Merge pull request #96 from scikit-hep/eduardo-README

README updated and matplotlib dependency made optional

.travis.yml

@@ -15,7 +15,7 @@ before_script:
 install:
   - pip install Cython
   - pip install numpy
-  - pip install "matplotlib<=3.0.0"
+  - pip install matplotlib
   - pip install iminuit
   - pip install pytest pytest-cov pytest-mpl
   - pip install flake8 pylint

README.rst

@@ -1,19 +1,39 @@
 .. -*- mode: rst -*-
 
-.. image:: https://travis-ci.org/scikit-hep/probfit.png?branch=master
-   :target: https://travis-ci.org/scikit-hep/probfit
+probfit
+=======
 
+.. image:: https://img.shields.io/pypi/v/probfit.svg
+   :target: https://pypi.python.org/pypi/probfit
 
-probfit
--------
+.. image:: https://zenodo.org/badge/DOI/10.5281/zenodo.1477853.svg
+   :target: https://doi.org/10.5281/zenodo.1477853
+
+.. image:: https://travis-ci.org/scikit-hep/probfit.png?branch=master
+   :target: https://travis-ci.org/scikit-hep/probfit
 
 *probfit* is a set of functions that helps you construct a complex fit. It's
 intended to be used with `iminuit <http://iminuit.readthedocs.org/>`_. The
-tool includes Binned/Unbinned Likelihood estimator, :math:`\chi^2` regression,
+tool includes Binned/Unbinned Likelihood estimators, :math:`\chi^2` regression,
 Binned :math:`\chi^2` estimator and Simultaneous fit estimator.
-Various functors for manipulating PDF such as Normalization and
-Convolution(with caching) and various builtin functions
-normally used in B physics is also provided.
+Various functors for manipulating PDFs such as Normalization and
+Convolution (with caching) and various built-in functions
+normally used in B physics are also provided.
+
+Strict dependencies
+-------------------
+
+- `Python <http://docs.python-guide.org/en/latest/starting/installation/>`__ (2.7+, 3.4+)
+- `Numpy <https://scipy.org/install.html>`__
+- `iminuit <http://iminuit.readthedocs.org/>`_
+
+Optional dependencies
+---------------------
+
+- `matplotlib <http://matplotlib.org/>`_ for the plotting functions
+
+Getting started
+---------------
 
 .. code-block:: python
 
@@ -26,14 +46,16 @@ normally used in B physics is also provided.
     minuit.migrad()
     unbinned_likelihood.draw(minuit)
 
+Documentation and Tutorial
+--------------------------
 
-* `MIT <http://opensource.org/licenses/MIT>`_ license (open source)
 * `Documentation <http://probfit.readthedocs.org/>`_
 * The tutorial is an IPython notebook that you can view online
   `here <http://nbviewer.ipython.org/urls/raw.github.com/scikit-hep/probfit/master/tutorial/tutorial.ipynb>`_.
   To run it locally: `cd tutorial; ipython notebook --pylab=inline tutorial.ipynb`.
-* Dependencies:
-   - `iminuit <http://iminuit.readthedocs.org/>`_
-   - `numpy <http://www.numpy.org/>`_
-   - `matplotlib <http://matplotlib.org/>`_ (optional, for plotting)
 * Developing probfit: see the `development page <http://probfit.readthedocs.io/en/latest/development.html>`_
+
+License
+-------
+
+The package is licensed under the `MIT <http://opensource.org/licenses/MIT>`_ license (open source).

probfit/costfunc.pyx

@@ -3,7 +3,6 @@ cimport cython
 import numpy as np
 cimport numpy as np
 from libc.math cimport exp, pow, fabs, log, tgamma, lgamma, sqrt
-from matplotlib import pyplot as plt
 from . import plotting
 from ._libstat cimport compute_nll, compute_chi2_f, compute_bin_chi2_f, \
     csum, compute_bin_lh_f, integrate1d
@@ -101,6 +100,7 @@ cdef class SimultaneousFit:
             :meth:`draw` for arguments.
 
         """
+        from matplotlib import pyplot as plt
         ret = self.draw(m)
         plt.show()
         return ret
@@ -324,6 +324,7 @@ cdef class UnbinnedLH:
             :meth:`draw` for arguments.
 
         """
+        from matplotlib import pyplot as plt
         ret = self.draw(*arg, **kwd)
         plt.show()
         return ret
@@ -575,6 +576,7 @@ cdef class BinnedLH:
             :meth:`draw` for arguments.
 
