mcmc_based_searches.py 81 KB
Newer Older
Gregory Ashton's avatar
Gregory Ashton committed
1
2
""" Searches using MCMC-based methods """

3
import sys
Gregory Ashton's avatar
Gregory Ashton committed
4
import os
5
import copy
Gregory Ashton's avatar
Gregory Ashton committed
6
import logging
7
from collections import OrderedDict
8
9
10
11
12
13
14
15

import numpy as np
import matplotlib
import matplotlib.pyplot as plt
import emcee
import corner
import dill as pickle

16
import core
Gregory Ashton's avatar
Gregory Ashton committed
17
from core import tqdm, args, earth_ephem, sun_ephem
18
from optimal_setup_functions import get_V_estimate
Gregory Ashton's avatar
Gregory Ashton committed
19
20
from optimal_setup_functions import get_optimal_setup
import helper_functions
21
22


23
class MCMCSearch(core.BaseSearchClass):
Gregory Ashton's avatar
Gregory Ashton committed
24
    """ MCMC search using ComputeFstat"""
Gregory Ashton's avatar
Gregory Ashton committed
25
    @helper_functions.initializer
Gregory Ashton's avatar
Gregory Ashton committed
26
27
    def __init__(self, label, outdir, theta_prior, tref, minStartTime,
                 maxStartTime, sftfilepath=None, nsteps=[100, 100],
28
29
                 nwalkers=100, ntemps=1, log10temperature_min=-5,
                 theta_initial=None, scatter_val=1e-10,
30
                 binary=False, BSGL=False, minCoverFreq=None,
31
                 maxCoverFreq=None, detectors=None, earth_ephem=None,
32
                 sun_ephem=None, injectSources=None, assumeSqrtSX=None):
33
34
35
36
        """
        Parameters
        label, outdir: str
            A label and directory to read/write data from/to
37
38
        sftfilepath: str
            File patern to match SFTs
39
        theta_prior: dict
40
41
42
43
            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.
44
45
46
47
        theta_initial: dict, array, (None)
            Either a dictionary of distribution about which to distribute the
            initial walkers about, an array (from which the walkers will be
            scattered by scatter_val, or  None in which case the prior is used.
48
        tref, minStartTime, maxStartTime: int
49
50
51
52
53
54
            GPS seconds of the reference time, start time and end time
        nsteps: list (m,)
            List specifying the number of steps to take, the last two entries
            give the nburn and nprod of the 'production' run, all entries
            before are for iterative initialisation steps (usually just one)
            e.g. [1000, 1000, 500].
55
56
57
58
59
60
61
62
        nwalkers, ntemps: int,
            The number of walkers and temperates to use in the parallel
            tempered PTSampler.
        log10temperature_min float < 0
            The  log_10(tmin) value, the set of betas passed to PTSampler are
            generated from np.logspace(0, log10temperature_min, ntemps).
        binary: Bool
            If true, search over binary parameters
63
        detectors: str
64
65
            Two character reference to the data to use, specify None for no
            contraint.
66
67
68
69
70
71
72
73
74
75
        minCoverFreq, maxCoverFreq: float
            Minimum and maximum instantaneous frequency which will be covered
            over the SFT time span as passed to CreateFstatInput
        earth_ephem, sun_ephem: str
            Paths of the two files containing positions of Earth and Sun,
            respectively at evenly spaced times, as passed to CreateFstatInput
            If None defaults defined in BaseSearchClass will be used

        """

Gregory Ashton's avatar
Gregory Ashton committed
76
77
        if os.path.isdir(outdir) is False:
            os.mkdir(outdir)
78
        self.add_log_file()
Gregory Ashton's avatar
Gregory Ashton committed
79
80
        logging.info(
            'Set-up MCMC search for model {} on data {}'.format(
81
                self.label, self.sftfilepath))
82
83
84
        self.pickle_path = '{}/{}_saved_data.p'.format(self.outdir, self.label)
        self.unpack_input_theta()
        self.ndim = len(self.theta_keys)
85
86
87
88
        if self.log10temperature_min:
            self.betas = np.logspace(0, self.log10temperature_min, self.ntemps)
        else:
            self.betas = None
89

90
91
92
93
94
95
96
97
        if earth_ephem is None:
            self.earth_ephem = self.earth_ephem_default
        if sun_ephem is None:
            self.sun_ephem = self.sun_ephem_default

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

Gregory Ashton's avatar
Gregory Ashton committed
98
99
100
101
102
103
        self.symbol_dictionary = dict(
            F0='$f$', F1='$\dot{f}$', F2='$\ddot{f}$', alpha=r'$\alpha$',
            delta='$\delta$')
        self.unit_dictionary = dict(
            F0='Hz', F1='Hz/s', F2='Hz/s$^2$', alpha=r'rad', delta='rad')

104
105
106
        self.log_input()

    def log_input(self):
107
        logging.info('theta_prior = {}'.format(self.theta_prior))
108
        logging.info('nwalkers={}'.format(self.nwalkers))
109
110
111
112
        logging.info('scatter_val = {}'.format(self.scatter_val))
        logging.info('nsteps = {}'.format(self.nsteps))
        logging.info('ntemps = {}'.format(self.ntemps))
        logging.info('log10temperature_min = {}'.format(
113
            self.log10temperature_min))
114

Gregory Ashton's avatar
Gregory Ashton committed
115
    def initiate_search_object(self):
116
        logging.info('Setting up search object')
117
        self.search = core.ComputeFstat(
118
119
120
            tref=self.tref, sftfilepath=self.sftfilepath,
            minCoverFreq=self.minCoverFreq, maxCoverFreq=self.maxCoverFreq,
            earth_ephem=self.earth_ephem, sun_ephem=self.sun_ephem,
121
            detectors=self.detectors, BSGL=self.BSGL, transient=False,
122
            minStartTime=self.minStartTime, maxStartTime=self.maxStartTime,
123
124
            binary=self.binary, injectSources=self.injectSources,
            assumeSqrtSX=self.assumeSqrtSX)
125
126

    def logp(self, theta_vals, theta_prior, theta_keys, search):
Gregory Ashton's avatar
Gregory Ashton committed
127
        H = [self.generic_lnprior(**theta_prior[key])(p) for p, key in
128
129
130
131
132
133
             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]
134
135
        FS = search.compute_fullycoherent_det_stat_single_point(
            *self.fixed_theta)
136
137
138
        return FS

    def unpack_input_theta(self):
139
        full_theta_keys = ['F0', 'F1', 'F2', 'Alpha', 'Delta']
140
141
142
        if self.binary:
            full_theta_keys += [
                'asini', 'period', 'ecc', 'tp', 'argp']
143
144
        full_theta_keys_copy = copy.copy(full_theta_keys)

145
146
        full_theta_symbols = ['$f$', '$\dot{f}$', '$\ddot{f}$', r'$\alpha$',
                              r'$\delta$']
147
148
149
150
        if self.binary:
            full_theta_symbols += [
                'asini', 'period', 'period', 'ecc', 'tp', 'argp']

151
152
        self.theta_keys = []
        fixed_theta_dict = {}
153
        for key, val in self.theta_prior.iteritems():
154
155
            if type(val) is dict:
                fixed_theta_dict[key] = 0
Gregory Ashton's avatar
Gregory Ashton committed
156
                self.theta_keys.append(key)
157
158
159
160
161
162
            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
163
            full_theta_keys_copy.pop(full_theta_keys_copy.index(key))
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179

        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]

    def check_initial_points(self, p0):
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
        for nt in range(self.ntemps):
            logging.info('Checking temperature {} chains'.format(nt))
            initial_priors = np.array([
                self.logp(p, self.theta_prior, self.theta_keys, self.search)
                for p in p0[nt]])
            number_of_initial_out_of_bounds = sum(initial_priors == -np.inf)

            if number_of_initial_out_of_bounds > 0:
                logging.warning(
                    'Of {} initial values, {} are -np.inf due to the prior'
                    .format(len(initial_priors),
                            number_of_initial_out_of_bounds))

                p0 = self.generate_new_p0_to_fix_initial_points(
                    p0, nt, initial_priors)

    def generate_new_p0_to_fix_initial_points(self, p0, nt, initial_priors):
        logging.info('Attempting to correct intial values')
        idxs = np.arange(self.nwalkers)[initial_priors == -np.inf]
        count = 0
        while sum(initial_priors == -np.inf) > 0 and count < 100:
            for j in idxs:
                p0[nt][j] = (p0[nt][np.random.randint(0, self.nwalkers)]*(
                             1+np.random.normal(0, 1e-10, self.ndim)))
            initial_priors = np.array([
                self.logp(p, self.theta_prior, self.theta_keys,
                          self.search)
                for p in p0[nt]])
            count += 1

        if sum(initial_priors == -np.inf) > 0:
            logging.info('Failed to fix initial priors')
        else:
            logging.info('Suceeded to fix initial priors')

        return p0
216

217
    def OLD_run_sampler_with_progress_bar(self, sampler, ns, p0):
218
219
        for result in tqdm(sampler.sample(p0, iterations=ns), total=ns):
            pass
Gregory Ashton's avatar
Gregory Ashton committed
220
221
        return sampler

222
223
    def setup_convergence_testing(
            self, convergence_period=10, convergence_length=10,
224
            convergence_burnin_fraction=0.25, convergence_threshold_number=10,
225
226
            convergence_threshold=1.2, convergence_prod_threshold=2,
            convergence_plot_upper_lim=2):
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
        """
        If called, convergence testing is used during the MCMC simulation

        This uses the Gelmanr-Rubin statistic based on the ratio of between and
        within walkers variance. The original statistic was developed for
        multiple (independent) MCMC simulations, in this context we simply use
        the walkers

