Blogue

Demo of the IPython notebook at Sage Paris group meeting

06 mars 2014 | Catégories: ipython, sage | View Comments

Today I am presenting the IPython notebook at the meeting of the Sage Paris group. This post gathers what I prepared.

Installation

First you can install the ipython notebook in Sage as explained in this previous blog post. If everything works, then you run:

sage -ipython notebook

and this will open a browser.

Turn on Sage preparsing

Create a new notebook and type:

In [1]: 3 + 3
6
In [2]: 2 / 3
0
In [3]: matrix
Traceback (most recent call last):
...
NameError: name 'matrix' is not defined

By default, Sage preparsing is turn off and Sage commands are not known. To turn on the Sage preparsing (thanks to a post of Jason on sage-devel):

%load_ext sage.misc.sage_extension

Since sage-6.2, according to sage-devel, the command is:

%load_ext sage

You now get Sage commands working in ipython:

In [4]: 3 + 4
Out[4]: 7
In [5]: 2 / 3
Out[5]: 2/3
In [6]: type(_)
Out[6]: <type 'sage.rings.rational.Rational'>
In [7]: matrix(3, range(9))
Out[7]:
[0 1 2]
[3 4 5]
[6 7 8]

Scroll and hide output

If the output is too big, click on Out to scroll or hide the output:

In [8]: range(1000)

Sage 3d Graphics

3D graphics works but open in a new Jmol window:

In [9]: sphere()

Sage 2d Graphics

Similarly, 2D graphics works but open in a new window:

In [10]: plot(sin(x), (x,0,10))

Inline Matplotlib graphics

To create inline matplotlib graphics, the notebook must be started with this command:

sage -ipython notebook --pylab=inline

Then, a matplotlib plot can be drawn inline (example taken from this notebook):

import matplotlib.pyplot as plt
import numpy as np
x = np.linspace(0, 3*np.pi, 500)
plt.plot(x, np.sin(x**2))
plt.title('A simple chirp');

Or with:

%load http://matplotlib.org/mpl_examples/showcase/integral_demo.py

According to the previous cited notebook, it seems, that the inline mode can also be decided from the notebook using a magic command, but with my version of ipython (0.13.2), I get an error:

In [11]: %matplotlib inline
ERROR: Line magic function `%matplotlib` not found.

Use latex in a markdown cell

Change an input cell into a markdown cell and then you may use latex:

Test $\alpha+\beta+\gamma$

Output in latex

The output can be shown with latex and mathjax using the ipython display function:

from IPython.display import display, Math
def my_show(obj): return display(Math(latex(obj)))
y = 1 / (x^2+1)
my_show(y)

ipynb format

Create a new notebook with only one cell. Name it range_10 and save:

In [1]: range(10)
Out[1]: [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]

The file range_10.ipynb is saved in the directory. You can also download it from File > Download as > IPython (.ipynb). Here is the content of the file range_10.ipynb:

{
 "metadata": {
  "name": "range_10"
 },
 "nbformat": 3,
 "nbformat_minor": 0,
 "worksheets": [
  {
   "cells": [
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "range(10)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "pyout",
       "prompt_number": 1,
       "text": [
        "[0, 1, 2, 3, 4, 5, 6, 7, 8, 9]"
       ]
      }
     ],
     "prompt_number": 1
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [],
     "language": "python",
     "metadata": {},
     "outputs": []
    }
   ],
   "metadata": {}
  }
 ]
}

ipynb is just json

A ipynb file is written in json format. Below, we use json to open the file `range_10.ipynb as a Python dictionnary.

