60 - Custom PDF.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Creating your own pdf\n",
    "\n",
    "A core feature of zfit is the ability to create custom pdfs and functions in an simple and straightforward way.\n",
    "\n",
    "There are two main possibilities to create a custom pdf, an easier for most use-cases and an advanced way..\n",
    "\n",
    "## The simple way\n",
    "\n",
    "While the same works for functions, an example with a PDF is shown here"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import zfit\n",
    "from zfit import ztf\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "class MyGauss(zfit.pdf.ZPDF):\n",
    "    _N_OBS = 1  # dimension, can be omitted\n",
    "    _PARAMS = ['mean', 'std']  # the name of the parameters\n",
    "    \n",
    "    def _unnormalized_pdf(self, x):\n",
    "        x = ztf.unstack_x(x)  # returns a list with the columns: do x, y, z = ztf.unstack_x(x) for 3D\n",
    "        mean = self.params['mean']\n",
    "        std = self.params['std']\n",
    "        return ztf.exp(- ((x - mean)/std)**2)\n",
    "        "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Done. Now we can use our pdf already!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "obs = zfit.Space('obs1', limits=(-3, 6))\n",
    "\n",
    "data_np = np.random.random(size=1000)\n",
    "data = zfit.data.Data.from_numpy(array=data_np, obs=obs)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Create two parameters and an instance of your own pdf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /home/jonas/anaconda3/envs/zfit36/lib/python3.6/site-packages/tensorflow/python/ops/resource_variable_ops.py:435: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Colocations handled automatically by placer.\n"
     ]
    }
   ],
   "source": [
    "mean = zfit.Parameter(\"mean\", 1.)\n",
    "std = zfit.Parameter(\"std\", 1.)\n",
    "my_gauss = MyGauss(obs='obs1', mean=mean, std=std)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "probs = my_gauss.pdf(data, norm_range=(-3, 4))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0.54061221 0.26987065 0.55748768 0.55287316 0.35744102 0.56147295\n",
      " 0.24496626 0.27563618 0.469028   0.50871657 0.2807743  0.23819832\n",
      " 0.28492248 0.56415543 0.49875545 0.25174901 0.5481325  0.21933145\n",
      " 0.33557439 0.48277319]\n"
     ]
    }
   ],
   "source": [
    "probs_np = zfit.run(probs)\n",
    "print(probs_np[:20])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We could improve our PDF by registering an integral"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def gauss_integral_from_any_to_any(limits, params, model):\n",
    "    lower, upper = limits.limit1d\n",
    "    mean = params['mean']\n",
    "    std = params['std']\n",
    "    # write your integral here\n",
    "    return 42.  # dummy integral, must be a scalar!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "limits = zfit.Space.from_axes(axes=0, limits=(zfit.Space.ANY_LOWER, zfit.Space.ANY_UPPER))\n",
    "MyGauss.register_analytic_integral(func=gauss_integral_from_any_to_any, limits=limits)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### n-dimensional"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Advanced Custom PDF\n",
    "\n",
    "Subclass BasePDF. The `_unnormalized_pdf` has to be overriden and, in addition, the `__init__`.\n",
    "\n",
    "Any of the public main methods (`pdf`, `integrate`, `partial_integrate` etc.) can **always** be overriden by implementing the function with a leading underscore, e.g. implement `_pdf` to directly controls `pdf`, the API is the same as the public function without the name. In case, during execution of your own method, it is found to be a bad idea to have overridden the default methods, throwing a `NotImplementedError` will restore the default behavior."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# TOBEDONE"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}