Question

我正在尝试开始使用scipy，并且在使用“线性”以外的其他类型（我尝试'零'和'立方'）时似乎无法从interpolation.interp1d获得示例。

我在Google上搜索时找不到有同样问题的人，所以我觉得这对我来说很愚蠢。

我使用scipy 0.11在python 2.73上使用OSX 10.8

代码不起作用：

from scipy import interpolate
import numpy as np

x = np.arange(0, 10)
y = np.exp(-x/3.0)
f = interpolate.interp1d(x, y, kind="zero")
xnew = np.arange(0,9, 0.1)
ynew = f(xnew)   # use interpolation function returned by `interp1d`
plt.plot(x, y, 'o', xnew, ynew, '-')
plt.show()

IndexError                                Traceback (most recent call last)
<ipython-input-1-23bb96a1589b> in <module>()
  4 f = interpolate.interp1d(x, y, kind="zero")
  5 xnew = np.arange(0,9, 0.1)
----> 6 ynew = f(xnew)   # use interpolation function returned by `interp1d`
  7 plt.plot(x, y, 'o', xnew, ynew, '-')
  8 plt.show()

/usr/local/lib/python2.7/site-packages/scipy/interpolate/interpolate.pyc in __call__(self, x_new)
394         out_of_bounds = self._check_bounds(x_new)
395 
--> 396         y_new = self._call(x_new)
397 
398         # Rotate the values of y_new back so that they correspond to the

/usr/local/lib/python2.7/site-packages/scipy/interpolate/interpolate.pyc in _call_spline(self, x_new)
370     def _call_spline(self, x_new):
371         x_new =np.asarray(x_new)
--> 372         result = spleval(self._spline,x_new.ravel())
373         return result.reshape(x_new.shape+result.shape[1:])
374 

/usr/local/lib/python2.7/site-packages/scipy/interpolate/interpolate.pyc in spleval((xj, cvals, k), xnew, deriv)
833             res[sl].imag = _fitpack._bspleval(xx,xj,cvals.imag[sl],k,deriv)
834         else:
--> 835             res[sl] = _fitpack._bspleval(xx,xj,cvals[sl],k,deriv)
836     res.shape = oldshape + sh
837     return res

IndexError: too many indices

在这里启动调试控制台时，我可以将其缩小到cvals [sl]，导致错误

sl = (slice(None, None, None), 0) # <-- I don't really get the slice part here...
cvals = array([ 1.        ,  0.71653131,  0.51341712,  0.36787944,  0.26359714,
    0.1888756 ,  0.13533528,  0.09697197,  0.06948345])

有人可以重现这个或者我的机器出了什么问题吗？

Answer 1

看来，在OSX上遵循scipy的安装说明时，我安装了numpy的非发布版本而没有意识到它。

用pip重新安装numpy，scipy和matplotlib解决了这个问题。

非常感谢unutbu指出问题的根源。

带有interp1d IndexError的scipy样条插值

1 个答案: