-
Notifications
You must be signed in to change notification settings - Fork 0
/
average_green.py
234 lines (200 loc) · 8.38 KB
/
average_green.py
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
import os
import sys
from socket import gethostname
from uuid import uuid1
from mpi4py import MPI
from numpy import *
from functions import getDensityFromGmat, rotate_all, irotate
from share_fun import interp_root, divideTasks
def averageGreen(delta0, mu0, w, SelfEnergy, parms, Nd, Ntot, tuneup, extra):
N_LAYERS = int(parms['N_LAYERS'])
FLAVORS = int(parms['FLAVORS'])
SPINS = int(parms['SPINS'])
rot_mat = extra['rot_mat']
parallel = int(parms.get('KINT_PARALLEL', 2))
# calculate intersite Coulomb energy here
Vc = zeros(N_LAYERS, dtype=float)
# convert self energy to the C++ form
SelfEnergy_rot = array([irotate(SelfEnergy[s], rot_mat)
for s in range(SPINS)])
SE = array([array([s.flatten() for s in SelfEnergy_rot[n]])
for n in range(SPINS)])
v_delta = array([])
ddelta = 0.
delta_step = 1.
v_nd = array([])
dmu = 0.
mu_step = 0.5
tol = 0.003
firsttime = True
initial_Gasymp = extra['G_asymp_coefs'] if 'G_asymp_coefs' in extra.keys()\
else None
starting_error = 0.
# Delta loop
while True:
delta = delta0 + ddelta
if initial_Gasymp is not None:
extra['G_asymp_coefs'][:N_LAYERS*FLAVORS] = initial_Gasymp[:N_LAYERS*FLAVORS] - ddelta
v_mu = array([])
v_n = array([])
# mu loop
while True:
mu = mu0 + dmu
if initial_Gasymp is not None:
extra['G_asymp_coefs'] = initial_Gasymp - dmu
Gavg = integrate(w, delta, mu, SE, parms, extra, parallel)
Gavg_diag = array([[diag(Gavg[s, n]) for n in range(size(Gavg,1))]
for s in range(SPINS)])
nf = getDensityFromGmat(Gavg_diag, float(parms['BETA']), extra)
my_ntot = sum(nf) if SPINS == 2 else 2*sum(nf)
print " adjust mu: %.5f %.5f %.5f"%(mu, dmu, my_ntot)
if firsttime:
starting_error = abs(Ntot - my_ntot)/N_LAYERS
Gavg0 = Gavg.copy()
firsttime = False
if Ntot < 0 or abs(Ntot - my_ntot)/N_LAYERS < tol or not tuneup:
break
v_mu = r_[v_mu, dmu]
v_n = r_[v_n, my_ntot]
if v_n.min() < Ntot and v_n.max() > Ntot:
dmu = interp_root(v_mu, v_n, Ntot)
else:
dmu += (1. if my_ntot < Ntot else -1.)*mu_step
my_nd = sum(nf[:, :N_LAYERS*FLAVORS])
if tuneup:
print ('adjust double counting: %.5f %.5f '
'%.5f %.5f')%(delta, ddelta, my_nd, my_nd/N_LAYERS)
if Nd < 0 or abs(Nd - my_nd)/N_LAYERS < tol or not tuneup:
break
v_delta = r_[v_delta, ddelta]
v_nd = r_[v_nd, my_nd]
if v_nd.min() < Nd and v_nd.max() > Nd:
ddelta = interp_root(v_delta, v_nd, Nd)
else:
ddelta += (1. if my_nd < Nd else -1.)*delta_step
# adjusted Gavg with mu_new = mu_0 + N*dmu and
# delta_new = delta_0 + N*ddelta;
N = float(parms.get('TUNEUP_FACTOR', 1))
if N != 1. and (ddelta != 0. or dmu != 0.) and starting_error < 50*tol:
mu = mu0 + N*dmu
delta = delta0 + N*ddelta
Gavg = integrate(w, delta, mu, SE, parms, extra, parallel)
print ('TUNEUP_FACTOR = %d final adjustment: mu = %.4f, dmu = %.4f, '
'delta = %.4f, ddelta = %.4f')%(N, mu, N*dmu, delta, N*ddelta)
Gavg = array([rotate_all(Gavg[s], rot_mat) for s in range(SPINS)])
Gavg0 = array([rotate_all(Gavg0[s], rot_mat, need_extra = True)
for s in range(SPINS)])
if initial_Gasymp is not None:
extra['G_asymp_coefs'] = initial_Gasymp
return Gavg, Gavg0, delta, mu, Vc
def integrate(w, DELTA, MU, SelfEnergy, parms, extra, parallel=0):
"""
Main function for k-integration
Input argument: parallel
0: no parallel computation (nthreads = 1)
1: OpenMP computation (nthreads = -1, use OMP_NUM_THREADS)
