
                             injectSources=None, ax=None, c='k', savefig=True,
                             title=None, plt_label=None, **kwargs):
        """ Plot the twoF value cumulatively

        Parameters
        ----------
        label, outdir : str
        add_pfs : bool
            If true, plot the predicted 2F and standard deviation
        N : int
            Number of points to use
        injectSources : dict
            See `ComputeFstat`
        ax : matplotlib.axes._subplots_AxesSubplot, optional
            Axis to add the plot to.
        c : str
            Colour
        savefig : bool
            If true, save the figure in outdir
        title, plt_label: str
            Figure title and label

        Returns
        -------
        tauS, tauF : ndarray shape (N,)
            If savefig, the times and twoF (cumulative) values
        ax : matplotlib.axes._subplots_AxesSubplot, optional
            If savefig is False

        """
        if ax is None:
            fig, ax = plt.subplots()
        if injectSources:
            pfs_input = dict(
                h0=injectSources['h0'], cosi=injectSources['cosi'],
                psi=injectSources['psi'], Alpha=injectSources['Alpha'],
                Delta=injectSources['Delta'], Freq=injectSources['fkdot'][0])
        else:
            pfs_input = None

        taus, twoFs = self.calculate_twoF_cumulative(**kwargs)
        ax.plot(taus/86400., twoFs, label=plt_label, color=c)
        if len(self.detector_names) > 1:
            detector_names = self.detector_names
            detectors = self.detectors
            for d in self.detector_names:
                self.detectors = d
                self.init_computefstatistic_single_point()
                taus, twoFs = self.calculate_twoF_cumulative(**kwargs)
                ax.plot(taus/86400., twoFs, label='{}'.format(d),
                        color=detector_colors[d.lower()])
            self.detectors = detectors
            self.detector_names = detector_names

        if add_pfs:
            times, pfs, pfs_sigma = self._calculate_predict_fstat_cumulative(
                N=N, label=label, outdir=outdir, pfs_input=pfs_input)
            ax.fill_between(
                (times-self.minStartTime)/86400., pfs-pfs_sigma, pfs+pfs_sigma,
                color=c,
                label=(r'Predicted $\langle 2\mathcal{F} '
                       r'\rangle\pm $ 1-$\sigma$ band'),
                zorder=-10, alpha=0.2)
            if len(self.detector_names) > 1:
                for d in self.detector_names:
                    out = self._calculate_predict_fstat_cumulative(
                        N=N, label=label, outdir=outdir, IFO=d.upper(),
                        pfs_input=pfs_input)
                    times, pfs, pfs_sigma = out
                    ax.fill_between(
                        (times-self.minStartTime)/86400., pfs-pfs_sigma,
                        pfs+pfs_sigma, color=detector_colors[d.lower()],
                        alpha=0.5,
                        label=(
                            'Predicted $2\mathcal{{F}}$ 1-$\sigma$ band ({})'
                            .format(d.upper())),
                        zorder=-10)

        ax.set_xlabel(r'Days from $t_{{\rm start}}={:.0f}$'.format(
            kwargs['tstart']))
        if self.BSGL:
            ax.set_ylabel(r'$\log_{10}(\mathrm{BSGL})_{\rm cumulative}$')
        else:
            ax.set_ylabel(r'$\widetilde{2\mathcal{F}}_{\rm cumulative}$')
        ax.set_xlim(0, taus[-1]/86400)
        if plt_label:
            ax.legend(frameon=False, loc=2, fontsize=6)
        if title:
            ax.set_title(title)
        if savefig:
            plt.tight_layout()
            plt.savefig('{}/{}_twoFcumulative.png'.format(outdir, label))
            return taus, twoFs
        else:
            return ax

    def write_atoms_to_file(self, fnamebase=''):
        multiFatoms = getattr(self.FstatResults, 'multiFatoms', None)
        if multiFatoms and multiFatoms[0]:
            dopplerName = lalpulsar.PulsarDopplerParams2String ( self.PulsarDopplerParams )
            #fnameAtoms = os.path.join(self.outdir,'Fstatatoms_%s.dat' % dopplerName)
            fnameAtoms = fnamebase + '_Fstatatoms_%s.dat' % dopplerName
            fo = lal.FileOpen(fnameAtoms, 'w')
            lalpulsar.write_MultiFstatAtoms_to_fp ( fo, multiFatoms[0] )
            del fo # instead of lal.FileClose() which is not SWIG-exported
        else:
            raise RuntimeError('Cannot print atoms vector to file: no FstatResults.multiFatoms, or it is None!')


    def __del__(self):
        """
        In pyCuda case without autoinit,
        we need to make sure the context is removed at the end
        """
        if hasattr(self,'gpu_context') and self.gpu_context:
            self.gpu_context.detach()


class SemiCoherentSearch(ComputeFstat):
    """ A semi-coherent search """

    @helper_functions.initializer
    def __init__(self, label, outdir, tref, nsegs=None, sftfilepattern=None,
                 binary=False, BSGL=False, minStartTime=None,
                 maxStartTime=None, minCoverFreq=None, maxCoverFreq=None,
                 detectors=None, injectSources=None, assumeSqrtSX=None,
                 SSBprec=None):
        """
        Parameters
        ----------
        label, outdir: str
            A label and directory to read/write data from/to.
        tref, minStartTime, maxStartTime: int
            GPS seconds of the reference time, and start and end of the data.
        nsegs: int
            The (fixed) number of segments
        sftfilepattern: str
            Pattern to match SFTs using wildcards (*?) and ranges [0-9];
            mutiple patterns can be given separated by colons.

