error_estimates_base.py 13.7 KB
Newer Older
Mohammad-Yasin Arif's avatar
Mohammad-Yasin Arif committed
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
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
216
217
218
219
220
221
222
223
224
225
226
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
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
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
from numpy import zeros, int32, unique, genfromtxt, sort, argsort, arange, \
        append, array
from ase import Atom
from ase.db import connect

def string2symbols(s):
    """Convert string to list of chemical symbols."""
    n = len(s)

    if n == 0:
        return []
    
    c = s[0]
    
    if c.isdigit():
        i = 1
        while i < n and s[i].isdigit():
            i += 1
        return int(s[:i]) * string2symbols(s[i:])

    if c == '(':
        p = 0
        for i, c in enumerate(s):
            if c == '(':
                p += 1
            elif c == ')':
                p -= 1
                if p == 0:
                    break
        j = i + 1
        while j < n and s[j].isdigit():
            j += 1
        if j > i + 1:
            m = int(s[i + 1:j])
        else:
            m = 1
        return m * string2symbols(s[1:i]) + string2symbols(s[j:])

    if c.isupper():
        i = 1
        if 1 < n and s[1].islower():
            i += 1
        j = i
        while j < n and s[j].isdigit():
            j += 1
        if j > i:
            m = int(s[i:j])
        else:
            m = 1
        return m * [s[:i]] + string2symbols(s[j:])
    else:
        raise ValueError

def get_data(con, name_dict, Z, code, category, keys, recommended_VASP,
             name_dict_monos_gpaw, name_dict_bins_gpaw):
    """Function for obtaining the data from the databases for each code
    Parameters:

    con: The database

    name_dict: dict
        The dictionary for el. solid/binaries

    Z: array like
        atomic numbers for el. solids/binaries

    code: str
        DFT-code

    category: str
        code specific category string

    keys: code specific parameters
  """
    N_monos = len(name_dict)
    data = zeros(N_monos)
    if code == 'VASP':
        rows = con.select(selection=[('category=' + category),
                                     ('precision=' + keys[0]),
                                     ('k_point_density=' + str(keys[1])),
                                     ('functional=' + keys[2])])
        # For VASP we have to select the recommended Pseudo-Potential:
        for row in rows:
            if category.find('binaries') == 0:
                A = unique(row.numbers)[0] - 1
                B = unique(row.numbers)[1] - 1
                rec_POT = ('POTCAR' +
                           recommended_VASP[A, 1].replace('POTCAR', '') +
                           recommended_VASP[B, 1].replace('POTCAR', ''))
                data_POT = row.potcar.replace('_h', '')
                rec_POT = rec_POT.replace('_3', '')
                rec_POT = rec_POT.replace('_2', '')
            else:
                rec_POT = recommended_VASP[Z[name_dict[row.name]]-1, 1]
                data_POT = row.potcar
            if data_POT == rec_POT:
                data[name_dict[row.name]] = row.total_energy

    elif code == 'FHI-aims':
        rows = con.select(selection=[('category=' + category),
                                     ('basis_set=' + keys[0]),
                                     ('k_point_density=' + str(keys[1])),
                                     ('functional=' + keys[2]),
                                     ('tiers=' + keys[3]),
                                     ('relativistic_treatment=' + keys[4])])
        for row in rows:
            data[name_dict[row.name]] = row.total_energy

    elif code == 'exciting':
        rows = con.select(selection=[('category=' + category),
                                     ('total_precision','=',float(keys[0])),
                                     ('k_point_density','=',float(keys[1])),
                                     ('xc_functional=' + keys[2])])
        for row in rows:
            data[name_dict[row.name]] = row.total_energy

    elif code == 'GPAW':
        rows = con.select(selection=[('category=' + category),
                                     ('ecut=' + str(keys[0])),
                                     ('k_point_density=' + str(keys[1]))])
        # Mapping from formula to system name
        for row in rows:
            if category.find('binaries') == 0:
                data[name_dict[name_dict_bins_gpaw[row.formula]]] = row.energy
            else:
                data[name_dict[name_dict_monos_gpaw[row.formula]]] = row.energy
    return data

def get_rows(con, name_dict, Z, code, category, keys, recommended_VASP,
             name_dict_monos_gpaw, name_dict_bins_gpaw):
    """Function for obtaining the rows from the databases for each code
    Parameters:

    con: The database

    name_dict: dict
        The dictionary for el. solid/binaries

    Z: array like
        atomic numbers for el. solids/binaries

    code: str
        DFT-code

    category: str
        code specific category string

    keys: code specific parameters
  """
    if code == 'VASP':
        rows = con.select(selection=[('category=' + category),
                                     ('precision=' + keys[0]),
                                     ('k_point_density=' + str(keys[1])),
                                     ('functional=' + keys[2])])