        Parameters
        ----------
        convergence_period: int
            period (in number of steps) at which to test convergence
        convergence_length: int
            number of steps to use in testing convergence - this should be
            large enough to measure the variance, but if it is too long
            this will result in incorect early convergence tests
        convergence_burnin_fraction: float [0, 1]
            the fraction of the burn-in period after which to start testing
        convergence_threshold_number: int
            the number of consecutive times where the test passes after which
            to break the burn-in and go to production
        convergence_threshold: float
            the threshold to use in diagnosing convergence. Gelman & Rubin
            recomend a value of 1.2, 1.1 for strict convergence
        convergence_prod_threshold: float
            the threshold to test the production values with
253
254
        convergence_plot_upper_lim: float
            the upper limit to use in the diagnostic plot
255
        """
256
257
258
259
260
261
262

        if convergence_length > convergence_period:
            raise ValueError('convergence_length must be < convergence_period')
        logging.info('Setting up convergence testing')
        self.convergence_length = convergence_length
        self.convergence_period = convergence_period
        self.convergence_burnin_fraction = convergence_burnin_fraction
263
        self.convergence_prod_threshold = convergence_prod_threshold
264
265
266
267
268
        self.convergence_diagnostic = []
        self.convergence_diagnosticx = []
        self.convergence_threshold_number = convergence_threshold_number
        self.convergence_threshold = convergence_threshold
        self.convergence_number = 0
269
        self.convergence_plot_upper_lim = convergence_plot_upper_lim
270

271
    def get_convergence_statistic(self, i, sampler):
272
273
274
275
276
277
278
279
280
        s = sampler.chain[0, :, i-self.convergence_length+1:i+1, :]
        within_std = 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)
        between_std = np.sqrt(np.mean((per_walker_mean-mean)**2, axis=0))
        W = within_std
        B_over_n = between_std**2 / self.convergence_period
        Vhat = ((self.convergence_period-1.)/self.convergence_period * W
                + B_over_n + B_over_n / float(self.nwalkers))
281
        c = np.sqrt(Vhat/W)
282
        self.convergence_diagnostic.append(c)
283
        self.convergence_diagnosticx.append(i - self.convergence_length/2)
284
285
        return c

286
    def burnin_convergence_test(self, i, sampler, nburn):
287
288
        if i < self.convergence_burnin_fraction*nburn:
            return False
289
        if np.mod(i+1, self.convergence_period) != 0:
290
291
            return False
        c = self.get_convergence_statistic(i, sampler)
292
293
        if np.all(c < self.convergence_threshold):
            self.convergence_number += 1
294
295
        else:
            self.convergence_number = 0
296
297
        return self.convergence_number > self.convergence_threshold_number

298
299
300
301
302
303
    def prod_convergence_test(self, i, sampler, nburn):
        testA = i > nburn + self.convergence_length
        testB = np.mod(i+1, self.convergence_period) == 0
        if testA and testB:
            self.get_convergence_statistic(i, sampler)

304
305
306
307
308
309
310
311
312
    def check_production_convergence(self, k):
        bools = np.any(
            np.array(self.convergence_diagnostic)[k:, :]
            > self.convergence_prod_threshold, axis=1)
        if np.any(bools):
            logging.warning(
                '{} convergence tests in the production run of {} failed'
                .format(np.sum(bools), len(bools)))

313
314
    def run_sampler(self, sampler, p0, nprod=0, nburn=0):
        if hasattr(self, 'convergence_period'):
315
316
317
318
            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):
319
                if self.burnin_convergence_test(i, sampler, nburn):
320
321
322
323
324
325
326
327
328
329
330
                    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
            k = len(self.convergence_diagnostic)
            for result in tqdm(sampler.sample(output[0], iterations=nprod),
                               total=nprod):
331
                self.prod_convergence_test(j, sampler, nburn)
332
333
                j += 1
            self.check_production_convergence(k)
334
335
336
337
338
339
            return sampler
        else:
            for result in tqdm(sampler.sample(p0, iterations=nburn+nprod),
                               total=nburn+nprod):
                pass
            return sampler
340

341
    def run(self, proposal_scale_factor=2, create_plots=True, **kwargs):
342

Gregory Ashton's avatar
Gregory Ashton committed
343
        self.old_data_is_okay_to_use = self.check_old_data_is_okay_to_use()
344
345
346
347
348
349
350
351
352
353
        if self.old_data_is_okay_to_use is True:
            logging.warning('Using saved data from {}'.format(
                self.pickle_path))
            d = self.get_saved_data()
            self.sampler = d['sampler']
            self.samples = d['samples']
            self.lnprobs = d['lnprobs']
            self.lnlikes = d['lnlikes']
            return

Gregory Ashton's avatar
Gregory Ashton committed
354
        self.initiate_search_object()
355
356
357
358

        sampler = emcee.PTSampler(
            self.ntemps, self.nwalkers, self.ndim, self.logl, self.logp,
            logpargs=(self.theta_prior, self.theta_keys, self.search),
359
            loglargs=(self.search,), betas=self.betas, a=proposal_scale_factor)
360

