mcmc_based_searches.py 95.1 KB
Newer Older
Gregory Ashton's avatar
Gregory Ashton committed
1
""" Searches using MCMC-based methods """
2
from __future__ import division, absolute_import, print_function
Gregory Ashton's avatar
Gregory Ashton committed
3

4
import sys
Gregory Ashton's avatar
Gregory Ashton committed
5
import os
6
import copy
Gregory Ashton's avatar
Gregory Ashton committed
7
import logging
8
from collections import OrderedDict
9
import subprocess
10 11 12 13

import numpy as np
import matplotlib
import matplotlib.pyplot as plt
Gregory Ashton's avatar
Gregory Ashton committed
14
from ptemcee import Sampler as PTSampler
15 16 17
import corner
import dill as pickle

18
import pyfstat.core as core
19 20
from pyfstat.core import tqdm, args, read_par
import pyfstat.optimal_setup_functions as optimal_setup_functions
21
import pyfstat.helper_functions as helper_functions
22 23


24
class MCMCSearch(core.BaseSearchClass):
Gregory Ashton's avatar
Gregory Ashton committed
25
    """MCMC search using ComputeFstat
26 27 28 29 30 31 32 33 34

    Parameters
    ----------
    theta_prior: dict
        Dictionary of priors and fixed values for the search parameters.
        For each parameters (key of the dict), if it is to be held fixed
        the value should be the constant float, if it is be searched, the
        value should be a dictionary of the prior.
    tref, minStartTime, maxStartTime: int
35 36 37 38 39 40 41
        GPS seconds of the reference time, start time and end time. While tref
        is requirede, minStartTime and maxStartTime default to None in which
        case all available data is used.
    label, outdir: str
        A label and output directory (optional, defaults is `'data'`) to
        name files
    sftfilepattern: str, optional
Gregory Ashton's avatar
Gregory Ashton committed
42 43
        Pattern to match SFTs using wildcards (*?) and ranges [0-9];
        mutiple patterns can be given separated by colons.
44
    detectors: str, optional
Gregory Ashton's avatar
Gregory Ashton committed
45 46
        Two character reference to the detectors to use, specify None for no
        contraint and comma separate for multiple references.
47
    nsteps: list (2,), optional
48 49 50
        Number of burn-in and production steps to take, [nburn, nprod]. See
        `pyfstat.MCMCSearch.setup_initialisation()` for details on adding
        initialisation steps.
51
    nwalkers, ntemps: int, optional
52 53
        The number of walkers and temperates to use in the parallel
        tempered PTSampler.
54
    log10beta_min float < 0, optional
55 56
        The  log_10(beta) value, if given the set of betas passed to PTSampler
        are generated from `np.logspace(0, log10beta_min, ntemps)` (given
Gregory Ashton's avatar
Gregory Ashton committed
57
        in descending order to ptemcee).
58
    theta_initial: dict, array, optional
59 60
        A dictionary of distribution about which to distribute the
        initial walkers about
61
    rhohatmax: float, optional
62 63 64
        Upper bound for the SNR scale parameter (required to normalise the
        Bayes factor) - this needs to be carefully set when using the
        evidence.
65
    binary: bool, optional
66
        If true, search over binary parameters
67
    BSGL: bool, optional
Gregory Ashton's avatar
Gregory Ashton committed
68
        If true, use the BSGL statistic
69
    SSBPrec: int, optional
Gregory Ashton's avatar
Gregory Ashton committed
70
        SSBPrec (SSB precision) to use when calling ComputeFstat
71
    minCoverFreq, maxCoverFreq: float, optional
72 73
        Minimum and maximum instantaneous frequency which will be covered
        over the SFT time span as passed to CreateFstatInput
74
    injectSources: dict, optional
Gregory Ashton's avatar
Gregory Ashton committed
75 76
        If given, inject these properties into the SFT files before running
        the search
77
    assumeSqrtSX: float, optional
Gregory Ashton's avatar
Gregory Ashton committed
78
        Don't estimate noise-floors, but assume (stationary) per-IFO sqrt{SX}
David Keitel's avatar
David Keitel committed
79 80 81 82 83 84
    transientWindowType: str
        If 'rect' or 'exp',
        compute atoms so that a transient (t0,tau) map can later be computed.
        ('none' instead of None explicitly calls the transient-window function,
        but with the full range, for debugging)
        Currently only supported for nsegs=1.
85 86 87
    tCWFstatMapVersion: str
        Choose between standard 'lal' implementation,
        'pycuda' for gpu, and some others for devel/debug.
88 89 90 91 92 93 94 95 96 97 98 99 100 101 102

    Attributes
    ----------
    symbol_dictionary: dict
        Key, val pairs of the parameters (i.e. `F0`, `F1`), to Latex math
        symbols for plots
    unit_dictionary: dict
        Key, val pairs of the parameters (i.e. `F0`, `F1`), and the
        units (i.e. `Hz`)
    transform_dictionary: dict
        Key, val pairs of the parameters (i.e. `F0`, `F1`), where the key is
        itself a dictionary which can item `multiplier`, `subtractor`, or
        `unit` by which to transform by and update the units.

