如何避免此列表中的舍入错误？

Question

这个列表应该有x[50]为零，两边对称，但是它稍微偏离中心，因为我假设是舍入误差。如何修改代码以避免这种情况

谢谢

import numpy as np
L=2*np.pi
s=101
ds=L/(s-1)
svals=np.arange(1,101)
x=[0]
x[0:s]=((svals-1)*ds)-L/2
print(x)

Answer 1

从浮点运算中获得精确的输出是很棘手的。您可以使用np.linspace获得所需的列表：

x = np.linspace(-np.pi, 0, num=51)
x = np.concatenate([x, np.linspace(x[-1] - x[-2], np.pi, num=50)])
print(x)

输出：

[-3.14159265 -3.0787608  -3.01592895 -2.95309709 -2.89026524 -2.82743339
 -2.76460154 -2.70176968 -2.63893783 -2.57610598 -2.51327412 -2.45044227
 -2.38761042 -2.32477856 -2.26194671 -2.19911486 -2.136283   -2.07345115
 -2.0106193  -1.94778745 -1.88495559 -1.82212374 -1.75929189 -1.69646003
 -1.63362818 -1.57079633 -1.50796447 -1.44513262 -1.38230077 -1.31946891
 -1.25663706 -1.19380521 -1.13097336 -1.0681415  -1.00530965 -0.9424778
 -0.87964594 -0.81681409 -0.75398224 -0.69115038 -0.62831853 -0.56548668
 -0.50265482 -0.43982297 -0.37699112 -0.31415927 -0.25132741 -0.18849556
 -0.12566371 -0.06283185  0.          0.06283185  0.12566371  0.18849556
  0.25132741  0.31415927  0.37699112  0.43982297  0.50265482  0.56548668
  0.62831853  0.69115038  0.75398224  0.81681409  0.87964594  0.9424778
  1.00530965  1.0681415   1.13097336  1.19380521  1.25663706  1.31946891
  1.38230077  1.44513262  1.50796447  1.57079633  1.63362818  1.69646003
  1.75929189  1.82212374  1.88495559  1.94778745  2.0106193   2.07345115
  2.136283    2.19911486  2.26194671  2.32477856  2.38761042  2.45044227
  2.51327412  2.57610598  2.63893783  2.70176968  2.76460154  2.82743339
  2.89026524  2.95309709  3.01592895  3.0787608   3.14159265]

它分两步进行，以避免在linspace范围跨越0时出现数字错误

x = np.linspace(-np.pi, np.pi, 101)

那么x[50]处的值是4.440892098500626e-16，而不是像预期的那样x[50]是0