    elif code == 'FHI-aims':
        rows = con.select(selection=[('category=' + category),
                                     ('basis_set=' + keys[0]),
                                     ('k_point_density=' + str(keys[1])),
                                     ('functional=' + keys[2]),
                                     ('tiers=' + keys[3]),
                                     ('relativistic_treatment=' + keys[4])])

    elif code == 'exciting':
        rows = con.select(selection=[('category=' + category),
                                     ('total_precision','=',float(keys[0])),
                                     ('k_point_density','=',float(keys[1])),
                                     ('xc_functional=' + keys[2])])

    elif code == 'GPAW':
        rows = con.select(selection=[('category=' + category),
                                     ('ecut=' + str(keys[0])),
                                     ('k_point_density=' + str(keys[1]))])
    return rows


def get_keys(code, prec, kpt, xc, tiers, rel):
    if code == 'VASP':
        keys = [prec, str(kpt), xc, tiers]
        ref_keys = [xc]
    elif code == 'FHI-aims':
        keys = [prec, str(kpt), xc, tiers, rel]
        ref_keys = [xc, rel]
    elif code == 'exciting':
        keys = [prec, str(kpt), xc]
        ref_keys = [xc]
    elif code == 'GPAW':
        keys = [prec, str(kpt)]
        ref_keys = [xc]
    return keys, ref_keys


def get_xy(Z, N, data_ref, data):
    plot_x = sort(Z)
    plot_y = ((data_ref-data)/N)[argsort(Z)]
    nonzero=((data!=0)*(data_ref!=0))[argsort(Z)]
    plot_x = plot_x[nonzero]
    plot_y = plot_y[nonzero]
    #print(plot_x[abs(plot_y)>0.001])
    return plot_x, abs(plot_y)


def get_xy_predict(N, data_ref, data):
    plot_x = ((data) / N)
    plot_y = data_ref / N
    nonzero = (plot_y != 0) * (plot_x != 0)
    plot_x = plot_x[nonzero]
    plot_y = plot_y[nonzero]
    #print(plot_x[abs(plot_y)>0.001])
    return abs(plot_x), abs(plot_y)


def get_binary_error_from_solids(error, binaries_to_monos_min, N_bins_min,
                                 binaries_to_monos_max, N_bins_max, pred):
    error_A = error[binaries_to_monos_min]
    error_B = error[binaries_to_monos_max]
    N_A = N_bins_min
    N_B = N_bins_max
    N_AB = N_A + N_B

    if pred == '1':
        pred_error=(error_A * N_A + error_B * N_B)
        rel_pred_error = pred_error / N_AB
    elif pred == '2':
        pred_error = (error_A * N_A + error_B * N_B)
        rel_pred_error = pred_error / N_AB
    return pred_error #, rel_pred_error


def do_plot(fig, ax, Z, N, data_ref, data, lab):
    plot_x = sort(Z)
    plot_y = ((data_ref - data) / N)[argsort(Z)]
    nonzero = plot_y != 0
    plot_x = plot_x[nonzero]
    plot_y = plot_y[nonzero]
    #print(plot_x[abs(plot_y)>0.001])
    ax.semilogy(plot_x, abs(plot_y), 'o', label=lab)
    ax.legend(numpoints=1, loc=4, fontsize=8)
    fig.canvas.draw_idle()


def do_plot_predict(fig, ax, N, data_ref, data, lab):
    plot_x = ((data) / N)
    plot_y = data_ref / N
    nonzero = (plot_y != 0) * (plot_x != 0)
    plot_x = plot_x[nonzero]
    plot_y = plot_y[nonzero]
    #print(plot_x[abs(plot_y)>0.001])
    ax.loglog(abs(plot_x), abs(plot_y), 'o', label=lab)
    ax.plot(ax.get_xlim(), ax.get_xlim(), '-k')
    ax.legend(numpoints=1, loc=4, fontsize=8)
    fig.canvas.draw_idle()


def get_xy(Z, N, data_ref, data):
    plot_x = sort(Z)
    plot_y = ((data_ref-data)/N)[argsort(Z)]
    nonzero = plot_y != 0
    plot_x = plot_x#[nonzero]
    plot_y = plot_y#[nonzero]
    #print(plot_x[abs(plot_y)>0.001])
    return plot_x, abs(plot_y)


def get_xy_predict(N, data_ref, data):
    plot_x = ((data) / N)
    plot_y = data_ref / N
    nonzero = (plot_y != 0) * (plot_x != 0)
    plot_x = plot_x[nonzero]
    plot_y = plot_y[nonzero]
    #print(plot_x[abs(plot_y)>0.001])
    return abs(plot_x), abs(plot_y)