         """
+        from matplotlib import pyplot as plt
         ret = self.draw(*arg, **kwd)
         plt.show()
         return ret
@@ -679,6 +681,7 @@ cdef class Chi2Regression:
             :meth:`draw` for arguments.
 
         """
+        from matplotlib import pyplot as plt
         ret = self.draw(*arg, **kwd)
         plt.show()
         return ret
@@ -827,6 +830,7 @@ cdef class BinnedChi2:
             :meth:`draw` for arguments.
 
         """
+        from matplotlib import pyplot as plt
         ret = self.draw(*arg, **kwd)
         plt.show()
         return ret

probfit/oneshot.py

@@ -1,7 +1,6 @@
 import itertools as itt
 import collections
 import numpy as np
-from matplotlib import pyplot as plt
 from iminuit import Minuit
 from .py23_compat import range
 from .costfunc import UnbinnedLH, BinnedChi2, BinnedLH
@@ -24,6 +23,7 @@ def fit_uml(f, data, quiet=False, print_level=0, *arg, **kwd):
     minuit.migrad()
     if not minuit.migrad_ok() or not minuit.matrix_accurate():
         if not quiet:
+            from matplotlib import pyplot as plt
             plt.figure()
             uml.show()
             print(minuit.values)
@@ -49,6 +49,7 @@ def fit_binx2(f, data, bins=30, bound=None, print_level=0, quiet=False, *arg, **
     minuit.migrad()
     if not minuit.migrad_ok() or not minuit.matrix_accurate():
         if not quiet:
+            from matplotlib import pyplot as plt
             plt.figure()
             uml.show()
             print(minuit.values)
@@ -83,6 +84,7 @@ def fit_binlh(f, data, bins=30,
     minuit.migrad()
     if not minuit.migrad_ok() or not minuit.matrix_accurate():
         if not quiet:
+            from matplotlib import pyplot as plt
             plt.figure()
             uml.show()
             print(minuit.values)
@@ -114,6 +116,7 @@ def pprint_arg(vnames, value):
 
 
 def try_uml(f, data, bins=40, fbins=1000, *arg, **kwd):
+    from matplotlib import pyplot as plt
     fom = UnbinnedLH(f, data)
     narg = f.func_code.co_argcount
     vnames = f.func_code.co_varnames[1:narg]
@@ -139,6 +142,7 @@ def try_uml(f, data, bins=40, fbins=1000, *arg, **kwd):
 
 def try_binlh(f, data, weights=None, bins=40, fbins=1000, show='both', extended=False,
               bound=None, *arg, **kwd):
+    from matplotlib import pyplot as plt
     if bound is None: bound = minmax(data)
     fom = BinnedLH(f, data, extended=extended, bound=bound)
     narg = f.func_code.co_argcount
@@ -171,6 +175,7 @@ def try_binlh(f, data, weights=None, bins=40, fbins=1000, show='both', extended=
 
 
 def try_chi2(f, data, weights=None, bins=40, fbins=1000, show='both', *arg, **kwd):
+    from matplotlib import pyplot as plt
     fom = BinnedChi2(f, data)
     narg = f.func_code.co_argcount
     vnames = f.func_code.co_varnames[1:narg]

probfit/plotting.py

@@ -5,12 +5,12 @@
 from math import sqrt, ceil, floor
 from warnings import warn
 import numpy as np
-from matplotlib import pyplot as plt
 from .nputil import mid, minmax, vector_apply
 from .util import parse_arg, describe
 
 
 def draw_simultaneous(self, minuit=None, args=None, errors=None, **kwds):
+    from matplotlib import pyplot as plt
     numf = len(self.allf)
     ret = []
     numraw = sqrt(numf)
@@ -72,6 +72,8 @@ def draw_ulh(self, minuit=None, bins=100, ax=None, bound=None,
              parmloc=(0.05, 0.95), nfbins=200, print_par=True, grid=True,
              args=None, errors=None, parts=False, show_errbars='normal',
              no_plot=False):
+    from matplotlib import pyplot as plt
+
     data_ret = None
     error_ret = None
     total_ret = None
@@ -170,6 +172,8 @@ def draw_residual(x, y, yerr, xerr,
 
