附加到数组开头的nil值

Question

我有一个foo.txt文件，其中包含以下行：

A 10 20 30
B 40 50 60
C 70 80 90

我将这些值转换为弧度并将其转换为旋转矩阵，然后将旋转矩阵与另一个旋转矩阵相乘，并将输出存储在新变量中，然后保存到新的文本文件中

import transforms3d.euler as t3d
import numpy as np
import pandas as pd


# Suppress scientific notation
np.set_printoptions(suppress=True)

file="foo.txt"

r1=np.array([[1,0,0],[0,1,0],[0,0,1]])
r2=np.array([[0.5,-0.30902,-0.80902],[0.30902,-0.80902,0.5],[-0.80902,-0.5,-0.30902]])
r3=np.array([[0.5,0.30902,-0.80902],[-0.30902,-0.80902,-0.5],[-0.80902,0.5,-0.30902]])
R=np.array([r1,r2,r3])

source_array=pd.read_csv(file, sep=" ", header=None)

name_array=source_array.iloc[:,0].to_numpy()
number_array=source_array.iloc[:,1:4].to_numpy()


fullstack=np.empty((1,4))
for i in range(number_array.shape[0]):
    inMat=t3d.euler2mat(np.radians(number_array[i][0]),\
                         np.radians(number_array[i][1]),\
                         np.radians(number_array[i][2]),'rzyz')
    for j in range(R.shape[0]):
        expMat=np.dot(inMat,R[j])
        expDeg=np.degrees(t3d.mat2euler(expMat))
        name=np.repeat(name_array[i],1)
        nameandDeg=np.hstack((name,expDeg)).reshape(1,4)
        fullstack=np.append(fullstack,nameandDeg,axis=0)
        np.savetxt('foo3.txt',fullstack,fmt='%s')

预期产量为

A 10.314104815618196 17.229396562958897 41.56670396614096
A -86.15202640927517 58.81289030540899 101.65925914885956
A 119.58912709295329 73.9812604867084 6.339258927995449
B 45.90468727333836 22.521012118111006 109.63942512488691
B -59.08770198050942 30.43292092265276 177.0466653835958
B -123.45325226252277 66.42599043668275 175.86346970407573
C 80.0 3.455054858914999e-15 160.0
C -63.59253616776351 -9.42984810562947 -140.45406233407002
C -127.56777089055139 26.411096630846163 -143.9360404337779

但是以当前形式运行代码后，在数组的开头会出现一行额外的代码

**0.0** **5e-324** **1e-323** **1.5e-323**
A 10.314104815618196 17.229396562958897 41.56670396614096
A -86.15202640927517 58.81289030540899 101.65925914885956
A 119.58912709295329 73.9812604867084 6.339258927995449
B 45.90468727333836 22.521012118111006 109.63942512488691
B -59.08770198050942 30.43292092265276 177.0466653835958
B -123.45325226252277 66.42599043668275 175.86346970407573
C 80.0 3.455054858914999e-15 160.0
C -63.59253616776351 -9.42984810562947 -140.45406233407002
C -127.56777089055139 26.411096630846163 -143.9360404337779

为什么会出现这条线？我怎样才能摆脱它

还有一个问题，是否有一种方法可以格式化输出，使所有包含浮动的列都表示为%0.3f

提前谢谢

Answer 1

获取行0.0 5e-324 1e-323 1.5e-323的原因是因为行fullstack=np.empty((1,4))。而在数值计算中np.empty(someshape)会产生有效等于0的条目，将它们保存到文件still writes the in memory numerical value，或：0.0 5e-324 1e-323 1.5e-323。只需将np.savetxt()行更改为

np.savetxt('foo3.txt',fullstack[1:],fmt='%s')

或者重新思考为什么首先需要创建一个空数组

还请注意，如果在j循环的每次迭代中都要附加到fullstack。您不需要在j循环的每次迭代中保存完整的堆栈，只需在j循环完成后保存一次

Answer 2

np.empty((1,4))创建给定形状的新数组without initializing entries。这意味着fullstack数组不是空的，甚至在循环中开始向其追加nameandDeg之前就包含值。当您将fullstack写入foo3.txt文件时，您看到的第一行是这些值