Jump to content

File:Donsker theorem for normal distributions.gif

Page contents not supported in other languages.
This is a file from the Wikimedia Commons
From Wikipedia, the free encyclopedia

Original file (2,000 × 600 pixels, file size: 4.5 MB, MIME type: image/gif, looped, 132 frames, 11 s)

Note: Due to technical limitations, thumbnails of high resolution GIF images such as this one will not be animated.

Summary

Description
English: ```python

import numpy as np import matplotlib.pyplot as plt import scipy

import tempfile import os import imageio

  1. generate random samples from

def plot_bridge(samples=None, n=1000, dist=scipy.stats.norm, dist_name="normal", low=-3, high=+3):

   fig, axes = plt.subplot_mosaic("ABB", figsize=(20, 6))
   ax1 = axes["A"]
   ax2 = axes["B"]
   if samples is None:
       samples = dist.rvs(size=n)
   else:
       n = len(samples)
   x = sorted(samples)
   y = dist.cdf(x)
   ref_x = np.linspace(low,high, 1000)
   ref_y = dist.cdf(ref_x)
   # calculate empirical cdf
   ecdf = np.arange(n) / len(samples)
   # plot cdf of standard normal distribution and empirical cdf of samples
   ax1.plot(ref_x, ref_y, label='Standard CDF')
   ax1.plot(x, ecdf, label='Empirical CDF')
   ax1.set_title()
   ax1.set_xlim(low,high)
   ax1.legend()
   ax1.set_ylim(0,1)
   ax2.plot([0.0] + dist.cdf(x)                        .tolist() + [1.0], 
            [0.0] + (np.sqrt(n) * (ecdf - dist.cdf(x))).tolist() + [0.0])
   ax2.set_title('centered, scaled, and re-timed')
   ax2.set_ylim(-0.9, 0.9)
   fig.suptitle(f"{dist_name}, with n = {n}")
   return fig

def interpolate_counts(counts, frames_per_step):

   interpolated_counts = [counts[0]]
   for i in range(1,len(counts)):
       interval = (counts[i] - counts[i-1]) // i
       interpolated_counts += list(range(counts[i-1], counts[i], interval))
   return interpolated_counts + [counts[-1]]

with tempfile.TemporaryDirectory() as temp_dir:

   dist = scipy.stats.norm
   dist_name = "normal"
   low, high = -3.5, +3.5
   
   n_steps = 16
   frames_per_step = 10
   sample_counts = interpolate_counts([2**n for n in range(n_steps)], frames_per_step)
   n_frames = len(sample_counts)-1
   samples = dist.rvs(size=sample_counts[0]).tolist()
   
   for i in range(n_frames):
       samples += dist.rvs(size=sample_counts[i+1]-sample_counts[i]).tolist()
       fig = plot_bridge(samples, dist=dist, dist_name=dist_name, low=low,high=high)
       filename = os.path.join(temp_dir, f"plot_{i:03d}.png")
       fig.savefig(filename)
       plt.close(fig)
   # Compile images into GIF
   fps = 12
   images = []
   for i in range(n_frames):
       filename = os.path.join(temp_dir, f"plot_{i:03d}.png")
       images.append(imageio.imread(filename))
   imageio.mimsave(f"{dist_name} Donsker theorem.gif", images, duration=1/fps)
```
Date
Source Own work
Author Cosmia Nebula

Licensing

I, the copyright holder of this work, hereby publish it under the following license:
w:en:Creative Commons
attribution share alike
This file is licensed under the Creative Commons Attribution-Share Alike 4.0 International license.
You are free:
  • to share – to copy, distribute and transmit the work
  • to remix – to adapt the work
Under the following conditions:
  • attribution – You must give appropriate credit, provide a link to the license, and indicate if changes were made. You may do so in any reasonable manner, but not in any way that suggests the licensor endorses you or your use.
  • share alike – If you remix, transform, or build upon the material, you must distribute your contributions under the same or compatible license as the original.

Captions

Donsker-Skorokhod-Kolmogorov theorem for normal distributions

3 March 2023

image/gif

File history

Click on a date/time to view the file as it appeared at that time.

Date/TimeThumbnailDimensionsUserComment
current06:09, 4 March 2023Thumbnail for version as of 06:09, 4 March 20232,000 × 600 (4.5 MB)Cosmia NebulaUploaded own work with UploadWizard

The following 2 pages use this file: