在SciPy 0.8.0中对rv\u continuous子类化

Question

我想使用“双高斯”（即两个正态分布的归一化和）对一些数据进行拟合。使用SciPy 0.18.1，以下代码正常工作：

import numpy as np
from scipy.stats import rv_continuous

class DoubleGauss(rv_continuous):
    """Double Gaussian distribution"""

    def _pdf(self, x, mu1, sigma1, mu2, sigma2, w):
        g1 = np.exp(-0.5*((x-mu1)/sigma1)**2) / np.sqrt(2.0*np.pi*sigma1**2)
        g2 = np.exp(-0.5*((x-mu2)/sigma2)**2) / np.sqrt(2.0*np.pi*sigma2**2)
        return w * g1 + (1.0-w) * g2

dg = DoubleGauss(name='DG')
print dg.rvs(mu1=0.001, sigma1=2.0, mu2=0.05, sigma2=4.0, w=0.8)

（这当然是一个最小的例子。最后，我还将重新定义\u cdf for speed和\u argcheck，以允许mu参数的非零值。）

因为我需要使用其他库读取要安装的数据，所以我必须在只有scipy0.8.0可用的环境中运行代码。但是，运行相同的代码会导致以下错误：

  File "test.py", line 13, in <module>
    print dg.rvs(mu1=0.01, sigma1=2.0, mu2=0.05, sigma2=4.0, w=0.8)
  File "/cvmfs/cms.cern.ch/slc6_amd64_gcc491/external/py2-scipy/0.8.0-cms/lib/python2.7/site-packages/scipy/stats/distributions.py", line 523, in rvs
    vals = self._rvs(*args)
  File "/cvmfs/cms.cern.ch/slc6_amd64_gcc491/external/py2-scipy/0.8.0-cms/lib/python2.7/site-packages/scipy/stats/distributions.py", line 848, in _rvs
    Y = self._ppf(U,*args)
  File "/cvmfs/cms.cern.ch/slc6_amd64_gcc491/external/py2-scipy/0.8.0-cms/lib/python2.7/site-packages/scipy/stats/distributions.py", line 861, in _ppf
    return self.vecfunc(q,*args)
  File "/cvmfs/cms.cern.ch/slc6_amd64_gcc491/external/py2-numpy/1.6.1-cms/lib/python2.7/site-packages/numpy/lib/function_base.py", line 1862, in __call__
    theout = self.thefunc(*newargs)
  File "/cvmfs/cms.cern.ch/slc6_amd64_gcc491/external/py2-scipy/0.8.0-cms/lib/python2.7/site-packages/scipy/stats/distributions.py", line 817, in _ppf_single_call
    return optimize.brentq(self._ppf_to_solve, self.xa, self.xb, args=(q,)+args, xtol=self.xtol)
  File "/cvmfs/cms.cern.ch/slc6_amd64_gcc491/external/py2-scipy/0.8.0-cms/lib/python2.7/site-packages/scipy/optimize/zeros.py", line 362, in brentq
    r = _zeros._brentq(f,a,b,xtol,maxiter,args,full_output,disp)
  File "/cvmfs/cms.cern.ch/slc6_amd64_gcc491/external/py2-scipy/0.8.0-cms/lib/python2.7/site-packages/scipy/stats/distributions.py", line 814, in _ppf_to_solve
    return apply(self.cdf, (x, )+args)-q
  File "/cvmfs/cms.cern.ch/slc6_amd64_gcc491/external/py2-scipy/0.8.0-cms/lib/python2.7/site-packages/scipy/stats/distributions.py", line 952, in cdf
    place(output,cond,self._cdf(*goodargs))
  File "/cvmfs/cms.cern.ch/slc6_amd64_gcc491/external/py2-scipy/0.8.0-cms/lib/python2.7/site-packages/scipy/stats/distributions.py", line 855, in _cdf
    return self.veccdf(x,*args)
  File "/cvmfs/cms.cern.ch/slc6_amd64_gcc491/external/py2-numpy/1.6.1-cms/lib/python2.7/site-packages/numpy/lib/function_base.py", line 1862, in __call__
    theout = self.thefunc(*newargs)
  File "/cvmfs/cms.cern.ch/slc6_amd64_gcc491/external/py2-scipy/0.8.0-cms/lib/python2.7/site-packages/scipy/stats/distributions.py", line 852, in _cdf_single_call
    return scipy.integrate.quad(self._pdf, self.a, x, args=args)[0]
  File "/cvmfs/cms.cern.ch/slc6_amd64_gcc491/external/py2-scipy/0.8.0-cms/lib/python2.7/site-packages/scipy/integrate/quadpack.py", line 247, in quad
    retval = _quad(func,a,b,args,full_output,epsabs,epsrel,limit,points)
  File "/cvmfs/cms.cern.ch/slc6_amd64_gcc491/external/py2-scipy/0.8.0-cms/lib/python2.7/site-packages/scipy/integrate/quadpack.py", line 313, in _quad
    return _quadpack._qagie(func,bound,infbounds,args,full_output,epsabs,epsrel,limit)
TypeError: _pdf() takes exactly 7 arguments (2 given)

在文档中，我发现关于如何使用rv\u continuous的描述没有差异。如何修复代码，使其与scipy0.8.0一起工作？你知道吗

Answer 1

在0.13之前，pdf等人不接受关键字参数。尝试将它们作为位置参数，而不是关键字（命名）参数。你知道吗

自v0.8以来，还有其他一些改进。我认为你需要在构造器中指定形状，还有一个叫做xa和xb的东西用于ppf，你可能需要在https://github.com/scipy/scipy/blob/v0.8.0/scipy/stats/distributions.py检查源代码