在嵌套循环中附加数据框

Question

我有以下代码根据 2 个变量（bh 和频率）在函数中进行一些计算。当我以固定值 bh 运行没有循环的代码时，我得到了一个正确的输出和保存在 .csv 文件中的数据框。：

    bh   frequency                Re                  Im
0  1e-05         1  5.86848609615851  -0.999374346845734
1  1e-05        11  4.34298196390882  -0.994875549720418
2  1e-05        21  3.93236459069042   -0.99112086235206
3  1e-05        31  3.68545733552675  -0.987695572513367
4  1e-05        41  3.50849758486341  -0.984487932588323

但是，当我在 bh 上的循环中编码时，我想在 bh 值和频率列表上循环，我得到与以前相同的输出，这意味着它不会循环。是否有人有解决方案来修改数据框或将 eaxh bh 循环输出保存在新的 .csv 中，以便稍后绘制数据。

from mpmath import *
import numpy as np
import cmath
import math
import pandas as pd

mp.dps = 15; mp.pretty = True
a = mpf(0.25)
b = mpf(0.25)
z = mpf(0.75)
frequency = np.arange(1, 50, 10)  # frequency range
bh = np.arange(10e-6, 30e-6, 10e-6) #10e-6 # width
print(bh)
D = 1e-6 #7.8e-4  # diffusivity
gamma = 0.5772 # Euler constant
v = []
w =[]
i = []
def q(frequency):
  for i in bh:
    # for f in frequency:
      omega = (((i ** 2) * 2 * math.pi * frequency) / D)  # depends on bh and frequency
      u = ((-j/(math.pi * omega))*meijerg([[1, 3/2], []], [[1, 1], [0.5, 0]], j*omega))
      v = np.real(u)
      w = np.imag(u)
      return i, frequency, v, w
#transpose arrays
T = np.vectorize(q)
print(T(frequency))
df = np.array(T(frequency)).T
print(df)
# create DataFrame
df1 = pd.DataFrame(data=df, columns=['bh', 'frequency','Re', 'Im'])
print(df1)
#save in .csv
df1.to_csv('C:\\Users\\calculations\\T.csv')

Answer 1

我不完全确定我是否理解您的问题域，但您似乎正在返回循环的第一次迭代。通过在循环中使用 return，您基本上可以提前终止 for 循环。这应该在循环之外。此外，您不会将值保存在数组 v、w 和 i 中。您正在覆盖变量。

我做了一些修改（根据您的问题域可能不正确），但它应该可以完成您想要完成的工作。

from mpmath import *
import numpy as np
import cmath
import math
import pandas as pd

mp.dps = 15; mp.pretty = True
a = mpf(0.25)
b = mpf(0.25)
z = mpf(0.75)
frequencies = np.arange(1, 50, 10)  # frequency range
bh = np.arange(10e-6, 30e-6, 10e-6) #10e-6 # width
print(bh)
D = 1e-6 #7.8e-4  # diffusivity
gamma = 0.5772 # Euler constant
v = []
w = []
i = []
bhs = []
freqs = []
def q(frequencies):
  for frequency in frequencies:
    for i in bh:
      # for f in frequency:
        omega = (((i ** 2) * 2 * math.pi * frequency) / D)  # depends on bh and frequency
        u = ((-j/(math.pi * omega))*meijerg([[1, 3/2], []], [[1, 1], [0.5, 0]], j*omega))
        v.append(np.real(u))
        w.append(np.imag(u))
        bhs.append(i)
        freqs.append(frequency)
  return bhs, freqs, v, w

data = np.array(q(frequencies)).T
# create DataFrame
df1 = pd.DataFrame(data=data, columns=['bh', 'frequency','Re', 'Im'])
df1

