mcmc_based_searches.py 94 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
88
89
90
91
92
93
94
95
96
97
98
99

    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.

    """
100
101

    symbol_dictionary = dict(
102
        F0='$f$', F1='$\dot{f}$', F2='$\ddot{f}$', Alpha=r'$\alpha$',
103
104
        Delta='$\delta$', asini='asini', period='P', ecc='ecc', tp='tp',
        argp='argp')
105
    unit_dictionary = dict(
106
107
        F0='Hz', F1='Hz/s', F2='Hz/s$^2$', Alpha=r'rad', Delta='rad',
        asini='', period='s', ecc='', tp='', argp='')
108
    transform_dictionary = {}
109

Gregory Ashton's avatar
Gregory Ashton committed
110
    @helper_functions.initializer
111
112
113
    def __init__(self, theta_prior, tref, label, outdir='data',
                 minStartTime=None, maxStartTime=None, sftfilepattern=None,
                 detectors=None, nsteps=[100, 100], nwalkers=100, ntemps=1,
114
                 log10beta_min=-5, theta_initial=None,
115
                 rhohatmax=1000, binary=False, BSGL=False,
Gregory Ashton's avatar
Gregory Ashton committed
116
                 SSBprec=None, minCoverFreq=None, maxCoverFreq=None,
David Keitel's avatar
David Keitel committed
117
118
                 injectSources=None, assumeSqrtSX=None,
                 transientWindowType=None):
119

Gregory Ashton's avatar
Gregory Ashton committed
120
121
        if os.path.isdir(outdir) is False:
            os.mkdir(outdir)
122
        self._add_log_file()
123
        logging.info('Set-up MCMC search for model {}'.format(self.label))
124
125
        if sftfilepattern:
            logging.info('Using data {}'.format(self.sftfilepattern))
126
        else:
127
            logging.info('No sftfilepattern given')
128
129
        if injectSources:
            logging.info('Inject sources: {}'.format(injectSources))
130
        self.pickle_path = '{}/{}_saved_data.p'.format(self.outdir, self.label)
131
        self._unpack_input_theta()
132
        self.ndim = len(self.theta_keys)
133
134
        if self.log10beta_min:
            self.betas = np.logspace(0, self.log10beta_min, self.ntemps)
135
136
        else:
            self.betas = None
137

138
139
140
        if args.clean and os.path.isfile(self.pickle_path):
            os.rename(self.pickle_path, self.pickle_path+".old")

141
        self._set_likelihoodcoef()
142
        self._log_input()
143
144
145

    def _set_likelihoodcoef(self):
        self.likelihoodcoef = np.log(70./self.rhohatmax**4)
146

147
    def _log_input(self):
148
        logging.info('theta_prior = {}'.format(self.theta_prior))
149
        logging.info('nwalkers={}'.format(self.nwalkers))
150
151
        logging.info('nsteps = {}'.format(self.nsteps))
        logging.info('ntemps = {}'.format(self.ntemps))
152
153
        logging.info('log10beta_min = {}'.format(
            self.log10beta_min))
154

155
    def _initiate_search_object(self):
156
        logging.info('Setting up search object')
157
        self.search = core.ComputeFstat(
158
            tref=self.tref, sftfilepattern=self.sftfilepattern,
159
            minCoverFreq=self.minCoverFreq, maxCoverFreq=self.maxCoverFreq,
David Keitel's avatar
David Keitel committed
160
161
            detectors=self.detectors, BSGL=self.BSGL,
            transientWindowType=self.transientWindowType,
162
            minStartTime=self.minStartTime, maxStartTime=self.maxStartTime,
163
            binary=self.binary, injectSources=self.injectSources,
164
            assumeSqrtSX=self.assumeSqrtSX, SSBprec=self.SSBprec)
165
166
167
168
        if self.minStartTime is None:
            self.minStartTime = self.search.minStartTime
        if self.maxStartTime is None:
            self.maxStartTime = self.search.maxStartTime
169
170

    def logp(self, theta_vals, theta_prior, theta_keys, search):
171
        H = [self._generic_lnprior(**theta_prior[key])(p) for p, key in
172
173
174
175
176
177
             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]
178
179
180
        twoF = search.get_fullycoherent_twoF(
            self.minStartTime, self.maxStartTime, *self.fixed_theta)
        return twoF/2.0 + self.likelihoodcoef
181