Gregory Ashton's avatar
Gregory Ashton committed
361
362
        p0 = self.generate_initial_p0()
        p0 = self.apply_corrections_to_p0(p0)
363
364
365
366
367
        self.check_initial_points(p0)

        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
368
                j, ninit_steps, n))
369
            sampler = self.run_sampler(sampler, p0, nburn=n)
370
371
            logging.info("Mean acceptance fraction: {}"
                         .format(np.mean(sampler.acceptance_fraction, axis=1)))
372
373
374
            if self.ntemps > 1:
                logging.info("Tswap acceptance fraction: {}"
                             .format(sampler.tswap_acceptance_fraction))
375
376
377
378
379
380
            if create_plots:
                fig, axes = self.plot_walkers(sampler,
                                              symbols=self.theta_symbols,
                                              **kwargs)
                fig.tight_layout()
                fig.savefig('{}/{}_init_{}_walkers.png'.format(
381
                    self.outdir, self.label, j), dpi=400)
382

383
            p0 = self.get_new_p0(sampler)
Gregory Ashton's avatar
Gregory Ashton committed
384
            p0 = self.apply_corrections_to_p0(p0)
385
386
387
            self.check_initial_points(p0)
            sampler.reset()

Gregory Ashton's avatar
Gregory Ashton committed
388
389
390
391
        if len(self.nsteps) > 1:
            nburn = self.nsteps[-2]
        else:
            nburn = 0
392
393
394
        nprod = self.nsteps[-1]
        logging.info('Running final burn and prod with {} steps'.format(
            nburn+nprod))
395
        sampler = self.run_sampler(sampler, p0, nburn=nburn, nprod=nprod)
396
397
        logging.info("Mean acceptance fraction: {}"
                     .format(np.mean(sampler.acceptance_fraction, axis=1)))
398
399
400
        if self.ntemps > 1:
            logging.info("Tswap acceptance fraction: {}"
                         .format(sampler.tswap_acceptance_fraction))
401

402
403
        if create_plots:
            fig, axes = self.plot_walkers(sampler, symbols=self.theta_symbols,
404
                                          nprod=nprod, **kwargs)
405
406
407
            fig.tight_layout()
            fig.savefig('{}/{}_walkers.png'.format(self.outdir, self.label),
                        dpi=200)
408
409
410
411
412
413
414
415
416
417

        samples = sampler.chain[0, :, nburn:, :].reshape((-1, self.ndim))
        lnprobs = sampler.lnprobability[0, :, nburn:].reshape((-1))
        lnlikes = sampler.lnlikelihood[0, :, nburn:].reshape((-1))
        self.sampler = sampler
        self.samples = samples
        self.lnprobs = lnprobs
        self.lnlikes = lnlikes
        self.save_data(sampler, samples, lnprobs, lnlikes)

418
    def plot_corner(self, figsize=(7, 7),  tglitch_ratio=False,
419
420
421
                    add_prior=False, nstds=None, label_offset=0.4,
                    dpi=300, rc_context={}, **kwargs):

Gregory Ashton's avatar
Gregory Ashton committed
422
423
424
425
426
427
428
429
430
431
        if self.ndim < 2:
            with plt.rc_context(rc_context):
                fig, ax = plt.subplots(figsize=figsize)
                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

432
433
434
435
436
437
        with plt.rc_context(rc_context):
            fig, axes = plt.subplots(self.ndim, self.ndim,
                                     figsize=figsize)

            samples_plt = copy.copy(self.samples)
            theta_symbols_plt = copy.copy(self.theta_symbols)
438
439
            theta_symbols_plt = [s.replace('_{glitch}', r'_\textrm{glitch}')
                                 for s in theta_symbols_plt]
440
441
442
443
444

            if tglitch_ratio:
                for j, k in enumerate(self.theta_keys):
                    if k == 'tglitch':
                        s = samples_plt[:, j]
445
446
447
                        samples_plt[:, j] = (
                            s - self.minStartTime)/(
                                self.maxStartTime - self.minStartTime)
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
                        theta_symbols_plt[j] = r'$R_{\textrm{glitch}}$'

            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))
            else:
                _range = None

            fig_triangle = corner.corner(samples_plt,
                                         labels=theta_symbols_plt,
                                         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,
                                         **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:
                self.add_prior_to_corner(axes, samples_plt)

            fig_triangle.savefig('{}/{}_corner.png'.format(
                self.outdir, self.label), dpi=dpi)
490
491
492
493
494
495

    def add_prior_to_corner(self, axes, samples):
        for i, key in enumerate(self.theta_keys):
            ax = axes[i][i]
            xlim = ax.get_xlim()
            s = samples[:, i]
Gregory Ashton's avatar
Gregory Ashton committed
496
            prior = self.generic_lnprior(**self.theta_prior[key])
497
498
499
500
501
502
            x = np.linspace(s.min(), s.max(), 100)
            ax2 = ax.twinx()
            ax2.get_yaxis().set_visible(False)
            ax2.plot(x, [prior(xi) for xi in x], '-r')
            ax.set_xlim(xlim)