输出：

    bh  frequency             Re                 Im
0   1e-05   1   5.86848609615851    -0.999374346845734
1   2e-05   1   4.98625732244539    -0.99786698700645
2   1e-05   11  4.34298196390882    -0.994875549720418
3   2e-05   11  3.46384911041305    -0.983559190454865
4   1e-05   21  3.93236459069042    -0.99112086235206
5   2e-05   21  3.05626898509369    -0.972212391507732
6   1e-05   31  3.68545733552675    -0.987695572513367
7   2e-05   31  2.81234917403506    -0.962167989599812
8   1e-05   41  3.50849758486341    -0.984487932588323
9   2e-05   41  2.63833200578647    -0.952979213441469
10  1e-05   1   5.86848609615851    -0.999374346845734
11  2e-05   1   4.98625732244539    -0.99786698700645
12  1e-05   11  4.34298196390882    -0.994875549720418
13  2e-05   11  3.46384911041305    -0.983559190454865
14  1e-05   21  3.93236459069042    -0.99112086235206
15  2e-05   21  3.05626898509369    -0.972212391507732
16  1e-05   31  3.68545733552675    -0.987695572513367
17  2e-05   31  2.81234917403506    -0.962167989599812
18  1e-05   41  3.50849758486341    -0.984487932588323
19  2e-05   41  2.63833200578647    -0.952979213441469
20  1e-05   1   5.86848609615851    -0.999374346845734
21  2e-05   1   4.98625732244539    -0.99786698700645
22  1e-05   11  4.34298196390882    -0.994875549720418
23  2e-05   11  3.46384911041305    -0.983559190454865
24  1e-05   21  3.93236459069042    -0.99112086235206
25  2e-05   21  3.05626898509369    -0.972212391507732
26  1e-05   31  3.68545733552675    -0.987695572513367
27  2e-05   31  2.81234917403506    -0.962167989599812
28  1e-05   41  3.50849758486341    -0.984487932588323
29  2e-05   41  2.63833200578647    -0.952979213441469

Answer 2

我建议 (1) 提前生成 bh 和 frequency 的笛卡尔积，以及 (2) 仅矢量化您真正需要的部分，因为已知 np.vectorization 成本高昂（即 meijerg()不是矢量化函数）。笛卡尔积可以由pd.MultiIndex.from_product 完成（见this answer）。

# run your code until gamma = 0.5772

# Cartesian product of input variables
idx = pd.MultiIndex.from_product([bh, frequency], names=["bh", "frequency"])
df = pd.DataFrame(index=idx).reset_index()

# Omega is vectorized naturally.
omega = (df["bh"].values**2 * df["frequency"].values) * (2 * math.pi / D)

# vectorize meijerg() only, so other operations won't interrupt with this
def f_u(omega_elem):
    return (-j/(math.pi * omega_elem)) * meijerg([[1, 3/2], []], [[1, 1], [0.5, 0]], j*omega_elem)

f_u_vec = np.vectorize(f_u, otypes=[np.complex128]) # output complex

u = f_u_vec(omega)  # np.complex128
df["Re"] = np.real(u)
df["Im"] = np.imag(u)

# output (please make sure your arange was set correctly)
df
Out[35]: 
        bh  frequency        Re        Im
0  0.00001          1  5.868486 -0.999374
1  0.00001         11  4.342982 -0.994876
2  0.00001         21  3.932365 -0.991121
3  0.00001         31  3.685457 -0.987696
4  0.00001         41  3.508498 -0.984488
5  0.00002          1  4.986257 -0.997867
6  0.00002         11  3.463849 -0.983559
7  0.00002         21  3.056269 -0.972212
8  0.00002         31  2.812349 -0.962168
9  0.00002         41  2.638332 -0.952979

如果你想保存单独的 csv 文件，你可以这样做：

for bh_elem in bh:
    fname = f"bh={bh_elem:.4e}.csv"
    df_save = df[(df["bh"]==bh_elem)]
    df_save.to_csv(fname)

注意在 pandas 1.1.3 和 python 3.7、debian 10 64 位上测试