182
    def _unpack_input_theta(self):
183
        full_theta_keys = ['F0', 'F1', 'F2', 'Alpha', 'Delta']
184
185
186
        if self.binary:
            full_theta_keys += [
                'asini', 'period', 'ecc', 'tp', 'argp']
187
188
        full_theta_keys_copy = copy.copy(full_theta_keys)

189
190
        full_theta_symbols = ['$f$', '$\dot{f}$', '$\ddot{f}$', r'$\alpha$',
                              r'$\delta$']
191
192
        if self.binary:
            full_theta_symbols += [
193
                'asini', 'period', 'ecc', 'tp', 'argp']
194

195
196
        self.theta_keys = []
        fixed_theta_dict = {}
197
        for key, val in self.theta_prior.iteritems():
198
199
            if type(val) is dict:
                fixed_theta_dict[key] = 0
Gregory Ashton's avatar
Gregory Ashton committed
200
                self.theta_keys.append(key)
201
202
203
204
205
206
            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
207
            full_theta_keys_copy.pop(full_theta_keys_copy.index(key))
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222

        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]

223
224
225
226
227
228
229
230
231
    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

232
    def _check_initial_points(self, p0):
233
234
        for nt in range(self.ntemps):
            logging.info('Checking temperature {} chains'.format(nt))
235
236
            num = sum(self._evaluate_logpost(p0[nt]) == -np.inf)
            if num > 0:
237
238
                logging.warning(
                    'Of {} initial values, {} are -np.inf due to the prior'
239
                    .format(len(p0[0]), num))
240
                p0 = self._generate_new_p0_to_fix_initial_points(
241
                    p0, nt)
242

243
    def _generate_new_p0_to_fix_initial_points(self, p0, nt):
244
        logging.info('Attempting to correct intial values')
245
246
        init_logpost = self._evaluate_logpost(p0[nt])
        idxs = np.arange(self.nwalkers)[init_logpost == -np.inf]
247
        count = 0
248
        while sum(init_logpost == -np.inf) > 0 and count < 100:
249
250
251
            for j in idxs:
                p0[nt][j] = (p0[nt][np.random.randint(0, self.nwalkers)]*(
                             1+np.random.normal(0, 1e-10, self.ndim)))
252
            init_logpost = self._evaluate_logpost(p0[nt])
253
254
            count += 1

255
        if sum(init_logpost == -np.inf) > 0:
256
257
258
259
260
            logging.info('Failed to fix initial priors')
        else:
            logging.info('Suceeded to fix initial priors')

        return p0
261

262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
    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

286
287
288
289
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
#    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
399

400
401
        self.mean_acceptance_fraction = np.mean(
            sampler.acceptance_fraction, axis=1)
402
        logging.info("Mean acceptance fraction: {}"
403
                     .format(self.mean_acceptance_fraction))
404
        if self.ntemps > 1:
405
            self.tswap_acceptance_fraction = sampler.tswap_acceptance_fraction
406
407
            logging.info("Tswap acceptance fraction: {}"
                         .format(sampler.tswap_acceptance_fraction))
Gregory Ashton's avatar
Gregory Ashton committed
408
409
410
        self.autocorr_time = sampler.get_autocorr_time(window=window)
        logging.info("Autocorrelation length: {}".format(
            self.autocorr_time))
411
412
413

        return sampler

414
    def _estimate_run_time(self):
415
416
417
418
419
420
421
422
423
424
        """ 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
425
426
427
428
            tau0LD = 5.2e-7
            tau0T = 1.5e-8
            tau0S = 1.2e-4
            tau0C = 5.8e-6
429
        else:
Gregory Ashton's avatar
Gregory Ashton committed
430
            tau0LD = 1.3e-7
431
            tau0T = 1.5e-8
Gregory Ashton's avatar
Gregory Ashton committed
432
433
            tau0S = 9.1e-5
            tau0C = 5.5e-6
434
        Nsfts = (self.maxStartTime - self.minStartTime) / 1800.
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
        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

