gofee.py 16.3 KB
Newer Older
Malthe Kjær Bisbo's avatar
Malthe Kjær Bisbo committed
1
import numpy as np
Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
2
3
import pickle
from os.path import isfile
Malthe Kjær Bisbo's avatar
Malthe Kjær Bisbo committed
4

5
6
7
8
9
10
from surrogate.gpr import GPR
from population import population

from ase import Atoms
from ase.io import read, write, Trajectory
from ase.calculators.singlepoint import SinglePointCalculator
11
from ase.calculators.calculator import FileIOCalculator
12
13

from bfgslinesearch_zlim import BFGSLineSearch_zlim
14
from bfgslinesearch_constrained import BFGSLineSearch_constrained
15
16
from ase.ga.relax_attaches import VariansBreak
from parallel_utils import split, parallel_function_eval
17
18
19
20
21

from candidate_operations.candidate_generation import OperationSelector
from candidate_operations.basic_mutations import RattleMutation


22
23
24
25
26
27
28
from mpi4py import MPI
world = MPI.COMM_WORLD

import traceback
import sys


Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
29
def relax(structure, calc, Fmax=0.05, steps_max=200, dmax_cov=None):
30
31
32
33
34
35
36
    a = structure.copy()
    # Set calculator 
    a.set_calculator(calc)
    pos_init = a.get_positions()

    # Catch if linesearch fails
    try:
37
38
39
40
        dyn = BFGSLineSearch_constrained(a,
                                         logfile=None,
                                         pos_init=pos_init,
                                         dmax_cov=dmax_cov)
Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
41
        dyn.run(fmax = Fmax, steps = steps_max)
42
43
44
45
46
47
48
49
    except Exception as err:
        print('Error in surrogate-relaxation:', err, flush=True)
        traceback.print_exc()
        traceback.print_exc(file=sys.stderr)
    return a


class GOFEE():
50
51
    """GOFEE global structure search method.
        
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
    structures: Atoms-object, list of Atoms-objects or None.
    In initial structures from which to start the sesarch.
    If None, the startgenerator must be supplied.
    If less than Ninit structures is supplied, the remaining
    ones are generated using the startgenerator or by rattling
    the supplied structures, depending on wether the
    startgenerator is supplied.

    gpr: The Gaussian Process Regression model used as the
    surrogate model for the Potential energy surface.
    
    startgenerator: 

    trajectory:

    kappa: 
    
Malthe Kjær Bisbo's avatar
Malthe Kjær Bisbo committed
69
70
71
72
    N_relax_final_pop: int or None.
    If not None the best 'N_relax_final_pop' structures in the
    population is relaxed after the specified number of search
    iterations is reached.
73
    """
74
75
76
77
78
79
80
    def __init__(self, structures=None,
                 calc=None,
                 gpr=None,
                 startgenerator=None,
                 candidate_generator=None,
                 trajectory=None,
                 kappa=2,
Malthe Kjær Bisbo's avatar
Malthe Kjær Bisbo committed
81
                 max_steps=200,
Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
82
                 Ninit=10,
83
84
85
86
                 dmax_cov=3.5,
                 Ncandidates=30,
                 population_size=5,
                 dualpoint=True,
Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
87
                 min_certainty=0.7,
Malthe Kjær Bisbo's avatar
Malthe Kjær Bisbo committed
88
89
                 restart=None,
                 N_relax_final_pop=None):
90
        
91
92
93
94
        if structures is None:
            assert startgenerator is not None
            self.structures = None
        else:
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
            if isinstance(structures, Atoms):
                self.structures = [structures]
            elif isinstance(structures, list):
                assert isinstance(structures[0], Atoms)
                self.structures = structures
            elif isinstance(structures, str):
                self.structures = read(structures, index=':')
        
        if calc is not None:
            self.calc = calc
        else:
            assert structures is not None
            calc = structures[0].get_calculator()
            assert calc is not None and not isinstance(calc, SinglePointCalculator)
            print('Using calculator from supplied structure(s)')
            self.calc = calc

        if gpr is not None:
            self.gpr = gpr
        else:
            self.gpr = GPR()

117
        if startgenerator is None:
118
119
            assert structures is not None
            self.startgenerator = None
120
121
        else:
            self.startgenerator = startgenerator
122