2: MPI computation (but no OpenMP, nthreads = 1)
"""
data = {
'w' : w,
'DELTA' : DELTA,
'MU' : MU,
'SelfEnergy': SelfEnergy,
'Hf' : float(parms.get('H', 0)),
'N_LAYERS' : int(parms['N_LAYERS']),
'FLAVORS' : int(parms['FLAVORS']),
'SPINS' : int(parms['SPINS']),
'NORB' : int(parms['NORB']),
'nthreads' : 1,
'integrate_mod': parms.get('INTEGRATE_MOD', 'integrate'),
'extra' : extra,
}
# this should be changed depending on queuing system
if parallel < 2 or parms['np'] == 1:
if parallel == 1:
data['nthreads'] = -1
# print 'average_green.integrate: OpenMP parallelization'
# print 'average_green.integrate: no MPI parallelization'
return run_task(**data)
# else: print 'average_green.integrate: MPI parallelization'
# prepare for spawning
nprocs = parms['np'] - 1
myinfo = MPI.Info.Create()
for job_hostfile in ('LSB_DJOB_HOSTFILE', 'PBS_NODEFILE'):
if job_hostfile in os.environ:
myinfo.Set("hostfile", os.environ[job_hostfile])
print 'Found MPI hostfile: %s'%job_hostfile
running_script = __file__
comm = MPI.COMM_SELF.Spawn(sys.executable, args=[running_script, 'child'],
maxprocs=nprocs, info=myinfo)
# prepare data
w = data['w']
se = data['SelfEnergy']
del data['w']
del data['SelfEnergy']
# broadcast data
comm.bcast(data, root=MPI.ROOT)
ntasks, displs = divideTasks(nprocs+1, len(w)) # also count the parent
# this is data for parent process
data['w'] = w[displs[0]:displs[0] + ntasks[0]]
data['SelfEnergy'] = se[:, displs[0]:displs[0]+ntasks[0]]
ntasks = ntasks[1:]
displs = displs[1:]
for n in range(nprocs):
comm.send(w[displs[n]:displs[n]+ntasks[n]], dest=n, tag=100*n)
comm.send(se[:, displs[n]:displs[n]+ntasks[n]], dest=n, tag=100*n+1)
# run task from parent process
Gout = run_task(**data)
# collect data
for n in range(nprocs):
results = comm.recv(source=n, tag=100*n+3)
Gout = r_['1', Gout, results]
comm.Disconnect()
return Gout
def run_child_task():
comm = MPI.Comm.Get_parent()
rank = comm.Get_rank()
# print 'child task rank %d/%d at %s'%(rank, comm.Get_size(), gethostname())
data = comm.bcast(None, root=0)
data['w'] = comm.recv(source=0, tag=100*rank)
data['SelfEnergy'] = comm.recv(source=0, tag=100*rank+1)
out = run_task(**data)
comm.send(out, dest=0, tag=100*rank+3)
comm.Disconnect()
# task run by both parent and children
def run_task(w, DELTA, MU, SelfEnergy, Hf, N_LAYERS, FLAVORS, SPINS, NORB,
nthreads, integrate_mod, extra):
# print >> sys.stderr, gethostname(), len(w);
int_module = __import__(integrate_mod, fromlist=[])
nbin = len(w)
bp, wf = extra['GaussianData']
if 'HR' in extra:
Gavg = array([int_module.calc_Gavg(w, DELTA, MU, SelfEnergy[n].copy(),
extra['HR'], extra['R'],
Hf*(-1)**n, bp, wf,
nthreads).reshape(nbin, NORB, NORB)
for n in range(SPINS)])
# swap the Gavg to the format of my code
swap_vec = zeros((2, N_LAYERS*FLAVORS), dtype=int)
for L in range(N_LAYERS):
for f in range(FLAVORS):
swap_vec[:,f*N_LAYERS+L] = array([f*N_LAYERS+L, L*FLAVORS+f])
for s in range(SPINS):
for n in range(len(Gavg[s])):
Gavg[s, n,:,swap_vec[0]] = Gavg[s, n, :, swap_vec[1]]
Gavg[s, n,swap_vec[0],:] = Gavg[s, n, swap_vec[1], :]
else:
Gavg = array([int_module.calc_Gavg(w, DELTA, MU, SelfEnergy[n].copy(),
extra['tight_binding_parms'],
Hf*(-1)**n, bp, wf,
nthreads).reshape(nbin, NORB, NORB)
for n in range(SPINS)])
return Gavg
# main part just for child only
if __name__ == '__main__':
if sys.argv[1] == 'child':
run_child_task();
else:
print >> sys.stderr, ('This is a child process for MPI integrate. '
'Improper running.')
sys.exit()