        For all other parameters, see pyfstat.ComputeFStat.
        """

        self.fs_file_name = "{}/{}_FS.dat".format(self.outdir, self.label)
        self.set_ephemeris_files()
        self.transientWindowType = 'rect'
        self.t0Band  = None
        self.tauBand = None
        self.tCWFstatMapVersion = 'lal'
        self.cudaDeviceName = None
        self.init_computefstatistic_single_point()
        self.init_semicoherent_parameters()

    def init_semicoherent_parameters(self):
        logging.info(('Initialising semicoherent parameters from {} to {} in'
                      ' {} segments').format(
            self.minStartTime, self.maxStartTime, self.nsegs))
        self.transientWindowType = 'rect'
        self.whatToCompute = lalpulsar.FSTATQ_2F+lalpulsar.FSTATQ_ATOMS_PER_DET
        self.tboundaries = np.linspace(self.minStartTime, self.maxStartTime,
                                       self.nsegs+1)
        self.Tcoh = self.tboundaries[1] - self.tboundaries[0]

        if hasattr(self, 'SFT_timestamps'):
            if self.tboundaries[0] < self.SFT_timestamps[0]:
                logging.debug(
                    'Semi-coherent start time {} before first SFT timestamp {}'
                    .format(self.tboundaries[0], self.SFT_timestamps[0]))
            if self.tboundaries[-1] > self.SFT_timestamps[-1]:
                logging.debug(
                    'Semi-coherent end time {} after last SFT timestamp {}'
                    .format(self.tboundaries[-1], self.SFT_timestamps[-1]))

    def get_semicoherent_twoF(
            self, F0, F1, F2, Alpha, Delta, asini=None,
            period=None, ecc=None, tp=None, argp=None,
            record_segments=False):
        """ Returns twoF or ln(BSGL) semi-coherently at a single point """

        self.PulsarDopplerParams.fkdot = np.array([F0, F1, F2, 0, 0, 0, 0])
        self.PulsarDopplerParams.Alpha = Alpha
        self.PulsarDopplerParams.Delta = Delta
        if self.binary:
            self.PulsarDopplerParams.asini = asini
            self.PulsarDopplerParams.period = period
            self.PulsarDopplerParams.ecc = ecc
            self.PulsarDopplerParams.tp = tp
            self.PulsarDopplerParams.argp = argp

        lalpulsar.ComputeFstat(self.FstatResults,
                               self.FstatInput,
                               self.PulsarDopplerParams,
                               1,
                               self.whatToCompute
                               )

        #if not self.transientWindowType:
        #    if self.BSGL is False:
        #        return self.FstatResults.twoF[0]
        #    twoF = np.float(self.FstatResults.twoF[0])
        #    self.twoFX[0] = self.FstatResults.twoFPerDet(0)
        #    self.twoFX[1] = self.FstatResults.twoFPerDet(1)
        #    log10_BSGL = lalpulsar.ComputeBSGL(twoF, self.twoFX,
        #                                       self.BSGLSetup)
        #    return log10_BSGL/np.log10(np.exp(1))

        detStat = 0
        if record_segments:
            self.detStat_per_segment = []

        self.windowRange.tau = int(self.Tcoh)  # TYPE UINT4
        for tstart in self.tboundaries[:-1]:
            d_detStat = self._get_per_segment_det_stat(tstart)
            detStat += d_detStat
            if record_segments:
                self.detStat_per_segment.append(d_detStat)

        return detStat

    def _get_per_segment_det_stat(self, tstart):
        self.windowRange.t0 = int(tstart)  # TYPE UINT4