452
        logging.info('Estimated run-time = {} s = {:1.0f}:{:1.0f} m'.format(
453
            time, *divmod(time, 60)))
454

Gregory Ashton's avatar
Gregory Ashton committed
455
456
    def run(self, proposal_scale_factor=2, create_plots=True, window=50,
            **kwargs):
457
458
459
460
461
462
463
464
465
466
467
        """ 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
468
        window: int
469
470
            The minimum number of autocorrelation times needed to trust the
            result when estimating the autocorrelation time (see
Gregory Ashton's avatar
Gregory Ashton committed
471
            ptemcee.Sampler.get_autocorr_time for further details.
472
473
474
        **kwargs:
            Passed to _plot_walkers to control the figures

475
476
        Returns
        -------
Gregory Ashton's avatar
Gregory Ashton committed
477
478
        sampler: ptemcee.Sampler
            The ptemcee ptsampler object
479

480
        """
481

482
        self.old_data_is_okay_to_use = self._check_old_data_is_okay_to_use()
483
484
485
        if self.old_data_is_okay_to_use is True:
            logging.warning('Using saved data from {}'.format(
                self.pickle_path))
486
            d = self.get_saved_data_dictionary()
487
488
489
            self.samples = d['samples']
            self.lnprobs = d['lnprobs']
            self.lnlikes = d['lnlikes']
490
            self.all_lnlikelihood = d['all_lnlikelihood']
491
492
            return

493
        self._initiate_search_object()
494
        self._estimate_run_time()
495

Gregory Ashton's avatar
Gregory Ashton committed
496
497
498
        sampler = PTSampler(
            ntemps=self.ntemps, nwalkers=self.nwalkers, dim=self.ndim,
            logl=self.logl, logp=self.logp,
499
            logpargs=(self.theta_prior, self.theta_keys, self.search),
500
            loglargs=(self.search,), betas=self.betas, a=proposal_scale_factor)
501

502
503
504
        p0 = self._generate_initial_p0()
        p0 = self._apply_corrections_to_p0(p0)
        self._check_initial_points(p0)
505

506
        # Run initialisation steps if required
507
508
509
        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
510
                j, ninit_steps, n))
Gregory Ashton's avatar
Gregory Ashton committed
511
            sampler = self._run_sampler(sampler, p0, nburn=n, window=window)
512
            if create_plots:
513
                fig, axes = self._plot_walkers(sampler,
514
                                               **kwargs)
515
516
                fig.tight_layout()
                fig.savefig('{}/{}_init_{}_walkers.png'.format(
Gregory Ashton's avatar
Gregory Ashton committed
517
                    self.outdir, self.label, j))
518

519
520
521
            p0 = self._get_new_p0(sampler)
            p0 = self._apply_corrections_to_p0(p0)
            self._check_initial_points(p0)
522
523
            sampler.reset()

Gregory Ashton's avatar
Gregory Ashton committed
524
525
526
527
        if len(self.nsteps) > 1:
            nburn = self.nsteps[-2]
        else:
            nburn = 0
528
529
530
        nprod = self.nsteps[-1]
        logging.info('Running final burn and prod with {} steps'.format(
            nburn+nprod))
531
        sampler = self._run_sampler(sampler, p0, nburn=nburn, nprod=nprod)
532
        if create_plots:
533
            fig, axes = self._plot_walkers(sampler, nprod=nprod, **kwargs)
534
535
            fig.tight_layout()
            fig.savefig('{}/{}_walkers.png'.format(self.outdir, self.label),
Gregory Ashton's avatar
Gregory Ashton committed
536
                        )
537
538