    """
103 104

    symbol_dictionary = dict(
105
        F0='$f$', F1='$\dot{f}$', F2='$\ddot{f}$', Alpha=r'$\alpha$',
106 107
        Delta='$\delta$', asini='asini', period='P', ecc='ecc', tp='tp',
        argp='argp')
108
    unit_dictionary = dict(
109 110
        F0='Hz', F1='Hz/s', F2='Hz/s$^2$', Alpha=r'rad', Delta='rad',
        asini='', period='s', ecc='', tp='', argp='')
111
    transform_dictionary = {}
112

Gregory Ashton's avatar
Gregory Ashton committed
113
    @helper_functions.initializer
114 115 116
    def __init__(self, theta_prior, tref, label, outdir='data',
                 minStartTime=None, maxStartTime=None, sftfilepattern=None,
                 detectors=None, nsteps=[100, 100], nwalkers=100, ntemps=1,
117
                 log10beta_min=-5, theta_initial=None,
118
                 rhohatmax=1000, binary=False, BSGL=False,
Gregory Ashton's avatar
Gregory Ashton committed
119
                 SSBprec=None, minCoverFreq=None, maxCoverFreq=None,
David Keitel's avatar
David Keitel committed
120
                 injectSources=None, assumeSqrtSX=None,
121
                 transientWindowType=None, tCWFstatMapVersion='lal'):
122

Gregory Ashton's avatar
Gregory Ashton committed
123 124
        if os.path.isdir(outdir) is False:
            os.mkdir(outdir)
125
        self._add_log_file()
126
        logging.info('Set-up MCMC search for model {}'.format(self.label))
127 128
        if sftfilepattern:
            logging.info('Using data {}'.format(self.sftfilepattern))
129
        else:
130
            logging.info('No sftfilepattern given')
131 132
        if injectSources:
            logging.info('Inject sources: {}'.format(injectSources))
133
        self.pickle_path = '{}/{}_saved_data.p'.format(self.outdir, self.label)
134
        self._unpack_input_theta()
135
        self.ndim = len(self.theta_keys)
136 137
        if self.log10beta_min:
            self.betas = np.logspace(0, self.log10beta_min, self.ntemps)
138 139
        else:
            self.betas = None
140

141 142 143
        if args.clean and os.path.isfile(self.pickle_path):
            os.rename(self.pickle_path, self.pickle_path+".old")

144
        self._set_likelihoodcoef()
145
        self._log_input()
146 147 148

    def _set_likelihoodcoef(self):
        self.likelihoodcoef = np.log(70./self.rhohatmax**4)
149

150
    def _log_input(self):
151
        logging.info('theta_prior = {}'.format(self.theta_prior))
152
        logging.info('nwalkers={}'.format(self.nwalkers))
153 154
        logging.info('nsteps = {}'.format(self.nsteps))
        logging.info('ntemps = {}'.format(self.ntemps))
155 156
        logging.info('log10beta_min = {}'.format(
            self.log10beta_min))
157

158
    def _initiate_search_object(self):
159
        logging.info('Setting up search object')
160
        self.search = core.ComputeFstat(
161
            tref=self.tref, sftfilepattern=self.sftfilepattern,
162
            minCoverFreq=self.minCoverFreq, maxCoverFreq=self.maxCoverFreq,
David Keitel's avatar
David Keitel committed
163 164
            detectors=self.detectors, BSGL=self.BSGL,
            transientWindowType=self.transientWindowType,
165
            minStartTime=self.minStartTime, maxStartTime=self.maxStartTime,
166
            binary=self.binary, injectSources=self.injectSources,
167 168
            assumeSqrtSX=self.assumeSqrtSX, SSBprec=self.SSBprec,
            tCWFstatMapVersion=self.tCWFstatMapVersion)
169 170 171 172
        if self.minStartTime is None:
            self.minStartTime = self.search.minStartTime
        if self.maxStartTime is None:
            self.maxStartTime = self.search.maxStartTime
173 174

    def logp(self, theta_vals, theta_prior, theta_keys, search):
175
        H = [self._generic_lnprior(**theta_prior[key])(p) for p, key in
176 177 178 179 180 181
             zip(theta_vals, theta_keys)]
        return np.sum(H)

    def logl(self, theta, search):
        for j, theta_i in enumerate(self.theta_idxs):
            self.fixed_theta[theta_i] = theta[j]
182 183 184
        twoF = search.get_fullycoherent_twoF(
            self.minStartTime, self.maxStartTime, *self.fixed_theta)
        return twoF/2.0 + self.likelihoodcoef
185

186
    def _unpack_input_theta(self):
187
        full_theta_keys = ['F0', 'F1', 'F2', 'Alpha', 'Delta']
188 189 190
        if self.binary:
            full_theta_keys += [
                'asini', 'period', 'ecc', 'tp', 'argp']
191 192
        full_theta_keys_copy = copy.copy(full_theta_keys)

193 194
        full_theta_symbols = ['$f$', '$\dot{f}$', '$\ddot{f}$', r'$\alpha$',
                              r'$\delta$']
195 196
        if self.binary:
            full_theta_symbols += [
197
                'asini', 'period', 'ecc', 'tp', 'argp']
198