123
124
125
126
127
128
129
130
131
        if startgenerator is not None:
            self.n_to_optimize = len(self.startgenerator.stoichiometry)
        else:
            self.n_to_optimize = len(self.structures[0])
            for constraint in self.structures[0].constraints:
                if isinstance(constraint, FixAtoms):
                    indices_fixed = constraint.get_indices()
                    self.n_to_optimize -= len(indices_fixed)
                    break
Malthe Kjær Bisbo's avatar
Malthe Kjær Bisbo committed
132
        
133
134
135
        if candidate_generator is not None:
            self.candidate_generator = candidate_generator
        else:
136
137
138
139
            rattle = RattleMutation(self.n_to_optimize,
                                    Nrattle=3,
                                    rattle_range=4)
            self.candidate_generator = OperationSelector([1.0],[rattle])
140
141
142
143
144
145
146
147
148

        # Initialize population
        self.population = population(population_size=population_size, gpr=self.gpr, similarity2equal=0.9999)

        # Define parallel communication
        self.comm = world.Dup()  # Important to avoid mpi-problems from call to ase.parallel in BFGS
        self.master = self.comm.rank == 0

        self.kappa = kappa
Malthe Kjær Bisbo's avatar
Malthe Kjær Bisbo committed
149
        self.max_steps = max_steps
Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
150
        self.Ninit = Ninit
151
152
153
154
        self.dmax_cov = dmax_cov
        self.Ncandidates = Ncandidates
        self.dualpoint = dualpoint
        self.min_certainty = min_certainty
Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
155
        self.restart = restart
Malthe Kjær Bisbo's avatar
Malthe Kjær Bisbo committed
156
157
158
159
160
161
162
163
        if N_relax_final_pop is None:
            self.N_relax_final_pop = 0
        else:
            try:
                self.N_relax_final_pop = int(N_relax_final_pop)
            except TypeError:
                raise
                
Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
164
165
166
167
168
169
170
171
172
        
        if isinstance(trajectory, str):
            self.trajectory = Trajectory(filename=trajectory, mode='a', master=self.master)
            if self.restart:
                self.traj_name = trajectory
        elif isinstance(trajectory, Trajectory):
            self.trajectory = trajectory
        else:
            assert trajectory is None
173

Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
174
175
176
177
178
179
180
181
        if restart is None or not isfile(restart):
            self.initialize()
        else:
            self.read()
            self.comm.barrier()

    def initialize(self):
        self.steps = 0
182
183

    def get_initial_structures(self):
184
185
186
187
188
189
190
191
192
        """Method to prepare the initial structures for the search.
        
        The method makes sure that there are atleast self.Ninit
        initial structures.
        These structures are first of all the potentially supplied
        structures. If more structures are required, these are
        generated using self.startgenerator (if supplied), otherwise
        they are generated by heavily rattling the supplied structures.
        """
Malthe Kjær Bisbo's avatar
Malthe Kjær Bisbo committed
193
        
194
195
196
        # Collect potentially supplied structures and evaluate
        # energies and forces if not present.
        structures_init = []
Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
197
        if self.structures is not None:
198
            for a in self.structures:
199
200
201
202
203
204
205
                calc = a.get_calculator()
                if isinstance(calc, SinglePointCalculator):
                    if 'energy' in calc.results and 'forces' in calc.results:
                        # Add without evaluating.
                        structures_init.append(a)
                        self.write(a)
                        continue
Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
206
                a = self.evaluate(a)
207
                structures_init.append(a)
Malthe Kjær Bisbo's avatar
Malthe Kjær Bisbo committed
208
        
209
        Nremaining = self.Ninit - len(structures_init)
Malthe Kjær Bisbo's avatar
Malthe Kjær Bisbo committed
210
        
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
        if Nremaining > 0 and self.startgenerator is None:
            # Initialize rattle-mutation for all atoms.
            rattle = RattleMutation(self.n_to_optimize,
                                    Nrattle=self.n_to_optimize,
                                    rattle_range=2)

        # Generation of remaining initial-structures (up to self.Ninit).
        for i in range(Nremaining):
            if self.startgenerator is not None:
                a = self.startgenerator.get_new_candidate()
            else:
                # Perform two times rattle of all atoms.
                a0 = structures_init[i % len(structures_init)]
                a = rattle.get_new_candidate([a])
                a = rattle.get_new_candidate([a])
            a = self.evaluate(a)
            structures_init.append(a)

