From 524c0993f4f57ad158dc6dae481cae3ab09d7a2c Mon Sep 17 00:00:00 2001
From: Aaron LI <aly@aaronly.me>
Date: Wed, 30 Aug 2017 11:36:48 +0800
Subject: utils: Add analyze.py with function "countsdist_integrate()"

---
 fg21sim/utils/analyze.py | 66 ++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 66 insertions(+)
 create mode 100644 fg21sim/utils/analyze.py

(limited to 'fg21sim')

diff --git a/fg21sim/utils/analyze.py b/fg21sim/utils/analyze.py
new file mode 100644
index 0000000..f8137e9
--- /dev/null
+++ b/fg21sim/utils/analyze.py
@@ -0,0 +1,66 @@
+# Copyright (c) 2017 Weitian LI <weitian@aaronly.me>
+# MIT license
+
+"""
+Utilities to help analyze the simulation results.
+"""
+
+import logging
+
+import numpy as np
+
+
+logger = logging.getLogger(__name__)
+
+
+def inverse_cumsum(x):
+    """
+    Do cumulative sum reversely.
+
+    Credit: https://stackoverflow.com/a/28617608/4856091
+    """
+    x = np.asarray(x)
+    return x[::-1].cumsum()[::-1]
+
+
+def countdist_integrated(x, nbin, log=True):
+    """
+    Calculate the integrated counts distribution (i.e., luminosity
+    function), representing the counts (number of objects) with a
+    greater value.
+
+    Parameters
+    ----------
+    x : list[float]
+        Array of quantities of every object/source.
+    nbin : int
+        Number of bins to calculate the counts distribution.
+    log : bool, optional
+        Whether to take logarithm on the ``x`` quantities to determine
+        the bin edges?
+        Default: True
+
+    Returns
+    -------
+    counts : 1D `~numpy.ndarray`
+        The integrated counts for each bin, of length ``nbin``.
+    bins : 1D `~numpy.ndarray`
+        The central positions of every bin, of length ``nbin``.
+    binedges : 1D `~numpy.ndarray`
+        The edge positions of every bin, of length ``nbin+1``.
+    """
+    x = np.asarray(x)
+    if log is True:
+        x = np.log(x)
+
+    binedges = np.linspace(x.min(), x.max(), num=nbin+1)
+    bins = (binedges[1:] + binedges[:-1]) / 2
+    counts, __ = np.histogram(x, bins=binedges)
+    # Convert to the integrated counts distribution
+    counts = inverse_cumsum(counts)
+
+    if log is True:
+        bins = np.exp(bins)
+        binedges = np.exp(binedges)
+
+    return (counts, bins, binedges)
-- 
cgit v1.2.2