gofee.py 12.2 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
11
12
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

from bfgslinesearch_zlim import BFGSLineSearch_zlim
13
from bfgslinesearch_constrained import BFGSLineSearch_constrained
14
15
16
17
18
19
20
21
22
from ase.ga.relax_attaches import VariansBreak
from parallel_utils import split, parallel_function_eval
from mpi4py import MPI
world = MPI.COMM_WORLD

import traceback
import sys


Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
23
def relax(structure, calc, Fmax=0.05, steps_max=200, dmax_cov=None):
24
25
26
27
28
29
30
    a = structure.copy()
    # Set calculator 
    a.set_calculator(calc)
    pos_init = a.get_positions()

    # Catch if linesearch fails
    try:
31
32
33
34
        dyn = BFGSLineSearch_constrained(a,
                                         logfile=None,
                                         pos_init=pos_init,
                                         dmax_cov=dmax_cov)
Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
35
        dyn.run(fmax = Fmax, steps = steps_max)
36
37
38
39
40
41
42
43
    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():
44
45
46
47
    """GOFEE global structure search method.
        
    structures: 
    """
48
49
50
51
52
53
54
55
    def __init__(self, structures=None,
                 calc=None,
                 gpr=None,
                 startgenerator=None,
                 candidate_generator=None,
                 trajectory=None,
                 kappa=2,
                 Neval=200,
Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
56
                 Ninit=10,
57
58
59
60
                 dmax_cov=3.5,
                 Ncandidates=30,
                 population_size=5,
                 dualpoint=True,
Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
61
62
                 min_certainty=0.7,
                 restart=None):
63
        
64
65
66
67
        if structures is None:
            assert startgenerator is not None
            self.structures = None
        else:
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
            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()

90
        if startgenerator is None:
91
92
            assert structures is not None
            self.startgenerator = None
93
94
        else:
            self.startgenerator = startgenerator
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109

        if candidate_generator is not None:
            self.candidate_generator = candidate_generator
        else:
            assert False

        # 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
        self.Neval = Neval
Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
110
        self.Ninit = Ninit
111
112
113
114
        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
115
116
117
118
119
120
121
122
123
124
        self.restart = restart
        
        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
125

Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
126
127
128
129
130
131
132
133
        if restart is None or not isfile(restart):
            self.initialize()
        else:
            self.read()
            self.comm.barrier()

    def initialize(self):
        self.steps = 0
134
135
136

    def get_initial_structures(self):
        # Initial structures
Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
137
138

        if self.structures is not None:
139
140
141
142
            for a in self.structures:
                Ei = a.get_potential_energy()
                Fi = a.get_forces()
                self.gpr.memory.save_data([a])
Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
143
144
                self.trajectory.write(a)
                self.population.add(a)
Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
145
146
        elif self.startgenerator is not None:
            for i in range(int(np.ceil(self.Ninit/2))):
147
                a = self.startgenerator.get_new_candidate()
Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
148
149
150
151
152
153
154
                a = self.evaluate(a)
                a2 = self.get_dualpoint(a) 
                a2 = self.evaluate(a2)
                self.gpr.memory.save_data([a,a2])
                self.population.add(a2)
        else:
            assert False
155
156
                
    def run(self):
Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
157
        self.get_initial_structures()
158

Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
159
160
        while self.steps < self.Neval:
            self.print_master('steps:', self.steps)
161
162
163
164
165
            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
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
            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
185
            self.steps += 1
186
187

            # Add structure to population
Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
188
            self.population.add(a_add)
Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
189
190
191
192

            # Save search state
            self.dump((self.steps, self.population, np.random.get_state()))
            
193
194
            if self.master:
                print('anew pred:', anew.info['key_value_pairs']['Epred'], anew.info['key_value_pairs']['Epred_std'])
195
                print('E_true:', [a.get_potential_energy() for a in a_add])
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
                print('pop:', [a.get_potential_energy() for a in self.population.pop])
            
    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()
228
            a_mutated = self.candidate_generator.get_new_candidate(parents)
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
            # 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
251
        if self.steps % 3 == 0:
252
253
254
255
256
257
258
259
260
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
286
            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
287
        if self.steps < 50 or (self.steps % 10) == 0:
288
289
290
291
292
293
294
295
296
            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
297
        a_relaxed = relax(a, calc, dmax_cov=self.dmax_cov, Fmax=Fmax, steps_max=steps)
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315

        # 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
316
317
        a = self.comm.bcast(a, root=0)
        
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
        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)

        self.write(a)

        return a

    def write(self, a):
        if self.trajectory is not None:
            self.trajectory.write(a)

Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
334
335
336
    def dump(self, data):
        if self.comm.rank == 0 and self.restart is not None:
            pickle.dump(data, open(self.restart, "wb"), protocol=2)
337

Malthe Kjær Bisbo's avatar
update    
Malthe Kjær Bisbo committed
338
339
340
341
342
    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)