/
SEO_simulator_tf.py
120 lines (92 loc) · 3.33 KB
/
SEO_simulator_tf.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
from qubiter.SEO_simulator import *
import tensorflow as tf
class SEO_simulator_tf(SEO_simulator):
"""
TF= TensorFlow
This class is a child of SEO_simulator. The class replaces numpy methods
by TensorFlow (2.0, Eager) methods when those methods perform the state
vector evolution. This allows those tf methods to be taped, and the tape
to be run in reverse to calculate gradients.
At the beginning (in the constructor), the class converts the arrays in
`self.cur_st_vec_dict` from numpy to tf.
Certain "interface" methods of the parent class were written assuming
that the arrays of `self.cur_st_vec_dict` are numpy. Such methods call
convert_tensors_to_numpy() at their beginning to convert the tf arrays
to numpy. They also call convert_tensors_to_tf() at their end to
convert the numpy tensors back to tf, the way they were upon starting
the method.
Attributes
----------
"""
def __init__(self, file_prefix, num_qbits, init_st_vec=None, **kwargs):
"""
Constructor
Parameters
----------
file_prefix : str
num_qbits : int
init_st_vec : StateVec
kwargs : dict
keyword args of SEO_simulator
Returns
-------
"""
assert tf.executing_eagerly()
SEO_simulator.__init__(self, file_prefix, num_qbits,
init_st_vec=init_st_vec, **kwargs)
def do_more_init_before_reading(self):
"""
Overrides method in parent SEO_simulator.
Returns
-------
None
"""
self.convert_tensors_to_tf()
SEO_simulator.transpose = tf.transpose
SEO_simulator.tensordot = tf.tensordot
SEO_simulator.reshape = tf.reshape
self.use_tf = True
self.lib = 'tf'
def convert_tensors_to_tf(self):
"""
This method converts the arrays in `cur_st_vec_dict` from numpy to
tensorflow.
Returns
-------
None
"""
for br_key, st_vec in self.cur_st_vec_dict.items():
if isinstance(st_vec.arr, np.ndarray):
st_vec.arr = tf.convert_to_tensor(st_vec.arr)
def convert_tensors_to_numpy(self):
"""
This method converts the arrays in `cur_st_vec_dict` from tensorflow
to numpy.
Returns
-------
None
"""
for br_key, st_vec in self.cur_st_vec_dict.items():
if isinstance(st_vec.arr, tf.Tensor):
st_vec.arr = st_vec.arr.numpy()
if __name__ == "__main__":
def main():
# tf.enable_eager_execution()
# use test = 0 if want to run all tests at once.
test = 0
# test = 3
if test in [0, 1]:
# test on circuit for a quantum fourier transform
# (no loops, no internal measurements)
sim = SEO_simulator_tf('sim_test1', 6,
verbose=True)
if test in [0, 2]:
# test embedded loops
sim = SEO_simulator_tf('sim_test2', 4,
verbose=True)
if test in [0, 3]:
# test MEAS branching. Each kind 2 measurement doubles number of
# branches
sim = SEO_simulator_tf('sim_test3', 4,
verbose=True)
main()