199 200
        self.theta_keys = []
        fixed_theta_dict = {}
201
        for key, val in self.theta_prior.iteritems():
202 203
            if type(val) is dict:
                fixed_theta_dict[key] = 0
Gregory Ashton's avatar
Gregory Ashton committed
204
                self.theta_keys.append(key)
205 206 207 208 209 210
            elif type(val) in [float, int, np.float64]:
                fixed_theta_dict[key] = val
            else:
                raise ValueError(
                    'Type {} of {} in theta not recognised'.format(
                        type(val), key))
Gregory Ashton's avatar
Gregory Ashton committed
211
            full_theta_keys_copy.pop(full_theta_keys_copy.index(key))
212 213 214 215 216 217 218 219 220 221 222 223 224 225 226

        if len(full_theta_keys_copy) > 0:
            raise ValueError(('Input dictionary `theta` is missing the'
                              'following keys: {}').format(
                                  full_theta_keys_copy))

        self.fixed_theta = [fixed_theta_dict[key] for key in full_theta_keys]
        self.theta_idxs = [full_theta_keys.index(k) for k in self.theta_keys]
        self.theta_symbols = [full_theta_symbols[i] for i in self.theta_idxs]

        idxs = np.argsort(self.theta_idxs)
        self.theta_idxs = [self.theta_idxs[i] for i in idxs]
        self.theta_symbols = [self.theta_symbols[i] for i in idxs]
        self.theta_keys = [self.theta_keys[i] for i in idxs]

227 228 229 230 231 232 233 234 235
    def _evaluate_logpost(self, p0vec):
        init_logp = np.array([
            self.logp(p, self.theta_prior, self.theta_keys, self.search)
            for p in p0vec])
        init_logl = np.array([
            self.logl(p, self.search)
            for p in p0vec])
        return init_logl + init_logp

236
    def _check_initial_points(self, p0):
237 238
        for nt in range(self.ntemps):
            logging.info('Checking temperature {} chains'.format(nt))
239 240
            num = sum(self._evaluate_logpost(p0[nt]) == -np.inf)
            if num > 0:
241 242
                logging.warning(
                    'Of {} initial values, {} are -np.inf due to the prior'
243
                    .format(len(p0[0]), num))
244
                p0 = self._generate_new_p0_to_fix_initial_points(
245
                    p0, nt)
246

247
    def _generate_new_p0_to_fix_initial_points(self, p0, nt):
248
        logging.info('Attempting to correct intial values')
249 250
        init_logpost = self._evaluate_logpost(p0[nt])
        idxs = np.arange(self.nwalkers)[init_logpost == -np.inf]
251
        count = 0
252
        while sum(init_logpost == -np.inf) > 0 and count < 100:
253 254 255
            for j in idxs:
                p0[nt][j] = (p0[nt][np.random.randint(0, self.nwalkers)]*(
                             1+np.random.normal(0, 1e-10, self.ndim)))
256
            init_logpost = self._evaluate_logpost(p0[nt])
257 258
            count += 1

259
        if sum(init_logpost == -np.inf) > 0:
260 261 262 263 264
            logging.info('Failed to fix initial priors')
        else:
            logging.info('Suceeded to fix initial priors')

        return p0
265

266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289
    def setup_initialisation(self, nburn0, scatter_val=1e-10):
        """ Add an initialisation step to the MCMC run

        If called prior to `run()`, adds an intial step in which the MCMC
        simulation is run for `nburn0` steps. After this, the MCMC simulation
        continues in the usual manner (i.e. for nburn and nprod steps), but the
        walkers are reset scattered around the maximum likelihood position
        of the initialisation step.

        Parameters
        ----------
        nburn0: int
            Number of initialisation steps to take
        scatter_val: float
            Relative number to scatter walkers around the maximum likelihood
            position after the initialisation step

        """

        logging.info('Setting up initialisation with nburn0={}, scatter_val={}'
                     .format(nburn0, scatter_val))
        self.nsteps = [nburn0] + self.nsteps
        self.scatter_val = scatter_val

