-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathPhaseOneSolver.py
154 lines (131 loc) · 4.26 KB
/
PhaseOneSolver.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
import numpy as np
from NewtonSolver import *
from FunctionManager import FunctionManagerPhase1, FunctionManagerSOCPPhase1
class PhaseOneSolver:
def __init__(
self,
C=None,
d=None,
lower_bound=0,
upper_bound=None,
x0=None,
max_outer_iters=50,
max_inner_iters=20,
epsilon=1e-8,
inner_epsilon=1e-5,
linear_solve_method="cholesky",
max_cg_iters=50,
alpha=0.2,
beta=0.6,
mu=15,
t0=1,
suppress_print=False,
use_gpu=False,
track_loss=False,
n=None,
tol=0.1,
socp=False,
socp_params=None,
use_psd_condition=False,
update_slacks_every=0
):
# all attributes for LP
self.C = C
self.d = d
self.lb = lower_bound
self.ub = upper_bound
self.x = x0
self.n = n
self.max_outer_iters = max_outer_iters
self.max_inner_iters = max_inner_iters
self.epsilon = epsilon
self.inner_epsilon = inner_epsilon
self.linear_solve_method = linear_solve_method
self.max_cg_iters = max_cg_iters
self.alpha = alpha
self.beta = beta
self.mu = mu
self.suppress_print = suppress_print
self.use_gpu = use_gpu
self.track_loss = track_loss
self.tol = tol
self.t0 = t0
self.update_slacks_every = update_slacks_every
if not socp:
self.phase1_fm = FunctionManagerPhase1(
C=self.C,
d=self.d,
x0=self.x,
lower_bound=self.lb,
upper_bound=self.ub,
t=self.t0,
use_gpu=self.use_gpu,
n=self.n,
suppress_print=self.suppress_print,
)
else:
self.phase1_fm = FunctionManagerSOCPPhase1(
*socp_params,
x0=self.x,
lower_bound=self.lb,
upper_bound=self.ub,
t=self.t0,
use_gpu=self.use_gpu,
n=self.n,
suppress_print=self.suppress_print,
)
if self.use_gpu:
self.x = cp.append(self.x, self.phase1_fm.s)
else:
self.x = np.append(self.x, self.phase1_fm.s)
self.phase1_ns = NewtonSolverCholesky(
C=self.C,
d=self.d,
function_manager=self.phase1_fm,
lower_bound=self.lb,
upper_bound=self.ub,
max_iters=self.max_inner_iters,
epsilon=self.inner_epsilon,
suppress_print=self.suppress_print,
max_cg_iters=self.max_cg_iters,
alpha=self.alpha,
beta=self.beta,
mu=self.mu,
use_gpu=self.use_gpu,
track_loss=self.track_loss,
phase1_flag=True,
phase1_tol=self.tol,
use_psd_condition=use_psd_condition,
update_slacks_every=self.update_slacks_every
)
def solve(self, x0=None):
if x0 is not None:
self.phase1_fm.update_x(x0)
t = self.t0
self.outer_iters = 0
self.inner_iters = []
for iter in range(self.max_outer_iters):
if not self.suppress_print:
print(f"Current slack: {self.phase1_fm.s}")
self.x, _, numiters_t, _, success_flag = self.phase1_ns.solve(self.x, t)
self.outer_iters += 1
self.inner_iters.append(numiters_t)
obj_val = self.phase1_fm.objective(self.x)
if obj_val < -self.tol:
break
# if self.track_loss:
# objective_vals.append(obj_val)
# if obj_val < best_obj:
# best_obj = obj_val
# best_x = x.copy()
# else:
# break
if numiters_t >= self.max_inner_iters:
if not self.suppress_print:
print(
f"Reached max Newton steps during {iter}th centering step (t={t}) of phase 1"
)
# increment t for next outer iteration
t = min(t * self.mu, (self.n + 1.0) / self.epsilon)
self.phase1_fm.update_t(t)
return self.x[:-1], obj_val