        samples = sampler.chain[0, :, nburn:, :].reshape((-1, self.ndim))
Gregory Ashton's avatar
Gregory Ashton committed
539
540
541
        lnprobs = sampler.logprobability[0, :, nburn:].reshape((-1))
        lnlikes = sampler.loglikelihood[0, :, nburn:].reshape((-1))
        all_lnlikelihood = sampler.loglikelihood[:, :, nburn:]
542
543
544
        self.samples = samples
        self.lnprobs = lnprobs
        self.lnlikes = lnlikes
545
546
        self.all_lnlikelihood = all_lnlikelihood
        self._save_data(sampler, samples, lnprobs, lnlikes, all_lnlikelihood)
Gregory Ashton's avatar
Gregory Ashton committed
547
        return sampler
548

549
    def _get_rescale_multiplier_for_key(self, key):
550
        """ Get the rescale multiplier from the transform_dictionary
551
552
553
554
555

        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
        """
556
        if key not in self.transform_dictionary:
557
558
            return 1

559
560
        if 'multiplier' in self.transform_dictionary[key]:
            val = self.transform_dictionary[key]['multiplier']
561
562
563
            if type(val) == str:
                if hasattr(self, val):
                    multiplier = getattr(
564
                        self, self.transform_dictionary[key]['multiplier'])
565
566
567
568
569
570
571
572
573
                else:
                    raise ValueError(
                        "multiplier {} not a class attribute".format(val))
            else:
                multiplier = val
        else:
            multiplier = 1
        return multiplier

574
    def _get_rescale_subtractor_for_key(self, key):
575
        """ Get the rescale subtractor from the transform_dictionary
576
577
578
579
580

        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
        """
581
        if key not in self.transform_dictionary:
582
583
            return 0

584
585
        if 'subtractor' in self.transform_dictionary[key]:
            val = self.transform_dictionary[key]['subtractor']
586
587
588
            if type(val) == str:
                if hasattr(self, val):
                    subtractor = getattr(
589
                        self, self.transform_dictionary[key]['subtractor'])
590
591
592
593
594
595
596
597
598
                else:
                    raise ValueError(
                        "subtractor {} not a class attribute".format(val))
            else:
                subtractor = val
        else:
            subtractor = 0
        return subtractor

599
    def _scale_samples(self, samples, theta_keys):
600
        """ Scale the samples using the transform_dictionary """
601
        for key in theta_keys:
602
            if key in self.transform_dictionary:
603
604
                idx = theta_keys.index(key)
                s = samples[:, idx]
605
                subtractor = self._get_rescale_subtractor_for_key(key)
606
                s = s - subtractor
607
                multiplier = self._get_rescale_multiplier_for_key(key)
608
                s *= multiplier
609
610
                samples[:, idx] = s

611
612
        return samples

613
    def _get_labels(self, newline_units=False):
614
        """ Combine the units, symbols and rescaling to give labels """
615

616
617
618
619
620
621
        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]
622
623
624
625
626
627
628
            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']
629
            if label is None:
630
631
632
633
                if newline_units:
                    label = '{} \n [{}]'.format(s, u)
                else:
                    label = '{} [{}]'.format(s, u)
634
635
            labels.append(label)
        return labels
636

637
638
    def plot_corner(self, figsize=(7, 7), add_prior=False, nstds=None,
                    label_offset=0.4, dpi=300, rc_context={},
639
                    tglitch_ratio=False, fig_and_axes=None, save_fig=True,
640
                    **kwargs):
641
642
643
644
645
646
647
648
649
        """ 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)
650
651
652
        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.
653
654
655
656
657
658
659
660
661
662
663
664
665
666
        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
667
668
669
670
671
        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
672
673
        **kwargs:
            Passed to corner.corner
674

675
676
677
678
        Returns
        -------
        fig, axes:
            The matplotlib figure and axes, only returned if save_fig = False
679
680

        """
681

682
683
684
685
        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
686
687
        if self.ndim < 2:
            with plt.rc_context(rc_context):
688
689
690
691
                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
692
693
694
695
696
697
698
                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