     The matplotlib axis instance the plot was drawn on.
     """
+    from matplotlib import pyplot as plt
+
     ax = plt.gca() if ax is None else ax
 
     if show_errbars:
@@ -196,6 +200,8 @@ def draw_residual_ulh(self, minuit=None, bins=100, ax=None, bound=None,
                       parmloc=(0.05, 0.95), print_par=False,
                       args=None, errors=None, errbar_algo='normal', norm=False,
                       **kwargs):
+    from matplotlib import pyplot as plt
+
     ax = plt.gca() if ax is None else ax
 
     arg, error = _get_args_and_errors(self, minuit, args, errors)
@@ -237,6 +243,8 @@ def draw_residual_ulh(self, minuit=None, bins=100, ax=None, bound=None,
 
 def draw_x2(self, minuit=None, ax=None, parmloc=(0.05, 0.95), print_par=True,
             args=None, errors=None, grid=True, parts=False, no_plot=False):
+    from matplotlib import pyplot as plt
+
     data_ret = None
     error_ret = None
     total_ret = None
@@ -289,6 +297,8 @@ def draw_x2(self, minuit=None, ax=None, parmloc=(0.05, 0.95), print_par=True,
 
 def draw_x2_residual(self, minuit=None, ax=None, args=None, errors=None, grid=True,
                      norm=False):
+    from matplotlib import pyplot as plt
+
     ax = plt.gca() if ax is None else ax
 
     arg, error = _get_args_and_errors(self, minuit, args, errors)
@@ -319,6 +329,8 @@ def draw_x2_residual(self, minuit=None, ax=None, args=None, errors=None, grid=Tr
 def draw_bx2(self, minuit=None, parmloc=(0.05, 0.95), nfbins=500, ax=None,
              print_par=True, args=None, errors=None, parts=False, grid=True,
              no_plot=False):
+    from matplotlib import pyplot as plt
+
     data_ret = None
     error_ret = None
     total_ret = None
@@ -379,6 +391,8 @@ def draw_bx2(self, minuit=None, parmloc=(0.05, 0.95), nfbins=500, ax=None,
 def draw_blh(self, minuit=None, parmloc=(0.05, 0.95),
              nfbins=1000, ax=None, print_par=True, grid=True,
              args=None, errors=None, parts=False, no_plot=False):
+    from matplotlib import pyplot as plt
+
     data_ret = None
     error_ret = None
     total_ret = None
@@ -436,6 +450,8 @@ def draw_blh(self, minuit=None, parmloc=(0.05, 0.95),
 def draw_residual_blh(self, minuit=None, parmloc=(0.05, 0.95),
                       ax=None, print_par=False, args=None, errors=None,
                       norm=False, **kwargs):
+    from matplotlib import pyplot as plt
+
     ax = plt.gca() if ax is None else ax
 
     arg, error = _get_args_and_errors(self, minuit, args, errors)
@@ -473,6 +489,8 @@ def draw_compare(f, arg, edges, data, errors=None, ax=None, grid=True, normed=Fa
     """
     TODO: this needs to be rewritten
     """
+    from matplotlib import pyplot as plt
+
     # arg is either map or tuple
     ax = plt.gca() if ax is None else ax
     arg = parse_arg(f, arg, 1) if isinstance(arg, dict) else arg
@@ -557,6 +575,8 @@ def draw_pdf_with_edges(f, arg, edges, ax=None, scale=1.0, density=True,
 
 
 def draw_pdf_with_midpoints(f, arg, x, ax=None, scale=1.0, normed_pdf=False, no_plot=False, **kwds):
+    from matplotlib import pyplot as plt
+
     ax = plt.gca() if ax is None and not no_plot else ax
     arg = parse_arg(f, arg, 1) if isinstance(arg, dict) else arg
     yf = vector_apply(f, x, *arg)
@@ -601,6 +621,8 @@ def draw_compare_hist(f, arg, data, bins=100, bound=None, ax=None, weights=None,
         - **parts** draw parts of pdf. (Works with AddPdf and Add2PdfNorm).
           Default False.
     """
+    from matplotlib import pyplot as plt
+
     ax = plt.gca() if ax is None else ax
     bound = minmax(data) if bound is None else bound
     h, e = np.histogram(data, bins=bins, range=bound, weights=weights)

probfit/toy.py

@@ -1,7 +1,6 @@
 from warnings import warn
 import numpy as np
 import numpy.random as npr
-from matplotlib import pyplot as plt
 from ._libstat import compute_cdf, invert_cdf, _vector_apply
 from .util import describe
 from .probfit_warnings import SmallIntegralWarning
@@ -62,6 +61,7 @@ def gen_toy(f, nsample, bound, accuracy=10000, quiet=True, **kwd):
 
     if not quiet:
         # move this to plotting
+        from matplotlib import pyplot as plt
         plt.figure()
         plt.title('comparison')
         numbin = 100

setup.py

@@ -50,8 +50,7 @@ def get_version():
     install_requires=[
         'setuptools',
         'numpy',
-        'iminuit',
-        'matplotlib'
+        'iminuit'
     ],
     classifiers=[
         "Programming Language :: Python",