503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
    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]
            prior_func = self.generic_lnprior(**prior_dict)
            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
            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)
522
523
524
525
526
            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
527
528
529
530
531
            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]
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
            else:
                raise ValueError('Not implemented for prior type {}'.format(
                    prior_dict['type']))
            priorln = ax.plot(x, prior, 'r', label='prior')
            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))

555
    def plot_cumulative_max(self, **kwargs):
Gregory Ashton's avatar
Gregory Ashton committed
556
557
558
559
        d, maxtwoF = self.get_max_twoF()
        for key, val in self.theta_prior.iteritems():
            if key not in d:
                d[key] = val
560
561

        if hasattr(self, 'search') is False:
Gregory Ashton's avatar
Gregory Ashton committed
562
            self.initiate_search_object()
563
564
565
        if self.binary is False:
            self.search.plot_twoF_cumulative(
                self.label, self.outdir, F0=d['F0'], F1=d['F1'], F2=d['F2'],
566
                Alpha=d['Alpha'], Delta=d['Delta'],
567
                tstart=self.minStartTime, tend=self.maxStartTime,
568
                **kwargs)
569
570
571
572
573
        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'],
574
                tstart=self.minStartTime, tend=self.maxStartTime, **kwargs)
Gregory Ashton's avatar
Gregory Ashton committed
575

Gregory Ashton's avatar
Gregory Ashton committed
576
    def generic_lnprior(self, **kwargs):
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
        """ Return a lambda function of the pdf

        Parameters
        ----------
        kwargs: dict
            A dictionary containing 'type' of pdf and shape parameters

        """

        def logunif(x, a, b):
            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

        def halfnorm(x, loc, scale):
601
            if x < loc:
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
                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':
            return lambda x: logunif(x, kwargs['lower'], kwargs['upper'])
        elif kwargs['type'] == 'halfnorm':
            return lambda x: halfnorm(x, kwargs['loc'], kwargs['scale'])
619
620
        elif kwargs['type'] == 'neghalfnorm':
            return lambda x: halfnorm(-x, kwargs['loc'], kwargs['scale'])
621
622
623
624
625
626
627
        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")

Gregory Ashton's avatar
Gregory Ashton committed
628
    def generate_rv(self, **kwargs):
629
630
631
632
633
634
635
636
        dist_type = kwargs.pop('type')
        if dist_type == "unif":
            return np.random.uniform(low=kwargs['lower'], high=kwargs['upper'])
        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']))
637
638
639
        if dist_type == "neghalfnorm":
            return -1 * np.abs(np.random.normal(loc=kwargs['loc'],
                                                scale=kwargs['scale']))
640
641
642
643
644
645
        if dist_type == "lognorm":
            return np.random.lognormal(
                mean=kwargs['loc'], sigma=kwargs['scale'])
        else:
            raise ValueError("dist_type {} unknown".format(dist_type))

Gregory Ashton's avatar
Gregory Ashton committed
646
    def plot_walkers(self, sampler, symbols=None, alpha=0.4, color="k", temp=0,
647
                     lw=0.1, nprod=0, add_det_stat_burnin=False,
648
                     fig=None, axes=None, xoffset=0, plot_det_stat=True,
649
                     context='classic', subtractions=None, labelpad=0.05):
650
651
        """ Plot all the chains from a sampler """

652
653
654
        if np.ndim(axes) > 1:
            axes = axes.flatten()

655
656
657
658
659
660
661
662
663
664
665
666
667
        shape = sampler.chain.shape
        if len(shape) == 3:
            nwalkers, nsteps, ndim = shape
            chain = sampler.chain[:, :, :]
        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))
            chain = sampler.chain[temp, :, :, :]

668
669
        if subtractions is None:
            subtractions = [0 for i in range(ndim)]
670
671
672
        else:
            if len(subtractions) != self.ndim:
                raise ValueError('subtractions must be of length ndim')
673

674
        with plt.style.context((context)):
Gregory Ashton's avatar
Gregory Ashton committed
675
            plt.rcParams['text.usetex'] = True
Gregory Ashton's avatar
Gregory Ashton committed
676
            if fig is None and axes is None:
677
                fig = plt.figure(figsize=(4, 3.0*ndim))
Gregory Ashton's avatar
Gregory Ashton committed
678
                ax = fig.add_subplot(ndim+1, 1, 1)
Gregory Ashton's avatar
Gregory Ashton committed
679
                axes = [ax] + [fig.add_subplot(ndim+1, 1, i)
Gregory Ashton's avatar
Gregory Ashton committed
680
                               for i in range(2, ndim+1)]
681