        # Potentially do a few relaxation steps.
        ### missing code ###
            
        self.gpr.memory.save_data(structures_init)
        self.population.add(structures_init)
234
235
                
    def run(self):
Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
236
        self.get_initial_structures()
237

Malthe Kjær Bisbo's avatar
Malthe Kjær Bisbo committed
238
        while self.steps < self.max_steps:
Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
239
            self.print_master('steps:', self.steps)
240
241
242
243
244
            self.train_surrogate()
            self.update_population()
            relaxed_candidates = self.get_surrogate_relaxed_candidates()

            kappa = self.kappa
Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
            a_add = []
            for _ in range(5):
                try:
                    anew = self.select_with_acquisition(relaxed_candidates, kappa)
                    self.print_master('aq done')
                    anew = self.evaluate(anew)
                    a_add.append(anew)
                    self.print_master('sp done')
                    if self.dualpoint:
                        adp = self.get_dualpoint(anew)
                        adp = self.evaluate(adp)
                        a_add.append(adp)
                    self.print_master('dp done')
                    break
                except Exception as err:
                    kappa /=2
                    if self.master:
                        traceback.print_exc(file=sys.stderr)
            self.gpr.memory.save_data(a_add)
Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
264
            self.steps += 1
265
266

            # Add structure to population
Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
267
            self.population.add(a_add)
Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
268
269
270
271

            # Save search state
            self.dump((self.steps, self.population, np.random.get_state()))
            
272
273
            if self.master:
                print('anew pred:', anew.info['key_value_pairs']['Epred'], anew.info['key_value_pairs']['Epred_std'])
274
                print('E_true:', [a.get_potential_energy() for a in a_add])
275
                print('pop:', [a.get_potential_energy() for a in self.population.pop])
Malthe Kjær Bisbo's avatar
Malthe Kjær Bisbo committed
276
277

        self.relax_final_population()
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
            
    def get_dualpoint(self, a, lmax=0.10, Fmax_flat=5):
        """Returns dual-point structure, i.e. the original structure
        perturbed slightly along the forces.
        
        lmax: The atom with the largest force will be displaced by
        this distance
        
        Fmax_flat: maximum atomic displacement. is increased linearely
        with force until Fmax = Fmax_flat, after which it remains
        constant as lmax.
        """
        F = a.get_forces()
        a_dp = a.copy()

        # Calculate and set new positions
        Fmax = np.sqrt((F**2).sum(axis=1).max())
        pos_displace = lmax * F*min(1/Fmax_flat, 1/Fmax)
        pos_dp = a.positions + pos_displace
        a_dp.set_positions(pos_dp)
        return a_dp

    def print_master(self, *args):
        self.comm.barrier()
        if self.master:
            print(*args, flush=True)

    def generate_candidate(self):
        Ntrials = 5
        for i_trial in range(Ntrials):
            parents = self.population.get_structure_pair()
309
            a_mutated = self.candidate_generator.get_new_candidate(parents)
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
            # break trial loop if successful
            if a_mutated is not None:
                break
            # If no success in max number of trials
        if a_mutated is None:
            a_mutated = parents[0].copy()
        return a_mutated

    def get_surrogate_relaxed_candidates(self):
        Njobs = self.Ncandidates
        task_split = split(Njobs, self.comm.size)
        def func1():
            return [self.generate_candidate() for i in task_split[self.comm.rank]]
        candidates = parallel_function_eval(self.comm, func1)
        
        def func2():
            return [self.surrogate_relaxation(candidates[i], Fmax=0.1, steps=200, kappa=self.kappa)
                    for i in task_split[self.comm.rank]]
        relaxed_candidates = parallel_function_eval(self.comm, func2)
        relaxed_candidates = self.certainty_filter(relaxed_candidates)