290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402
#    def setup_burnin_convergence_testing(
#            self, n=10, test_type='autocorr', windowed=False, **kwargs):
#        """ Set up convergence testing during the MCMC simulation
#
#        Parameters
#        ----------
#        n: int
#            Number of steps after which to test convergence
#        test_type: str ['autocorr', 'GR']
#            If 'autocorr' use the exponential autocorrelation time (kwargs
#            passed to `get_autocorr_convergence`). If 'GR' use the Gelman-Rubin
#            statistic (kwargs passed to `get_GR_convergence`)
#        windowed: bool
#            If True, only calculate the convergence test in a window of length
#            `n`
#        **kwargs:
#            Passed to either `_test_autocorr_convergence()` or
#            `_test_GR_convergence()` depending on `test_type`.
#
#        """
#        logging.info('Setting up convergence testing')
#        self.convergence_n = n
#        self.convergence_windowed = windowed
#        self.convergence_test_type = test_type
#        self.convergence_kwargs = kwargs
#        self.convergence_diagnostic = []
#        self.convergence_diagnosticx = []
#        if test_type in ['autocorr']:
#            self._get_convergence_test = self._test_autocorr_convergence
#        elif test_type in ['GR']:
#            self._get_convergence_test = self._test_GR_convergence
#        else:
#            raise ValueError('test_type {} not understood'.format(test_type))
#
#
#    def _test_autocorr_convergence(self, i, sampler, test=True, n_cut=5):
#        try:
#            acors = np.zeros((self.ntemps, self.ndim))
#            for temp in range(self.ntemps):
#                if self.convergence_windowed:
#                    j = i-self.convergence_n
#                else:
#                    j = 0
#                x = np.mean(sampler.chain[temp, :, j:i, :], axis=0)
#                acors[temp, :] = emcee.autocorr.exponential_time(x)
#            c = np.max(acors, axis=0)
#        except emcee.autocorr.AutocorrError:
#            logging.info('Failed to calculate exponential autocorrelation')
#            c = np.zeros(self.ndim) + np.nan
#        except AttributeError:
#            logging.info('Unable to calculate exponential autocorrelation')
#            c = np.zeros(self.ndim) + np.nan
#
#        self.convergence_diagnosticx.append(i - self.convergence_n/2.)
#        self.convergence_diagnostic.append(list(c))
#
#        if test:
#            return i > n_cut * np.max(c)
#
#    def _test_GR_convergence(self, i, sampler, test=True, R=1.1):
#        if self.convergence_windowed:
#            s = sampler.chain[0, :, i-self.convergence_n+1:i+1, :]
#        else:
#            s = sampler.chain[0, :, :i+1, :]
#        N = float(self.convergence_n)
#        M = float(self.nwalkers)
#        W = np.mean(np.var(s, axis=1), axis=0)
#        per_walker_mean = np.mean(s, axis=1)
#        mean = np.mean(per_walker_mean, axis=0)
#        B = N / (M-1.) * np.sum((per_walker_mean-mean)**2, axis=0)
#        Vhat = (N-1)/N * W + (M+1)/(M*N) * B
#        c = np.sqrt(Vhat/W)
#        self.convergence_diagnostic.append(c)
#        self.convergence_diagnosticx.append(i - self.convergence_n/2.)
#
#        if test and np.max(c) < R:
#            return True
#        else:
#            return False
#
#    def _test_convergence(self, i, sampler, **kwargs):
#        if np.mod(i+1, self.convergence_n) == 0:
#            return self._get_convergence_test(i, sampler, **kwargs)
#        else:
#            return False
#
#    def _run_sampler_with_conv_test(self, sampler, p0, nprod=0, nburn=0):
#        logging.info('Running {} burn-in steps with convergence testing'
#                     .format(nburn))
#        iterator = tqdm(sampler.sample(p0, iterations=nburn), total=nburn)
#        for i, output in enumerate(iterator):
#            if self._test_convergence(i, sampler, test=True,
#                                      **self.convergence_kwargs):
#                logging.info(
#                    'Converged at {} before max number {} of steps reached'
#                    .format(i, nburn))
#                self.convergence_idx = i
#                break
#        iterator.close()
#        logging.info('Running {} production steps'.format(nprod))
#        j = nburn
#        iterator = tqdm(sampler.sample(output[0], iterations=nprod),
#                        total=nprod)
#        for result in iterator:
#            self._test_convergence(j, sampler, test=False,
#                                   **self.convergence_kwargs)
#            j += 1
#        return sampler

    def _run_sampler(self, sampler, p0, nprod=0, nburn=0, window=50):
        for result in tqdm(sampler.sample(p0, iterations=nburn+nprod),
                           total=nburn+nprod):
            pass
403

404 405
        self.mean_acceptance_fraction = np.mean(
            sampler.acceptance_fraction, axis=1)
406
        logging.info("Mean acceptance fraction: {}"
407
                     .format(self.mean_acceptance_fraction))
408
        if self.ntemps > 1:
409
            self.tswap_acceptance_fraction = sampler.tswap_acceptance_fraction
410 411
            logging.info("Tswap acceptance fraction: {}"
                         .format(sampler.tswap_acceptance_fraction))
Gregory Ashton's avatar
Gregory Ashton committed
412 413 414
        self.autocorr_time = sampler.get_autocorr_time(window=window)
        logging.info("Autocorrelation length: {}".format(
            self.autocorr_time))
415 416 417

        return sampler

418
    def _estimate_run_time(self):
419 420 421 422 423 424 425 426 427 428
        """ Print the estimated run time

        Uses timing coefficients based on a Lenovo T460p Intel(R)
        Core(TM) i5-6300HQ CPU @ 2.30GHz.

        """
        # Todo: add option to time on a machine, and move coefficients to
        # ~/.pyfstat.conf
        if (type(self.theta_prior['Alpha']) == dict or
                type(self.theta_prior['Delta']) == dict):
Gregory Ashton's avatar
Gregory Ashton committed
429 430 431 432
            tau0LD = 5.2e-7
            tau0T = 1.5e-8
            tau0S = 1.2e-4
            tau0C = 5.8e-6
433
        else:
Gregory Ashton's avatar
Gregory Ashton committed
434
            tau0LD = 1.3e-7
435
            tau0T = 1.5e-8
Gregory Ashton's avatar
Gregory Ashton committed
436 437
            tau0S = 9.1e-5
            tau0C = 5.5e-6
438
        Nsfts = (self.maxStartTime - self.minStartTime) / 1800.
439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455
        if hasattr(self, 'run_setup'):
            ts = []
            for row in self.run_setup:
                nsteps = row[0]
                nsegs = row[1]
                numb_evals = np.sum(nsteps)*self.nwalkers*self.ntemps
                t = (tau0S + tau0LD*Nsfts) * numb_evals
                if nsegs > 1:
                    t += (tau0C + tau0T*Nsfts)*nsegs*numb_evals
                ts.append(t)
            time = np.sum(ts)
        else:
            numb_evals = np.sum(self.nsteps)*self.nwalkers*self.ntemps
            time = (tau0S + tau0LD*Nsfts) * numb_evals
            if getattr(self, 'nsegs', 1) > 1:
                time += (tau0C + tau0T*Nsfts)*self.nsegs*numb_evals

456
        logging.info('Estimated run-time = {} s = {:1.0f}:{:1.0f} m'.format(
457
            time, *divmod(time, 60)))
458

Gregory Ashton's avatar
Gregory Ashton committed
459 460
    def run(self, proposal_scale_factor=2, create_plots=True, window=50,
            **kwargs):
461 462 463 464 465 466 467 468 469 470 471
        """ Run the MCMC simulatation