Gregory Ashton's avatar
Gregory Ashton committed
682
            idxs = np.arange(chain.shape[1])
683
684
685
686
687
            burnin_idx = chain.shape[1] - nprod
            if hasattr(self, 'convergence_idx'):
                convergence_idx = self.convergence_idx
            else:
                convergence_idx = burnin_idx
688
689
            if ndim > 1:
                for i in range(ndim):
690
                    axes[i].ticklabel_format(useOffset=False, axis='y')
Gregory Ashton's avatar
Gregory Ashton committed
691
                    cs = chain[:, :, i].T
692
693
694
                    if burnin_idx > 0:
                        axes[i].plot(xoffset+idxs[:convergence_idx],
                                     cs[:convergence_idx]-subtractions[i],
695
                                     color="r", alpha=alpha,
Gregory Ashton's avatar
Gregory Ashton committed
696
                                     lw=lw)
697
698
                    axes[i].plot(xoffset+idxs[burnin_idx:],
                                 cs[burnin_idx:]-subtractions[i],
Gregory Ashton's avatar
Gregory Ashton committed
699
                                 color="k", alpha=alpha, lw=lw)
700
                    if symbols:
701
                        if subtractions[i] == 0:
702
                            axes[i].set_ylabel(symbols[i], labelpad=labelpad)
703
704
                        else:
                            axes[i].set_ylabel(
705
706
                                symbols[i]+'$-$'+symbols[i]+'$_0$',
                                labelpad=labelpad)
707

708
709
                    if hasattr(self, 'convergence_diagnostic'):
                        ax = axes[i].twinx()
710
711
                        c_x = np.array(self.convergence_diagnosticx)
                        c_y = np.array(self.convergence_diagnostic)
712
713
714
715
                        break_idx = np.argmin(np.abs(c_x - burnin_idx))
                        ax.plot(c_x[:break_idx], c_y[:break_idx, i], '-b')
                        ax.plot(c_x[break_idx:], c_y[break_idx:, i], '-b')
                        ax.set_ylabel('PSRF')
716
                        ax.ticklabel_format(useOffset=False)
717
                        ax.set_ylim(1, self.convergence_plot_upper_lim)
718
            else:
Gregory Ashton's avatar
Gregory Ashton committed
719
                axes[0].ticklabel_format(useOffset=False, axis='y')
Gregory Ashton's avatar
Gregory Ashton committed
720
                cs = chain[:, :, temp].T
Gregory Ashton's avatar
Gregory Ashton committed
721
722
723
724
725
726
                if burnin_idx:
                    axes[0].plot(idxs[:burnin_idx], cs[:burnin_idx],
                                 color="r", alpha=alpha, lw=lw)
                axes[0].plot(idxs[burnin_idx:], cs[burnin_idx:], color="k",
                             alpha=alpha, lw=lw)
                if symbols:
727
                    axes[0].set_ylabel(symbols[0], labelpad=labelpad)
728

729
            if plot_det_stat:
730
731
732
                if len(axes) == ndim:
                    axes.append(fig.add_subplot(ndim+1, 1, ndim+1))

733
734
735
                lnl = sampler.lnlikelihood[temp, :, :]
                if burnin_idx and add_det_stat_burnin:
                    burn_in_vals = lnl[:, :burnin_idx].flatten()
736
737
738
739
740
741
742
                    try:
                        axes[-1].hist(burn_in_vals[~np.isnan(burn_in_vals)],
                                      bins=50, histtype='step', color='r')
                    except ValueError:
                        logging.info('Det. Stat. hist failed, most likely all '
                                     'values where the same')
                        pass
743
744
745
                else:
                    burn_in_vals = []
                prod_vals = lnl[:, burnin_idx:].flatten()
746
747
748
749
750
751
752
                try:
                    axes[-1].hist(prod_vals[~np.isnan(prod_vals)], bins=50,
                                  histtype='step', color='k')
                except ValueError:
                    logging.info('Det. Stat. hist failed, most likely all '
                                 'values where the same')
                    pass
753
754
755
756
757
758
759
760
761
762
763
764
                if self.BSGL:
                    axes[-1].set_xlabel(r'$\mathcal{B}_\mathrm{S/GL}$')
                else:
                    axes[-1].set_xlabel(r'$\widetilde{2\mathcal{F}}$')
                axes[-1].set_ylabel(r'$\textrm{Counts}$')
                combined_vals = np.append(burn_in_vals, prod_vals)
                if len(combined_vals) > 0:
                    minv = np.min(combined_vals)
                    maxv = np.max(combined_vals)
                    Range = abs(maxv-minv)
                    axes[-1].set_xlim(minv-0.1*Range, maxv+0.1*Range)

765
                xfmt = matplotlib.ticker.ScalarFormatter()
766
                xfmt.set_powerlimits((-4, 4))
767
768
                axes[-1].xaxis.set_major_formatter(xfmt)

769
            axes[-2].set_xlabel(r'$\textrm{Number of steps}$', labelpad=0.2)
770
771
        return fig, axes

Gregory Ashton's avatar
Gregory Ashton committed
772
773
774
775
776
    def apply_corrections_to_p0(self, p0):
        """ Apply any correction to the initial p0 values """
        return p0

    def generate_scattered_p0(self, p):
777
        """ Generate a set of p0s scattered about p """
Gregory Ashton's avatar
Gregory Ashton committed
778
        p0 = [[p + self.scatter_val * p * np.random.randn(self.ndim)
779
780
781
782
               for i in xrange(self.nwalkers)]
              for j in xrange(self.ntemps)]
        return p0

