是否有方法在zfit中扩展预先定义的多项式pdf？

Question

​我试着用扩展的Chebyshev来做一个契合。为此，我使用预定义的zfit.pdf.Chebyshev()。我在这里错过了什么？

import tensorflow as tf
import zfit
from zfit import ztf

obs = zfit.Space('mass', limits=(-1000., 1000.)) 
slope = zfit.Parameter("slope", -0.1, -1., 1., floating = True)
Nbkg = zfit.Parameter("Nbkg", 10, 0., 1000)
bkg_pdf =  Nbkg*zfit.pdf.Chebyshev(obs =obs, coeffs = [slope])

对于所有预定义的多项式(Chebyshev/Legendre/Hermite)，我得到以下错误：

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-3-a86d6be1cc9d> in <module>
      2 slope = zfit.Parameter("slope", -0.1, -1., 1., floating = True)
      3 Nbkg = zfit.Parameter("Nbkg", 10, 0., 1000)
----> 4 bkg_pdf =  Nbkg*zfit.pdf.Chebyshev(obs =obs, coeffs = [slope])
      5 
      6 

~/soft/minconda3/envs/my-analysis-env/lib/python3.7/site-packages/zfit/core/parameter.py in __mul__(self, other)
    289             from . import operations
    290             with suppress(NotImplementedError):
--> 291                 return operations.multiply(self, other)
    292         return super().__mul__(other)
    293 

~/soft/minconda3/envs/my-analysis-env/lib/python3.7/site-packages/zfit/core/operations.py in multiply(object1, object2)
     34             new_object = multiply_param_func(param=object1, func=object2)
     35         elif isinstance(object2, ZfitPDF):
---> 36             new_object = multiply_param_pdf(param=object1, pdf=object2)
     37         else:
     38             assert False, "This code should never be reached due to logical reasons. Mistakes happen..."

~/soft/minconda3/envs/my-analysis-env/lib/python3.7/site-packages/zfit/core/operations.py in multiply_param_pdf(param, pdf)
     86     if pdf.is_extended:
     87         raise AlreadyExtendedPDFError()
---> 88     new_pdf = pdf.create_extended(param, name_addition="_autoextended")
     89     return new_pdf
     90 

~/soft/minconda3/envs/my-analysis-env/lib/python3.7/site-packages/zfit/core/basepdf.py in create_extended(self, yield_, name_addition)
    462         if self.is_extended:
    463             raise AlreadyExtendedPDFError("This PDF is already extended, cannot create an extended one.")
--> 464         new_pdf = self.copy(name=self.name + str(name_addition))
    465         new_pdf._set_yield_inplace(value=yield_)
    466         return new_pdf

~/soft/minconda3/envs/my-analysis-env/lib/python3.7/site-packages/zfit/core/basepdf.py in copy(self, **override_parameters)
    579         #     parameters.update(distribution=self.distribution)
    580         yield_ = parameters.pop('yield', None)
--> 581         new_instance = type(self)(**parameters)
    582         if yield_ is not None:
    583             new_instance._set_yield_inplace(yield_)

TypeError: __init__() got an unexpected keyword argument 'c_0'

我也尝试过zfit.pdf.Chebyshev().create_extended(Nbkg)，为pdf设置了规范化和coeff0参数。

Answer 1

实际上，这是一个在版本> 0.3.6 (当前开发分支)中修复的bug。

您可以通过修复来安装开发分支。

pip install git+https://github.com/zfit/zfit