def get_xy_Ecoh(Z, data_monos,data_bins,zeroinds,binaries_to_monos_min,binaries_to_monos_max,N_bins_min,N_bins_max):
    error_A = data_monos[binaries_to_monos_min]
    error_B = data_monos[binaries_to_monos_max]
    N_A = N_bins_min
    N_B = N_bins_max
    N_AB = N_A + N_B
    plot_x=Z[zeroinds]
    plot_y=((data_bins)[zeroinds]-(error_A * N_A + error_B * N_B)[zeroinds])/N_AB[zeroinds]
    return plot_x, plot_y
  
def get_mono_ind(formula,name_dict_monos):
    symbols = string2symbols(formula) 
    ind = zeros(0,dtype='int32')
    N_f = len(symbols)
    notinset=False
    for s in symbols:
        at = Atom(s)
        if at.number < 10:
            try:
                ind = append(ind, name_dict_monos['0'+str(at.number)+'_'+s])
            except KeyError:
                print("Element "+s+" is not in the set of elementary solids.")
                notinset=True
                break
        else:
            try:
                ind = append(ind, name_dict_monos[str(at.number)+'_'+s])
            except KeyError:
                print("Element "+s+" is not in the set of elementary solids.")
                notinset=True
                break
    return N_f, ind, notinset

def intermediate_delta_energy_exciting(elementname, intermediate_prec, con, name_dict_monos_exciting, keys, data_ref, N_mono):
    import numpy as np
    import sys
    import matplotlib.pyplot as mpl
    import scipy
    import scipy.stats

    available_precs=np.asarray([30,40,50,60,70,80]) # this list needs to be sorted
    natoms=N_mono[name_dict_monos_exciting[elementname]]
    total_energy_ref=data_ref['exciting','monomers',keys[2]][name_dict_monos_exciting[elementname]]
    
    if intermediate_prec < available_precs[0]:
        #sys.exit('intermediate_delta_energy_exciting: precision of one element in a binary is below '+str(available_precs[0]))
        number_of_points=2
        x__considered_precs=available_precs[0:number_of_points]
    elif intermediate_prec > available_precs[-1]:
        #print('CAUTION: extrapolation! Precision of one element in a binary is above '+str(available_precs[-1]))
        number_of_points=3
        number_of_points=-1*number_of_points
        x__considered_precs=available_precs[number_of_points:]
    else:
        for icount in range(1,len(available_precs)):
            if intermediate_prec < available_precs[icount]:
                x__considered_precs=available_precs[[icount-1,icount]]
                break
                
    y__log_error_per_atom=[]
    #for prec in x__considered_precs:
    rows = con.select(selection=[('category=monomers'),
                                        ('total_precision','<=',x__considered_precs[-1]),
                                        ('total_precision','>=',x__considered_precs[0]),
                                        ('k_point_density','=',float(keys[1])),
                                        ('xc_functional=' + keys[2]),
                                        ('element1='+elementname)])
    for row in rows:
        total_energy=row.total_energy        
        y__log_error_per_atom.append(np.log10((total_energy - total_energy_ref)/natoms))

    linregr_results = scipy.stats.linregress(x__considered_precs,y__log_error_per_atom)
    p = np.asarray([linregr_results[0], linregr_results[1]])
    delta_energy_per_atom = np.power(10,np.polyval(p,intermediate_prec))
    # want to plot results of linregress?
    if False:
        # get all monomers data
        monomers_delta_energy_per_atom=[]
        rows = con.select(selection=[('category=monomers'),
                                        ('total_precision','<=',available_precs[-1]),
                                        ('total_precision','>=',available_precs[0]),                                        
                                        ('k_point_density','=',float(keys[1])),
                                        ('xc_functional=' + keys[2]),
                                        ('element1='+elementname)])
        for row in rows:
            total_energy=row.total_energy        
            monomers_delta_energy_per_atom.append((total_energy - total_energy_ref)/natoms)
        # plot
        fig3 = figure(3,(10,10))
        rect = fig3.patch
        rect.set_facecolor('white')
        mpl.semilogy(available_precs, monomers_delta_energy_per_atom,'ko-', ms=7.0, markeredgewidth=1, linewidth=1.5, label="data")
        mpl.semilogy(x__considered_precs, np.power(10,y__log_error_per_atom),'go', ms=11.0, markeredgewidth=1, label="accounted data in linregress")
        mpl.semilogy(intermediate_prec, delta_energy_per_atom,'rx', ms=11.0, markeredgewidth=3, label="result from linregress")
        axes = mpl.gca()
        axes.grid(True)
        mpl.xlabel("Precision %", fontsize = "x-large")
        mpl.ylabel("$\Delta E$ per atom [eV]", fontsize = "x-large")
        mpl.legend(loc="lower left", frameon=True, fontsize = "x-large", numpoints=1)
        mpl.title(elementname)
        mpl.show()
    return delta_energy_per_atom