Gregory Ashton's avatar
Gregory Ashton committed
783
    def generate_initial_p0(self):
784
785
786
        """ Generate a set of init vals for the walkers """

        if type(self.theta_initial) == dict:
787
            logging.info('Generate initial values from initial dictionary')
788
            if hasattr(self, 'nglitch') and self.nglitch > 1:
789
                raise ValueError('Initial dict not implemented for nglitch>1')
Gregory Ashton's avatar
Gregory Ashton committed
790
            p0 = [[[self.generate_rv(**self.theta_initial[key])
791
792
793
                    for key in self.theta_keys]
                   for i in range(self.nwalkers)]
                  for j in range(self.ntemps)]
794
795
796
797
798
799
        elif type(self.theta_initial) == list:
            logging.info('Generate initial values from list of theta_initial')
            p0 = [[[self.generate_rv(**val)
                    for val in self.theta_initial]
                   for i in range(self.nwalkers)]
                  for j in range(self.ntemps)]
800
        elif self.theta_initial is None:
801
            logging.info('Generate initial values from prior dictionary')
Gregory Ashton's avatar
Gregory Ashton committed
802
            p0 = [[[self.generate_rv(**self.theta_prior[key])
803
804
805
806
                    for key in self.theta_keys]
                   for i in range(self.nwalkers)]
                  for j in range(self.ntemps)]
        elif len(self.theta_initial) == self.ndim:
Gregory Ashton's avatar
Gregory Ashton committed
807
            p0 = self.generate_scattered_p0(self.theta_initial)
808
809
810
811
812
        else:
            raise ValueError('theta_initial not understood')

        return p0

813
    def get_new_p0(self, sampler):
814
815
816
817
818
819
        """ Returns new initial positions for walkers are burn0 stage

        This returns new positions for all walkers by scattering points about
        the maximum posterior with scale `scatter_val`.

        """
Gregory Ashton's avatar
Gregory Ashton committed
820
821
822
823
        temp_idx = 0
        pF = sampler.chain[temp_idx, :, :, :]
        lnl = sampler.lnlikelihood[temp_idx, :, :]
        lnp = sampler.lnprobability[temp_idx, :, :]
824
825

        # General warnings about the state of lnp
Gregory Ashton's avatar
Gregory Ashton committed
826
        if np.any(np.isnan(lnp)):
827
828
            logging.warning(
                "Of {} lnprobs {} are nan".format(
Gregory Ashton's avatar
Gregory Ashton committed
829
830
                    np.shape(lnp), np.sum(np.isnan(lnp))))
        if np.any(np.isposinf(lnp)):
831
832
            logging.warning(
                "Of {} lnprobs {} are +np.inf".format(
Gregory Ashton's avatar
Gregory Ashton committed
833
834
                    np.shape(lnp), np.sum(np.isposinf(lnp))))
        if np.any(np.isneginf(lnp)):
835
836
            logging.warning(
                "Of {} lnprobs {} are -np.inf".format(
Gregory Ashton's avatar
Gregory Ashton committed
837
                    np.shape(lnp), np.sum(np.isneginf(lnp))))
838

839
840
        lnp_finite = copy.copy(lnp)
        lnp_finite[np.isinf(lnp)] = np.nan
Gregory Ashton's avatar
Gregory Ashton committed
841
842
        idx = np.unravel_index(np.nanargmax(lnp_finite), lnp_finite.shape)
        p = pF[idx]
843
        p0 = self.generate_scattered_p0(p)
844

845
846
847
848
849
850
851
852
        self.search.BSGL = False
        twoF = self.logl(p, self.search)
        self.search.BSGL = self.BSGL

        logging.info(('Gen. new p0 from pos {} which had det. stat.={:2.1f},'
                      ' twoF={:2.1f} and lnp={:2.1f}')
                     .format(idx[1], lnl[idx], twoF, lnp_finite[idx]))

853
854
855
856
857
        return p0

    def get_save_data_dictionary(self):
        d = dict(nsteps=self.nsteps, nwalkers=self.nwalkers,
                 ntemps=self.ntemps, theta_keys=self.theta_keys,
Gregory Ashton's avatar
Gregory Ashton committed
858
                 theta_prior=self.theta_prior, scatter_val=self.scatter_val,
859
                 log10temperature_min=self.log10temperature_min,
860
                 BSGL=self.BSGL)
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
        return d

    def save_data(self, sampler, samples, lnprobs, lnlikes):
        d = self.get_save_data_dictionary()
        d['sampler'] = sampler
        d['samples'] = samples
        d['lnprobs'] = lnprobs
        d['lnlikes'] = lnlikes

        if os.path.isfile(self.pickle_path):
            logging.info('Saving backup of {} as {}.old'.format(
                self.pickle_path, self.pickle_path))
            os.rename(self.pickle_path, self.pickle_path+".old")
        with open(self.pickle_path, "wb") as File:
            pickle.dump(d, File)

    def get_saved_data(self):
        with open(self.pickle_path, "r") as File:
            d = pickle.load(File)
        return d

    def check_old_data_is_okay_to_use(self):
883
884
885
886
        if args.use_old_data:
            logging.info("Forcing use of old data")
            return True

