-
Notifications
You must be signed in to change notification settings - Fork 18
/
tiGadgets.py
171 lines (137 loc) · 4.74 KB
/
tiGadgets.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
"""some gadgets for mathmatical usages"""
import taichi as ti
@ti.kernel
def c_equals_a_minus_b(c: ti.template(), a: ti.template(), b: ti.template()):
"""c = a - b"""
for I in ti.grouped(c):
c[I] = a[I] - b[I]
@ti.kernel
def a_equals_b_plus_c_mul_d(a:ti.template(), b:ti.template(), c: float, d: ti.template()):
"""a = b + c * d"""
for I in ti.grouped(a):
a[I] = b[I] + c * d[I]
@ti.kernel
def field_abs_max(f: ti.template()) -> float:
"""get the maximum absolute value of a scaler field"""
ans = 0.
for I in ti.grouped(f):
ti.atomic_max(ans, ti.abs(f[I]))
return ans
@ti.kernel
def field_norm(f: ti.template()) -> float:
"""get modified Euclidean norm of the scaler field, (the modified 2nd norm) """
ans = 0.
for I in ti.grouped(f):
ans += f[I] ** 2
N = 1
for i in ti.static(range(len(f.shape))):
N *= f.shape[i]
return (ans / N) ** 0.5
@ti.kernel
def field_max(f: ti.template()) -> float:
"""get the max value of a scaler field"""
ans = -float("inf")
for I in ti.grouped(f):
ti.atomic_max(ans, f[I])
return ans
@ti.kernel
def vectorField_max(f: ti.template()) -> float:
"""get the max value of a vector field"""
ans = -float("inf")
for I in ti.grouped(f):
ti.atomic_max(ans, f[I].max())
return ans
@ti.kernel
def field_min(f: ti.template()) -> float:
"""get the min value of a scaler field"""
ans = float("inf")
for I in ti.grouped(f):
ti.atomic_min(ans, f[I])
return ans
@ti.kernel
def field_multiply(field: ti.template(), num: float):
for i in field:
field[i] *= num
@ti.kernel
def field_add(field: ti.template(), num: float):
for i in field:
field[i] = field[i] + num # do not use +=, beacuse that is atomic add which could lose precision
@ti.kernel
def field_addVec(field: ti.template(), vec: ti.template()):
for i in field:
field[i] = field[i] + vec # do not use +=, beacuse that is atomic add which could lose precision
@ti.func
def sorted_tiVec(arr):
"""for small vector, naive bubble sort is acceptable"""
for i in ti.static(range(1, arr.n)):
for j in ti.static(range(0, arr.n - i)):
if arr[j] > arr[j+1]:
arr[j], arr[j + 1] = arr[j + 1], arr[j]
return arr
@ti.func
def get_index_ti(arr, val) -> int:
"""get the index of val in vector arr"""
index = -1
for i in ti.static(range(arr.n)):
if arr[i] == val:
index = i
return index
@ti.kernel
def scalerField_from_matrixField(f1: ti.template(), f2: ti.template(), i: int, j: int):
"""fill the scaler field with index [i, j] of a matrix field"""
for I in ti.grouped(f1):
f1[I] = f2[I][i, j]
def fraction_reduction(a: int, b: int):
"""fraction a/b is reduced by x/y
https://blog.csdn.net/Cosmos53/article/details/116330862 """
x, y = a, b
while b > 0:
a, b = b, a % b
return x // a, y // a
def relative_error(a, b):
"""
get the relative error between a and b
"""
maxVal = max(abs(a), abs(b))
if maxVal > 1.e-9:
return abs(a - b) / maxVal # return relative error
else:
return abs(a - b) # return absolute error if a and b are almost 0
@ti.func
def vec_mul_voigtMtrx(vec, mtrx): # vector (line of a matrix) multiply voigt matrix
"""
dot product of vector and voigt matrix
for 2-dimension
matrix = [
[mtrx[0], mtrx[2]],
[mtrx[2], mtrx[1]]
]
thus, vec * matrix = [
vec[0] * mtrx[0, :] + vec[1] * mtrx[2, :],
vec[0] * mtrx[2, :] + vec[1] * mtrx[1, :],
]
for 3-dimension
matrix = [
[mtrx[0], mtrx[3], mtrx[4]],
[mtrx[3], mtrx[1], mtrx[5]],
[mtrx[4], mtrx[5], mtrx[2]],
]
thus, vec * matrix = [
vec[0] * mtrx[0, :] + vec[1] * mtrx[3, :] + vec[2] * mtrx[4, :],
vec[0] * mtrx[3, :] + vec[1] * mtrx[1, :] + vec[2] * mtrx[5, :],
vec[0] * mtrx[4, :] + vec[1] * mtrx[5, :] + vec[2] * mtrx[2, :],
]
"""
ans = ti.Matrix.zero(ti.f64, vec.n, mtrx.m) #([[0. for _ in range(mtrx.m)] for _ in range(vec.m)])
if vec.n == 2:
ans[0, :] = vec[0] * mtrx[0, :] + vec[1] * mtrx[2, :]
ans[1, :] = vec[0] * mtrx[2, :] + vec[1] * mtrx[1, :]
elif vec.n == 3:
ans[0, :] = vec[0] * mtrx[0, :] + vec[1] * mtrx[3, :] + vec[2] * mtrx[4, :]
ans[1, :] = vec[0] * mtrx[3, :] + vec[1] * mtrx[1, :] + vec[2] * mtrx[5, :]
ans[2, :] = vec[0] * mtrx[4, :] + vec[1] * mtrx[5, :] + vec[2] * mtrx[2, :]
return ans
if __name__ == "__main__":
while True:
a, b = map(int, input("a, b = ").split(","))
print("x, y =", fraction_reduction(a, b))