sage: s = open('range_10.ipynb','r').read()
sage: import json
sage: D = json.loads(s)
sage: type(D)
dict
sage: D.keys()
[u'nbformat', u'nbformat_minor', u'worksheets', u'metadata']
sage: D
{u'metadata': {u'name': u'range_10'},
 u'nbformat': 3,
 u'nbformat_minor': 0,
 u'worksheets': [{u'cells': [{u'cell_type': u'code',
     u'collapsed': False,
     u'input': [u'range(10)'],
     u'language': u'python',
     u'metadata': {},
     u'outputs': [{u'output_type': u'pyout',
       u'prompt_number': 1,
       u'text': [u'[0, 1, 2, 3, 4, 5, 6, 7, 8, 9]']}],
     u'prompt_number': 1},
    {u'cell_type': u'code',
     u'collapsed': False,
     u'input': [],
     u'language': u'python',
     u'metadata': {},
     u'outputs': []}],
   u'metadata': {}}]}

Load vaucanson.ipynb

Download the file vaucanson.ipynb from the last meeting of Paris Sage Users. You can view the complete demo including pictures of automaton even if you are not able to install vaucanson on your machine.

IPython notebook from a Python file

In a Python file, separate your code with the following line to create cells:

# <codecell>

For example, create the following Python file. Then, import it in the notebook. It will get translated to ipynb format automatically.

# -*- coding: utf-8 -*-
# <nbformat>3.0</nbformat>

# <codecell>

%load_ext sage.misc.sage_extension

# <codecell>

matrix(4, range(16))

# <codecell>

factor(2^40-1)

More conversion

Since release 1.0 of IPython, many conversion from ipynb to other format are possible (html, latex, slides, markdown, rst, python). Unfortunately, the version of IPython in Sage is still 0.13.2 as of today but the version 1.2.1 will be in sage-6.2.

Read and Post Comments

Using sage in the new ipython notebook

10 février 2013 | Catégories: ipython, sage | View Comments

(NEW: See also the demo I made at the Sage Paris group meeting in March 2014.)

Ticket #12719 (Upgrade to IPython 0.13) was merged into sage-5.7.beta1. This took a lot of energy (see the number of comments in the ticket and especially the number of patches and dependencies). Big thanks to Volker Braun, Mike Hansen, Jason Grout, Jeroen Demeyer who worked on this since almost one year. Note that in December 2012, the IPython project has received a $1.15M grant from the Alfred P. Sloan foundation, that will support IPython development for the next two years. I really like this IPython sage command line interface so it is really good news!

The IPython notebook

Since version 0.12 (December 21 2011), IPython is released with its own notebook. The differences with the Sage Notebook are explained by Fernando Perez, leader of IPython project, in the blog post The IPython notebook: a historical retrospective he wrote in January 2012. One of the differences is that the IPython Notebook run in its own directory whereas each cell of the Sage Notebook lives in its directory. As William Stein says in the presentation Browser-based notebook interfaces for mathematical software - past, present and future he gave last December at ICERM, there are plenty of projects and directions these days for those interfaces.

In May 2012, I tested the same ticket which was to upgrade to IPython 0.12 at that time. Today, I was curious to test it again.

First, I installed sage-5.7.beta4:

./sage -version
Sage Version 5.7.beta4, Release Date: 2013-02-09

Install tornado:

./sage -sh
easy_install-2.7 tornado

[update March 6th, 2014] Note that some linux user have to install libssl-dev before tornado:

sudo apt-get install libssl-dev

Install zeromq and pyzmq:

./sage -i zeromq
./sage -i pyzmq

Start the ipython notebook:

./sage -ipython notebook
[NotebookApp] Using existing profile dir: u'/Users/slabbe/.sage/ipython-0.12/profile_default'
[NotebookApp] Serving notebooks from /Users/slabbe/Applications/sage-5.7.beta4
[NotebookApp] The IPython Notebook is running at: http://127.0.0.1:8888/
[NotebookApp] Use Control-C to stop this server and shut down all kernels.
/Files/2013/ipython_notebook_home.png

Create a new notebook. One may use sage commands by adding the line from sage.all import * in the first cell.

/Files/2013/ipython_notebook_nb.png

The next things I want to look at are:

  • Test the conversion of files from .py to .pynb.
  • Test the conversion of files from .rst to .pynb.
  • Test the qtconsole.
  • Test the parallel computing capacities of the IPython.
Read and Post Comments