699
        with plt.rc_context(rc_context):
700
701
702
703
704
            if fig_and_axes is None:
                fig, axes = plt.subplots(self.ndim, self.ndim,
                                         figsize=figsize)
            else:
                fig, axes = fig_and_axes
705
706

            samples_plt = copy.copy(self.samples)
707
            labels = self._get_labels(newline_units=True)
708

709
            samples_plt = self._scale_samples(samples_plt, self.theta_keys)
710
711
712
713
714

            if tglitch_ratio:
                for j, k in enumerate(self.theta_keys):
                    if k == 'tglitch':
                        s = samples_plt[:, j]
715
716
717
                        samples_plt[:, j] = (
                            s - self.minStartTime)/(
                                self.maxStartTime - self.minStartTime)
718
                        labels[j] = r'$R_{\textrm{glitch}}$'
719
720
721
722
723
724
725

            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))
726
727
            elif 'range' in kwargs:
                _range = kwargs.pop('range')
728
729
730
            else:
                _range = None

731
732
733
734
            hist_kwargs = kwargs.pop('hist_kwargs', dict())
            if 'normed' not in hist_kwargs:
                hist_kwargs['normed'] = True

735
            fig_triangle = corner.corner(samples_plt,
736
                                         labels=labels,
737
738
739
740
741
742
743
744
745
                                         fig=fig,
                                         bins=50,
                                         max_n_ticks=4,
                                         plot_contours=True,
                                         plot_datapoints=True,
                                         label_kwargs={'fontsize': 8},
                                         data_kwargs={'alpha': 0.1,
                                                      'ms': 0.5},
                                         range=_range,
746
                                         hist_kwargs=hist_kwargs,
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
                                         **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)
            plt.tight_layout(h_pad=0.0, w_pad=0.0)
            fig.subplots_adjust(hspace=0.05, wspace=0.05)

            if add_prior:
763
                self._add_prior_to_corner(axes, self.samples, add_prior)
764

765
766
767
768
769
            if save_fig:
                fig_triangle.savefig('{}/{}_corner.png'.format(
                    self.outdir, self.label), dpi=dpi)
            else:
                return fig, axes
770

771
    def _add_prior_to_corner(self, axes, samples, add_prior):
772
773
774
        for i, key in enumerate(self.theta_keys):
            ax = axes[i][i]
            s = samples[:, i]
775
776
777
778
779
780
781
782
783
784
            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)
785
786
            multiplier = self._get_rescale_multiplier_for_key(key)
            subtractor = self._get_rescale_subtractor_for_key(key)
787
788
789
790
791
792
793
794
            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])
795

796
797
798
799
800
801
802
803
    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]
804
            prior_func = self._generic_lnprior(**prior_dict)
805
806
807
808
809
            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
810
811
812
813
814
            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]
815
816
817
818
819
            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)
820
821
822
823
824
            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
825
826
827
828
829
            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]
830
831
832
            else:
                raise ValueError('Not implemented for prior type {}'.format(
                    prior_dict['type']))
833
            priorln = ax.plot(x, prior, 'C3', label='prior')
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
            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))

853
    def plot_cumulative_max(self, **kwargs):
854
855
856
857
        """ 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
858
859
860
861
        d, maxtwoF = self.get_max_twoF()
        for key, val in self.theta_prior.iteritems():
            if key not in d:
                d[key] = val
862

863
864
865
        if 'add_pfs' in kwargs:
            self.generate_loudest()

866
        if hasattr(self, 'search') is False:
867
            self._initiate_search_object()
868
869
870
        if self.binary is False:
            self.search.plot_twoF_cumulative(
                self.label, self.outdir, F0=d['F0'], F1=d['F1'], F2=d['F2'],
871
                Alpha=d['Alpha'], Delta=d['Delta'],
872
                tstart=self.minStartTime, tend=self.maxStartTime,
873
                **kwargs)
874
875
876
877
878
        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'],
879
                tstart=self.minStartTime, tend=self.maxStartTime, **kwargs)
Gregory Ashton's avatar
Gregory Ashton committed
880