        Parameters
        ----------
        proposal_scale_factor: float
            The proposal scale factor used by the sampler, see Goodman & Weare
            (2010). If the acceptance fraction is too low, you can raise it by
            decreasing the a parameter; and if it is too high, you can reduce
            it by increasing the a parameter [Foreman-Mackay (2013)].
        create_plots: bool
            If true, save trace plots of the walkers
Gregory Ashton's avatar
Gregory Ashton committed
472
        window: int
473 474
            The minimum number of autocorrelation times needed to trust the
            result when estimating the autocorrelation time (see
Gregory Ashton's avatar
Gregory Ashton committed
475
            ptemcee.Sampler.get_autocorr_time for further details.
476 477 478
        **kwargs:
            Passed to _plot_walkers to control the figures

479 480
        Returns
        -------
Gregory Ashton's avatar
Gregory Ashton committed
481 482
        sampler: ptemcee.Sampler
            The ptemcee ptsampler object
483

484
        """
485

486
        self.old_data_is_okay_to_use = self._check_old_data_is_okay_to_use()
487 488 489
        if self.old_data_is_okay_to_use is True:
            logging.warning('Using saved data from {}'.format(
                self.pickle_path))
490
            d = self.get_saved_data_dictionary()
491 492 493
            self.samples = d['samples']
            self.lnprobs = d['lnprobs']
            self.lnlikes = d['lnlikes']
494
            self.all_lnlikelihood = d['all_lnlikelihood']
495
            self.chain = d['chain']
496 497
            return

498
        self._initiate_search_object()
499
        self._estimate_run_time()
500

Gregory Ashton's avatar
Gregory Ashton committed
501 502 503
        sampler = PTSampler(
            ntemps=self.ntemps, nwalkers=self.nwalkers, dim=self.ndim,
            logl=self.logl, logp=self.logp,
504
            logpargs=(self.theta_prior, self.theta_keys, self.search),
505
            loglargs=(self.search,), betas=self.betas, a=proposal_scale_factor)
506

507 508 509
        p0 = self._generate_initial_p0()
        p0 = self._apply_corrections_to_p0(p0)
        self._check_initial_points(p0)
510

511
        # Run initialisation steps if required
512 513 514
        ninit_steps = len(self.nsteps) - 2
        for j, n in enumerate(self.nsteps[:-2]):
            logging.info('Running {}/{} initialisation with {} steps'.format(
Gregory Ashton's avatar
Gregory Ashton committed
515
                j, ninit_steps, n))
Gregory Ashton's avatar
Gregory Ashton committed
516
            sampler = self._run_sampler(sampler, p0, nburn=n, window=window)
517
            if create_plots:
518
                fig, axes = self._plot_walkers(sampler,
519
                                               **kwargs)
520 521
                fig.tight_layout()
                fig.savefig('{}/{}_init_{}_walkers.png'.format(
Gregory Ashton's avatar
Gregory Ashton committed
522
                    self.outdir, self.label, j))
523

524 525 526
            p0 = self._get_new_p0(sampler)
            p0 = self._apply_corrections_to_p0(p0)
            self._check_initial_points(p0)
527 528
            sampler.reset()

Gregory Ashton's avatar
Gregory Ashton committed
529 530 531 532
        if len(self.nsteps) > 1:
            nburn = self.nsteps[-2]
        else:
            nburn = 0
533 534 535
        nprod = self.nsteps[-1]
        logging.info('Running final burn and prod with {} steps'.format(
            nburn+nprod))
536
        sampler = self._run_sampler(sampler, p0, nburn=nburn, nprod=nprod)
537

538
        if create_plots:
539 540 541 542 543 544 545
            try:
                fig, axes = self._plot_walkers(sampler, nprod=nprod, **kwargs)
                fig.tight_layout()
                fig.savefig('{}/{}_walkers.png'.format(self.outdir, self.label))
            except RuntimeError as e:
                logging.warning("Failed to save walker plots due to Erro {}"
                                .format(e))
546 547

        samples = sampler.chain[0, :, nburn:, :].reshape((-1, self.ndim))
Gregory Ashton's avatar
Gregory Ashton committed
548 549 550
        lnprobs = sampler.logprobability[0, :, nburn:].reshape((-1))
        lnlikes = sampler.loglikelihood[0, :, nburn:].reshape((-1))
        all_lnlikelihood = sampler.loglikelihood[:, :, nburn:]
551
        self.samples = samples
552
        self.chain = sampler.chain
553 554
        self.lnprobs = lnprobs
        self.lnlikes = lnlikes
555
        self.all_lnlikelihood = all_lnlikelihood
556 557
        self._save_data(sampler, samples, lnprobs, lnlikes, all_lnlikelihood,
                        sampler.chain)
Gregory Ashton's avatar
Gregory Ashton committed
558
        return sampler
559

560
    def _get_rescale_multiplier_for_key(self, key):
561
        """ Get the rescale multiplier from the transform_dictionary
562 563 564 565 566

        Can either be a float, a string (in which case it is interpretted as
        a attribute of the MCMCSearch class, e.g. minStartTime, or non-existent
        in which case 0 is returned
        """
567
        if key not in self.transform_dictionary:
568 569
            return 1

570 571
        if 'multiplier' in self.transform_dictionary[key]:
            val = self.transform_dictionary[key]['multiplier']
572 573 574
            if type(val) == str:
                if hasattr(self, val):
                    multiplier = getattr(
575
                        self, self.transform_dictionary[key]['multiplier'])
576 577 578 579 580 581 582 583 584
                else:
                    raise ValueError(
                        "multiplier {} not a class attribute".format(val))
            else:
                multiplier = val
        else:
            multiplier = 1
        return multiplier