887
888
889
890
        if os.path.isfile(self.pickle_path) is False:
            logging.info('No pickled data found')
            return False

Gregory Ashton's avatar
Gregory Ashton committed
891
892
893
894
895
896
        if self.sftfilepath is not None:
            oldest_sft = min([os.path.getmtime(f) for f in
                              self.get_list_of_matching_sfts()])
            if os.path.getmtime(self.pickle_path) < oldest_sft:
                logging.info('Pickled data outdates sft files')
                return False
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911

        old_d = self.get_saved_data().copy()
        new_d = self.get_save_data_dictionary().copy()

        old_d.pop('samples')
        old_d.pop('sampler')
        old_d.pop('lnprobs')
        old_d.pop('lnlikes')

        mod_keys = []
        for key in new_d.keys():
            if key in old_d:
                if new_d[key] != old_d[key]:
                    mod_keys.append((key, old_d[key], new_d[key]))
            else:
912
                raise ValueError('Keys {} not in old dictionary'.format(key))
913
914
915
916
917
918
919
920
921

        if len(mod_keys) == 0:
            return True
        else:
            logging.warning("Saved data differs from requested")
            logging.info("Differences found in following keys:")
            for key in mod_keys:
                if len(key) == 3:
                    if np.isscalar(key[1]) or key[0] == 'nsteps':
922
                        logging.info("    {} : {} -> {}".format(*key))
923
                    else:
924
                        logging.info("    " + key[0])
925
926
927
928
929
                else:
                    logging.info(key)
            return False

    def get_max_twoF(self, threshold=0.05):
930
        """ Returns the max likelihood sample and the corresponding 2F value
931
932
933
934
935
936
937
938
939
940
941
942
943
944

        Note: the sample is returned as a dictionary along with an estimate of
        the standard deviation calculated from the std of all samples with a
        twoF within `threshold` (relative) to the max twoF

        """
        if any(np.isposinf(self.lnlikes)):
            logging.info('twoF values contain positive infinite values')
        if any(np.isneginf(self.lnlikes)):
            logging.info('twoF values contain negative infinite values')
        if any(np.isnan(self.lnlikes)):
            logging.info('twoF values contain nan')
        idxs = np.isfinite(self.lnlikes)
        jmax = np.nanargmax(self.lnlikes[idxs])
945
        maxlogl = self.lnlikes[jmax]
946
        d = OrderedDict()
947

948
949
        if self.BSGL:
            if hasattr(self, 'search') is False:
Gregory Ashton's avatar
Gregory Ashton committed
950
                self.initiate_search_object()
951
952
953
954
955
956
957
            p = self.samples[jmax]
            self.search.BSGL = False
            maxtwoF = self.logl(p, self.search)
            self.search.BSGL = self.BSGL
        else:
            maxtwoF = maxlogl

Gregory Ashton's avatar
Gregory Ashton committed
958
        repeats = []
959
        for i, k in enumerate(self.theta_keys):
Gregory Ashton's avatar
Gregory Ashton committed
960
961
962
963
964
965
966
967
968
969
            if k in d and k not in repeats:
                d[k+'_0'] = d[k]  # relabel the old key
                d.pop(k)
                repeats.append(k)
            if k in repeats:
                k = k + '_0'
                count = 1
                while k in d:
                    k = k.replace('_{}'.format(count-1), '_{}'.format(count))
                    count += 1
970
971
972
973
974
            d[k] = self.samples[jmax][i]
        return d, maxtwoF

    def get_median_stds(self):
        """ Returns a dict of the median and std of all production samples """
975
        d = OrderedDict()
Gregory Ashton's avatar
Gregory Ashton committed
976
        repeats = []
977
        for s, k in zip(self.samples.T, self.theta_keys):
Gregory Ashton's avatar
Gregory Ashton committed
978
979
980
981
982
983
984
985
986
987
988
989
990
            if k in d and k not in repeats:
                d[k+'_0'] = d[k]  # relabel the old key
                d[k+'_0_std'] = d[k+'_std']
                d.pop(k)
                d.pop(k+'_std')
                repeats.append(k)
            if k in repeats:
                k = k + '_0'
                count = 1
                while k in d:
                    k = k.replace('_{}'.format(count-1), '_{}'.format(count))
                    count += 1

991
992
993
994
            d[k] = np.median(s)
            d[k+'_std'] = np.std(s)
        return d

995
996
997
998
999
1000
    def check_if_samples_are_railing(self, threshold=0.01):
        return_flag = False
        for s, k in zip(self.samples.T, self.theta_keys):
            prior = self.theta_prior[k]
            if prior['type'] == 'unif':
                prior_range = prior['upper'] - prior['lower']
For faster browsing, not all history is shown. View entire blame