881
    def _generic_lnprior(self, **kwargs):
882
883
884
885
        """ Return a lambda function of the pdf

        Parameters
        ----------
886
        **kwargs:
887
888
889
890
            A dictionary containing 'type' of pdf and shape parameters

        """

Gregory Ashton's avatar
Gregory Ashton committed
891
        def log_of_unif(x, a, b):
892
893
894
895
896
897
898
899
900
901
902
903
904
            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
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
        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):
921
            if x < loc:
922
923
924
925
926
927
928
929
930
931
932
933
934
935
                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
936
937
938
939
            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'])
940
        elif kwargs['type'] == 'halfnorm':
Gregory Ashton's avatar
Gregory Ashton committed
941
            return lambda x: log_of_halfnorm(x, kwargs['loc'], kwargs['scale'])
942
        elif kwargs['type'] == 'neghalfnorm':
Gregory Ashton's avatar
Gregory Ashton committed
943
944
            return lambda x: log_of_halfnorm(
                -x, kwargs['loc'], kwargs['scale'])
945
946
947
948
949
950
951
        elif kwargs['type'] == 'norm':
            return lambda x: -0.5*((x - kwargs['loc'])**2/kwargs['scale']**2
                                   + np.log(2*np.pi*kwargs['scale']**2))
        else:
            logging.info("kwargs:", kwargs)
            raise ValueError("Print unrecognise distribution")

952
    def _generate_rv(self, **kwargs):
953
954
955
        dist_type = kwargs.pop('type')
        if dist_type == "unif":
            return np.random.uniform(low=kwargs['lower'], high=kwargs['upper'])
Gregory Ashton's avatar
Gregory Ashton committed
956
957
958
        if dist_type == "log10unif":
            return 10**(np.random.uniform(low=kwargs['log10lower'],
                                          high=kwargs['log10upper']))
959
960
961
962
963
        if dist_type == "norm":
            return np.random.normal(loc=kwargs['loc'], scale=kwargs['scale'])
        if dist_type == "halfnorm":
            return np.abs(np.random.normal(loc=kwargs['loc'],
                                           scale=kwargs['scale']))
964
965
966
        if dist_type == "neghalfnorm":
            return -1 * np.abs(np.random.normal(loc=kwargs['loc'],
                                                scale=kwargs['scale']))
967
968
969
970
971
972
        if dist_type == "lognorm":
            return np.random.lognormal(
                mean=kwargs['loc'], sigma=kwargs['scale'])
        else:
            raise ValueError("dist_type {} unknown".format(dist_type))

973
    def _plot_walkers(self, sampler, symbols=None, alpha=0.8, color="k",
974
975
                      temp=0, lw=0.1, nprod=0, add_det_stat_burnin=False,
                      fig=None, axes=None, xoffset=0, plot_det_stat=False,
976
                      context='ggplot', labelpad=5):
977
978
        """ Plot all the chains from a sampler """

979
980
        if symbols is None:
            symbols = self._get_labels()
981
982
983
984
985
        if context not in plt.style.available:
            raise ValueError((
                'The requested context {} is not available; please select a'
                ' context from `plt.style.available`').format(context))

986
987
988
        if np.ndim(axes) > 1:
            axes = axes.flatten()

989
990
991
        shape = sampler.chain.shape
        if len(shape) == 3:
            nwalkers, nsteps, ndim = shape
992
            chain = sampler.chain[:, :, :].copy()
993
994
995
996
997
998
999
        if len(shape) == 4:
            ntemps, nwalkers, nsteps, ndim = shape
            if temp < ntemps:
                logging.info("Plotting temperature {} chains".format(temp))
            else:
                raise ValueError(("Requested temperature {} outside of"
                                  "available range").format(temp))
1000
            chain = sampler.chain[temp, :, :, :].copy()
For faster browsing, not all history is shown. View entire blame