585
    def _get_rescale_subtractor_for_key(self, key):
586
        """ Get the rescale subtractor from the transform_dictionary
587 588 589 590 591

        Can either be a float, a string (in which case it is interpretted as
        a attribute of the MCMCSearch class, e.g. minStartTime, or non-existent
        in which case 0 is returned
        """
592
        if key not in self.transform_dictionary:
593 594
            return 0

595 596
        if 'subtractor' in self.transform_dictionary[key]:
            val = self.transform_dictionary[key]['subtractor']
597 598 599
            if type(val) == str:
                if hasattr(self, val):
                    subtractor = getattr(
600
                        self, self.transform_dictionary[key]['subtractor'])
601 602 603 604 605 606 607 608 609
                else:
                    raise ValueError(
                        "subtractor {} not a class attribute".format(val))
            else:
                subtractor = val
        else:
            subtractor = 0
        return subtractor

610
    def _scale_samples(self, samples, theta_keys):
611
        """ Scale the samples using the transform_dictionary """
612
        for key in theta_keys:
613
            if key in self.transform_dictionary:
614 615
                idx = theta_keys.index(key)
                s = samples[:, idx]
616
                subtractor = self._get_rescale_subtractor_for_key(key)
617
                s = s - subtractor
618
                multiplier = self._get_rescale_multiplier_for_key(key)
619
                s *= multiplier
620 621
                samples[:, idx] = s

622 623
        return samples

624
    def _get_labels(self, newline_units=False):
625
        """ Combine the units, symbols and rescaling to give labels """
626

627 628 629 630 631 632
        labels = []
        for key in self.theta_keys:
            label = None
            s = self.symbol_dictionary[key]
            s.replace('_{glitch}', r'_\textrm{glitch}')
            u = self.unit_dictionary[key]
633 634 635 636 637 638 639
            if key in self.transform_dictionary:
                if 'symbol' in self.transform_dictionary[key]:
                    s = self.transform_dictionary[key]['symbol']
                if 'label' in self.transform_dictionary[key]:
                    label = self.transform_dictionary[key]['label']
                if 'unit' in self.transform_dictionary[key]:
                    u = self.transform_dictionary[key]['unit']
640
            if label is None:
641 642 643 644
                if newline_units:
                    label = '{} \n [{}]'.format(s, u)
                else:
                    label = '{} [{}]'.format(s, u)
645 646
            labels.append(label)
        return labels
647

648 649
    def plot_corner(self, figsize=(7, 7), add_prior=False, nstds=None,
                    label_offset=0.4, dpi=300, rc_context={},
650
                    tglitch_ratio=False, fig_and_axes=None, save_fig=True,
651
                    **kwargs):
652 653 654 655 656 657 658 659 660
        """ Generate a corner plot of the posterior

        Using the `corner` package (https://pypi.python.org/pypi/corner/),
        generate estimates of the posterior from the production samples.

        Parameters
        ----------
        figsize: tuple (7, 7)
            Figure size in inches (passed to plt.subplots)
661 662 663
        add_prior: bool, str
            If true, plot the prior as a red line. If 'full' then for uniform
            priors plot the full extent of the prior.
664 665 666 667 668 669 670 671 672 673 674 675 676 677
        nstds: float
            The number of standard deviations to plot centered on the mean
        label_offset: float
            Offset the labels from the plot: useful to precent overlapping the
            tick labels with the axis labels
        dpi: int
            Passed to plt.savefig
        rc_context: dict
            Dictionary of rc values to set while generating the figure (see
            matplotlib rc for more details)
        tglitch_ratio: bool
            If true, and tglitch is a parameter, plot posteriors as the
            fractional time at which the glitch occurs instead of the actual
            time
678 679 680 681 682
        fig_and_axes: tuple
            fig and axes to plot on, the axes must be of the right shape,
            namely (ndim, ndim)
        save_fig: bool
            If true, save the figure, else return the fig, axes
683 684
        **kwargs:
            Passed to corner.corner
685

686 687 688 689
        Returns
        -------
        fig, axes:
            The matplotlib figure and axes, only returned if save_fig = False
690 691

        """
692

693 694 695 696
        if 'truths' in kwargs and len(kwargs['truths']) != self.ndim:
            logging.warning('len(Truths) != ndim, Truths will be ignored')
            kwargs['truths'] = None

Gregory Ashton's avatar
Gregory Ashton committed
697 698
        if self.ndim < 2:
            with plt.rc_context(rc_context):
699 700 701 702
                if fig_and_axes is None:
                    fig, ax = plt.subplots(figsize=figsize)
                else:
                    fig, ax = fig_and_axes
Gregory Ashton's avatar
Gregory Ashton committed
703 704 705 706 707 708 709
                ax.hist(self.samples, bins=50, histtype='stepfilled')
                ax.set_xlabel(self.theta_symbols[0])

            fig.savefig('{}/{}_corner.png'.format(
                self.outdir, self.label), dpi=dpi)
            return

710
        with plt.rc_context(rc_context):
711 712 713 714 715
            if fig_and_axes is None:
                fig, axes = plt.subplots(self.ndim, self.ndim,
                                         figsize=figsize)
            else:
                fig, axes = fig_and_axes
716 717

            samples_plt = copy.copy(self.samples)
718
            labels = self._get_labels(newline_units=True)
719