        #if (self.NsearchIter % self.Nuse_pop_as_candidates) == 0:
Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
332
        if self.steps % 3 == 0:
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
            relaxed_candidates = self.population.pop_MLrelaxed + relaxed_candidates
        
        if self.master:
            Epred = np.array([a.info['key_value_pairs']['Epred'] for a in relaxed_candidates])
            Epred_std = np.array([a.info['key_value_pairs']['Epred_std'] for a in relaxed_candidates])
            fitness = Epred - self.kappa*Epred_std
            print(np.c_[Epred, Epred_std, fitness])
        return relaxed_candidates
        
    def certainty_filter(self, structures):
        certainty = np.array([a.info['key_value_pairs']['Epred_std']
                              for a in structures]) / np.sqrt(self.gpr.K0)
        min_certainty = self.min_certainty
        for _ in range(5):
            filt = certainty < min_certainty
            if np.sum(filt.astype(int)) > 0:
                structures = [structures[i] for i in range(len(filt)) if filt[i]]
                break
            else:
                min_certainty = min_certainty + (1-min_certainty)/2
        return structures

    def update_population(self):
        Njobs = len(self.population.pop)
        task_split = split(Njobs, self.comm.size)
        func = lambda: [self.surrogate_relaxation(self.population.pop[i],
                                                  Fmax=0.001, steps=200, kappa=None)
                        for i in task_split[self.comm.rank]]
        self.population.pop_MLrelaxed = parallel_function_eval(self.comm, func)
        if self.master:
            print('ML-relaxed pop forces:\n',
                  [(a.get_forces()**2).sum(axis=1).max()**0.5 for a in self.population.pop_MLrelaxed])

    def train_surrogate(self):
        # Train
Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
368
        if self.steps < 50 or (self.steps % 10) == 0:
369
370
371
372
373
374
375
376
377
            self.gpr.optimize_hyperparameters(comm=self.comm)
        else:
            self.gpr.train()
        if self.master:
            print('kernel:', list(np.exp(self.gpr.kernel.theta)))
            print('lml:', self.gpr.lml)

    def surrogate_relaxation(self, a, Fmax=0.1, steps=200, kappa=None):
        calc = self.gpr.get_calculator(kappa)
Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
378
        a_relaxed = relax(a, calc, dmax_cov=self.dmax_cov, Fmax=Fmax, steps_max=steps)
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396

        # Evaluate uncertainty
        E, Estd = self.gpr.predict_energy(a_relaxed, eval_std=True)
        a_relaxed.info['key_value_pairs']['Epred'] = E
        a_relaxed.info['key_value_pairs']['Epred_std'] = Estd

        return a_relaxed

    def select_with_acquisition(self, structures, kappa):
        Epred = np.array([a.info['key_value_pairs']['Epred']
                          for a in structures])
        Epred_std = np.array([a.info['key_value_pairs']['Epred_std']
                              for a in structures])
        acquisition = Epred - kappa*Epred_std
        index_select = np.argmin(acquisition)
        return structures[index_select]

    def evaluate(self, a):
Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
397
398
        a = self.comm.bcast(a, root=0)
        
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
        if isinstance(self.calc, FileIOCalculator):
            if self.master:
                a.set_calculator(self.calc)
                E = a.get_potential_energy()
                F = a.get_forces()
                results = {'energy': E, 'forces': F}
                calc_sp = SinglePointCalculator(a, **results)
                a.set_calculator(calc_sp)
            a = self.comm.bcast(a, root=0)
        else:
            a.set_calculator(self.calc)
            E = a.get_potential_energy()
            F = a.get_forces()
            results = {'energy': E, 'forces': F}
            calc_sp = SinglePointCalculator(a, **results)
            a.set_calculator(calc_sp)
415
416
417
418
419

        self.write(a)

        return a

Malthe Kjær Bisbo's avatar
Malthe Kjær Bisbo committed
420
421
422
423
424
425
426
427
    def relax_final_population(self):
        if self.N_relax_final_pop > 0:
            relaxed_population_trajectory = Trajectory(filename='relaxed_final_population.traj',
                                                       mode='a', master=self.master)
            for a in self.population.pop[:self.N_relax_final_pop]:
                a = relax(a, self.calc, Fmax=0.05, steps_max=20)
                relaxed_population_trajectory.write(a)

428
429
430
431
    def write(self, a):
        if self.trajectory is not None:
            self.trajectory.write(a)

Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
432
433
434
    def dump(self, data):
        if self.comm.rank == 0 and self.restart is not None:
            pickle.dump(data, open(self.restart, "wb"), protocol=2)
435

Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
436
437
438
439
440
    def read(self):
        self.steps, self.population, random_state = pickle.load(open(self.restart, "rb"))
        np.random.set_state(random_state)
        training_structures = read(self.traj_name, index=':')
        self.gpr.memory.save_data(training_structures)