        FS = lalpulsar.ComputeTransientFstatMap(
            self.FstatResults.multiFatoms[0], self.windowRange, False)

        if self.BSGL is False:
            d_detStat = 2*FS.F_mn.data[0][0]
        else:
            FstatResults_single = copy.copy(self.FstatResults)
            FstatResults_single.lenth = 1
            FstatResults_single.data = self.FstatResults.multiFatoms[0].data[0]
            FS0 = lalpulsar.ComputeTransientFstatMap(
                FstatResults_single.multiFatoms[0], self.windowRange, False)
            FstatResults_single.data = self.FstatResults.multiFatoms[0].data[1]
            FS1 = lalpulsar.ComputeTransientFstatMap(
                FstatResults_single.multiFatoms[0], self.windowRange, False)

            self.twoFX[0] = 2*FS0.F_mn.data[0][0]
            self.twoFX[1] = 2*FS1.F_mn.data[0][0]
            log10_BSGL = lalpulsar.ComputeBSGL(
                    2*FS.F_mn.data[0][0], self.twoFX, self.BSGLSetup)
            d_detStat = log10_BSGL/np.log10(np.exp(1))
        if np.isnan(d_detStat):
            logging.debug('NaNs in semi-coherent twoF treated as zero')
            d_detStat = 0

        return d_detStat


class SemiCoherentGlitchSearch(ComputeFstat):
    """ A semi-coherent glitch search

    This implements a basic `semi-coherent glitch F-stat in which the data
    is divided into segments either side of the proposed glitches and the
    fully-coherent F-stat in each segment is summed to give the semi-coherent
    F-stat
    """

    @helper_functions.initializer
    def __init__(self, label, outdir, tref, minStartTime, maxStartTime,
                 nglitch=1, sftfilepattern=None, theta0_idx=0, BSGL=False,
                 minCoverFreq=None, maxCoverFreq=None, assumeSqrtSX=None,
                 detectors=None, SSBprec=None, injectSources=None):
        """
        Parameters
        ----------
        label, outdir: str
            A label and directory to read/write data from/to.
        tref, minStartTime, maxStartTime: int
            GPS seconds of the reference time, and start and end of the data.
        nglitch: int
            The (fixed) number of glitches; this can zero, but occasionally
            this causes issue (in which case just use ComputeFstat).
        sftfilepattern: str
            Pattern to match SFTs using wildcards (*?) and ranges [0-9];
            mutiple patterns can be given separated by colons.
        theta0_idx, int
            Index (zero-based) of which segment the theta refers to - uyseful
            if providing a tight prior on theta to allow the signal to jump
            too theta (and not just from)

        For all other parameters, see pyfstat.ComputeFStat.
        """

        self.fs_file_name = "{}/{}_FS.dat".format(self.outdir, self.label)
        self.set_ephemeris_files()
        self.transientWindowType = 'rect'
        self.t0Band  = None
        self.tauBand = None
        self.tCWFstatMapVersion = 'lal'
        self.cudaDeviceName = None
        self.binary  = False
        self.init_computefstatistic_single_point()

    def get_semicoherent_nglitch_twoF(self, F0, F1, F2, Alpha, Delta, *args):
        """ Returns the semi-coherent glitch summed twoF """

        args = list(args)
        tboundaries = ([self.minStartTime] + args[-self.nglitch:]
                       + [self.maxStartTime])
        delta_F0s = args[-3*self.nglitch:-2*self.nglitch]
        delta_F1s = args[-2*self.nglitch:-self.nglitch]
        delta_F2 = np.zeros(len(delta_F0s))
        delta_phi = np.zeros(len(delta_F0s))
        theta = [0, F0, F1, F2]
        delta_thetas = np.atleast_2d(
                np.array([delta_phi, delta_F0s, delta_F1s, delta_F2]).T)

        thetas = self._calculate_thetas(theta, delta_thetas, tboundaries,
                                        theta0_idx=self.theta0_idx)

        twoFSum = 0
        for i, theta_i_at_tref in enumerate(thetas):
            ts, te = tboundaries[i], tboundaries[i+1]
            if te - ts > 1800:
                twoFVal = self.get_fullycoherent_twoF(
                    ts, te, theta_i_at_tref[1], theta_i_at_tref[2],
                    theta_i_at_tref[3], Alpha, Delta)
                twoFSum += twoFVal

        if np.isfinite(twoFSum):
            return twoFSum
        else:
            return -np.inf

    def compute_glitch_fstat_single(self, F0, F1, F2, Alpha, Delta, delta_F0,
                                    delta_F1, tglitch):
        """ Returns the semi-coherent glitch summed twoF for nglitch=1

        Note: OBSOLETE, used only for testing
        """

        theta = [F0, F1, F2]
        delta_theta = [delta_F0, delta_F1, 0]
        tref = self.tref

        theta_at_glitch = self._shift_coefficients(theta, tglitch - tref)
        theta_post_glitch_at_glitch = theta_at_glitch + delta_theta
        theta_post_glitch = self._shift_coefficients(
            theta_post_glitch_at_glitch, tref - tglitch)

        twoFsegA = self.get_fullycoherent_twoF(
            self.minStartTime, tglitch, theta[0], theta[1], theta[2], Alpha,
            Delta)

        if tglitch == self.maxStartTime:
            return twoFsegA

        twoFsegB = self.get_fullycoherent_twoF(
            tglitch, self.maxStartTime, theta_post_glitch[0],
            theta_post_glitch[1], theta_post_glitch[2], Alpha,
            Delta)

        return twoFsegA + twoFsegB