720
            samples_plt = self._scale_samples(samples_plt, self.theta_keys)
721 722 723 724 725

            if tglitch_ratio:
                for j, k in enumerate(self.theta_keys):
                    if k == 'tglitch':
                        s = samples_plt[:, j]
726 727 728
                        samples_plt[:, j] = (
                            s - self.minStartTime)/(
                                self.maxStartTime - self.minStartTime)
729
                        labels[j] = r'$R_{\textrm{glitch}}$'
730 731 732 733 734 735 736

            if type(nstds) is int and 'range' not in kwargs:
                _range = []
                for j, s in enumerate(samples_plt.T):
                    median = np.median(s)
                    std = np.std(s)
                    _range.append((median - nstds*std, median + nstds*std))
737 738
            elif 'range' in kwargs:
                _range = kwargs.pop('range')
739 740 741
            else:
                _range = None

742 743 744 745
            hist_kwargs = kwargs.pop('hist_kwargs', dict())
            if 'normed' not in hist_kwargs:
                hist_kwargs['normed'] = True

746
            fig_triangle = corner.corner(samples_plt,
747
                                         labels=labels,
748 749 750 751 752
                                         fig=fig,
                                         bins=50,
                                         max_n_ticks=4,
                                         plot_contours=True,
                                         plot_datapoints=True,
753
                                         #label_kwargs={'fontsize': 12},
754 755 756
                                         data_kwargs={'alpha': 0.1,
                                                      'ms': 0.5},
                                         range=_range,
757
                                         hist_kwargs=hist_kwargs,
758 759 760 761 762 763 764 765 766 767 768 769
                                         **kwargs)

            axes_list = fig_triangle.get_axes()
            axes = np.array(axes_list).reshape(self.ndim, self.ndim)
            plt.draw()
            for ax in axes[:, 0]:
                ax.yaxis.set_label_coords(-label_offset, 0.5)
            for ax in axes[-1, :]:
                ax.xaxis.set_label_coords(0.5, -label_offset)
            for ax in axes_list:
                ax.set_rasterized(True)
                ax.set_rasterization_zorder(-10)
770 771

                for tick in ax.xaxis.get_major_ticks():
772
                    #tick.label.set_fontsize(8)
773 774
                    tick.label.set_rotation('horizontal')
                for tick in ax.yaxis.get_major_ticks():
775
                    #tick.label.set_fontsize(8)
776 777
                    tick.label.set_rotation('vertical')

778 779 780 781
            plt.tight_layout(h_pad=0.0, w_pad=0.0)
            fig.subplots_adjust(hspace=0.05, wspace=0.05)

            if add_prior:
782
                self._add_prior_to_corner(axes, self.samples, add_prior)
783

784 785 786 787 788
            if save_fig:
                fig_triangle.savefig('{}/{}_corner.png'.format(
                    self.outdir, self.label), dpi=dpi)
            else:
                return fig, axes
789

790
    def _add_prior_to_corner(self, axes, samples, add_prior):
791 792 793
        for i, key in enumerate(self.theta_keys):
            ax = axes[i][i]
            s = samples[:, i]
794 795 796 797 798 799 800 801 802 803
            lnprior = self._generic_lnprior(**self.theta_prior[key])
            if add_prior == 'full' and self.theta_prior[key]['type'] == 'unif':
                lower = self.theta_prior[key]['lower']
                upper = self.theta_prior[key]['upper']
                r = upper-lower
                xlim = [lower-0.05*r, upper+0.05*r]
                x = np.linspace(xlim[0], xlim[1], 1000)
            else:
                xlim = ax.get_xlim()
                x = np.linspace(s.min(), s.max(), 1000)
804 805
            multiplier = self._get_rescale_multiplier_for_key(key)
            subtractor = self._get_rescale_subtractor_for_key(key)
806 807 808 809 810 811 812 813
            ax.plot((x-subtractor)*multiplier,
                    [np.exp(lnprior(xi)) for xi in x], '-C3',
                    label='prior')

            for j in range(i, self.ndim):
                axes[j][i].set_xlim(xlim[0], xlim[1])
            for k in range(0, i):
                axes[i][k].set_ylim(xlim[0], xlim[1])
814

815 816 817 818 819 820 821 822
    def plot_prior_posterior(self, normal_stds=2):
        """ Plot the posterior in the context of the prior """
        fig, axes = plt.subplots(nrows=self.ndim, figsize=(8, 4*self.ndim))
        N = 1000
        from scipy.stats import gaussian_kde

        for i, (ax, key) in enumerate(zip(axes, self.theta_keys)):
            prior_dict = self.theta_prior[key]
823
            prior_func = self._generic_lnprior(**prior_dict)
824 825 826 827 828
            if prior_dict['type'] == 'unif':
                x = np.linspace(prior_dict['lower'], prior_dict['upper'], N)
                prior = prior_func(x)
                prior[0] = 0
                prior[-1] = 0
Gregory Ashton's avatar
Gregory Ashton committed
829 830 831 832 833
            elif prior_dict['type'] == 'log10unif':
                upper = prior_dict['log10upper']
                lower = prior_dict['log10lower']
                x = np.linspace(lower, upper, N)
                prior = [prior_func(xi) for xi in x]
834 835 836 837 838
            elif prior_dict['type'] == 'norm':
                lower = prior_dict['loc'] - normal_stds * prior_dict['scale']
                upper = prior_dict['loc'] + normal_stds * prior_dict['scale']
                x = np.linspace(lower, upper, N)
                prior = prior_func(x)
839 840 841 842 843
            elif prior_dict['type'] == 'halfnorm':
                lower = prior_dict['loc']
                upper = prior_dict['loc'] + normal_stds * prior_dict['scale']
                x = np.linspace(lower, upper, N)
                prior = [prior_func(xi) for xi in x]
Gregory Ashton's avatar
Gregory Ashton committed
844 845 846 847 848
            elif prior_dict['type'] == 'neghalfnorm':
                upper = prior_dict['loc']
                lower = prior_dict['loc'] - normal_stds * prior_dict['scale']
                x = np.linspace(lower, upper, N)
                prior = [prior_func(xi) for xi in x]
849 850 851
            else:
                raise ValueError('Not implemented for prior type {}'.format(
                    prior_dict['type']))
852
            priorln = ax.plot(x, prior, 'C3', label='prior')
853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871
            ax.set_xlabel(self.theta_symbols[i])

            s = self.samples[:, i]
            while len(s) > 10**4:
                # random downsample to avoid slow calculation of kde
                s = np.random.choice(s, size=int(len(s)/2.))
            kde = gaussian_kde(s)
            ax2 = ax.twinx()
            postln = ax2.plot(x, kde.pdf(x), 'k', label='posterior')
            ax2.set_yticklabels([])
            ax.set_yticklabels([])

        lns = priorln + postln
        labs = [l.get_label() for l in lns]
        axes[0].legend(lns, labs, loc=1, framealpha=0.8)

        fig.savefig('{}/{}_prior_posterior.png'.format(
            self.outdir, self.label))

872
    def plot_cumulative_max(self, **kwargs):
873 874 875 876
        """ Plot the cumulative twoF for the maximum posterior estimate

        See the pyfstat.core.plot_twoF_cumulative function for further details
        """
Gregory Ashton's avatar
Gregory Ashton committed
877 878 879 880
        d, maxtwoF = self.get_max_twoF()
        for key, val in self.theta_prior.iteritems():
            if key not in d:
                d[key] = val
881

882 883 884
        if 'add_pfs' in kwargs:
            self.generate_loudest()

885
        if hasattr(self, 'search') is False:
886
            self._initiate_search_object()
887 888 889
        if self.binary is False:
            self.search.plot_twoF_cumulative(
                self.label, self.outdir, F0=d['F0'], F1=d['F1'], F2=d['F2'],
890
                Alpha=d['Alpha'], Delta=d['Delta'],
891
                tstart=self.minStartTime, tend=self.maxStartTime,
892
                **kwargs)
893 894 895 896 897
        else:
            self.search.plot_twoF_cumulative(
                self.label, self.outdir, F0=d['F0'], F1=d['F1'], F2=d['F2'],
                Alpha=d['Alpha'], Delta=d['Delta'], asini=d['asini'],
                period=d['period'], ecc=d['ecc'], argp=d['argp'], tp=d['argp'],
898
                tstart=self.minStartTime, tend=self.maxStartTime, **kwargs)
Gregory Ashton's avatar
Gregory Ashton committed
899

900
    def _generic_lnprior(self, **kwargs):
901 902 903 904
        """ Return a lambda function of the pdf

        Parameters
        ----------
905
        **kwargs:
906 907 908 909
            A dictionary containing 'type' of pdf and shape parameters

        """

Gregory Ashton's avatar
Gregory Ashton committed
910
        def log_of_unif(x, a, b):
911 912 913 914 915 916 917 918 919 920 921 922 923
            above = x < b
            below = x > a
            if type(above) is not np.ndarray:
                if above and below:
                    return -np.log(b-a)
                else:
                    return -np.inf
            else:
                idxs = np.array([all(tup) for tup in zip(above, below)])
                p = np.zeros(len(x)) - np.inf
                p[idxs] = -np.log(b-a)
                return p

Gregory Ashton's avatar
Gregory Ashton committed
924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939
        def log_of_log10unif(x, log10lower, log10upper):
            log10x = np.log10(x)
            above = log10x < log10upper
            below = log10x > log10lower
            if type(above) is not np.ndarray:
                if above and below:
                    return -np.log(x*np.log(10)*(log10upper-log10lower))
                else:
                    return -np.inf
            else:
                idxs = np.array([all(tup) for tup in zip(above, below)])
                p = np.zeros(len(x)) - np.inf
                p[idxs] = -np.log(x*np.log(10)*(log10upper-log10lower))
                return p

        def log_of_halfnorm(x, loc, scale):
940
            if x < loc:
941 942 943 944 945 946 947 948 949 950 951 952 953 954
                return -np.inf
            else:
                return -0.5*((x-loc)**2/scale**2+np.log(0.5*np.pi*scale**2))

        def cauchy(x, x0, gamma):
            return 1.0/(np.pi*gamma*(1+((x-x0)/gamma)**2))

        def exp(x, x0, gamma):
            if x > x0:
                return np.log(gamma) - gamma*(x - x0)
            else:
                return -np.inf

        if kwargs['type'] == 'unif':
Gregory Ashton's avatar
Gregory Ashton committed
955 956 957 958
            return lambda x: log_of_unif(x, kwargs['lower'], kwargs['upper'])
        if kwargs['type'] == 'log10unif':
            return lambda x: log_of_log10unif(
                x, kwargs['log10lower'], kwargs['log10upper'])
959
        elif kwargs['type'] == 'halfnorm':
Gregory Ashton's avatar
Gregory Ashton committed
960
            return lambda x: log_of_halfnorm(x, kwargs['loc'], kwargs['scale'])
961