Download Introduction to Python for Scientific Computing

Introduction to Python
for Scientific Computing
http://tinyurl.com/cq-intro-python-20160217
By: Bart Oldeman, Calcul Québec – McGill HPC
[email protected], [email protected]
1
Partners and Sponsors
2
Outline of the workshop
• What is Python? Why is it useful?
• How do we run Python codes, including on the Guillimin cluster?
• How to program with Python?
• How to use scientific packages in Python (NumPy, SciPy,
Matplotlib)?
• Practical exercises on Guillimin
• Note: These slides are based on the lectures notes here:
http://scipy-lectures.github.io/
3
What is Python?
• Python is a high-level, general-purpose programming language.
• Readable code.
• (Usually) Interpreted, not compiled.
• Dynamically typed.
• Original author and “Benevolent Dictator For Life (BFDL)”: Guido
van Rossum (originally at CWI, Amsterdam, now at Dropbox).
• Named after Monty Python, Van Rossum’s favourite TV show.
• History: Started 1989, Python 1.0 (1994), Python 2.0 (2000),
Python 3.0 (2008). Using Python 3.5.0 in this workshop.
4
Why is it useful?
• Provides a general-purpose, open-source, high-level glue between
computations.
• Alternative to programs such as MATLAB, R, Julia, IDL, etc., for
many numerical computations. Python is less specialized, which has
pros:
• easier to interface with non-scientific code
• can do almost everything
and cons
• slightly more cumbersome syntax for arrays
• Used in many disciplines, e.g. bioinformatics, mathematics,
statistics, biology, economics, physics, electrical engineering,
geosciences, signal processing.
• Groups on Guillimin use it, for instance, to study
•
•
•
•
Gene flow in animal populations
Climate change
Searching for pulsars
Computational Fluid Dynamics
5
Scientific Python stack
Python The base language. It has “batteries included” but no fast
array type by default.
Numpy Provides efficient powerful numerical array type with
associated routines. Almost all scientific software written in
Python uses Numpy.
Scipy Higher level than Numpy, provides many data processing
routines, for instance optimization, regression,
interpolation.
Matplotlib Most popular 2D (and basic 3D) plotting library.
IPython Advance Python shell that integrates well with scientific
Python packages.
Mayavi 3D visualization.
Mpi4py Use MPI from Python.
Cython C-extensions for Python.
Sympy Symbolic mathematics.
6
Exercise 1:
Log in to Guillimin, setting up the environment
Windows users: please use http://mobaxterm.mobatek.net/.
1) Log in to Guillimin:
ssh -X class##@guillimin.hpc.mcgill.ca
2) Check for loaded software modules:
guillimin> module list
3) See all available modules:
guillimin> module av
4) Load necessary modules: (needs ~/.lmod legacy until March 15)
guillimin> module load iomkl/2015b Python/3.5.0
5) Check loaded modules again
6) Verify that you have access to the correct Python package:
guillimin> which python3
output: /software/CentOS6/eb/software/Toolchain/iomkl/2015b/Python/3.5.0/bin/python3
7
Interactive job submission
1) Copy all workshop files to your home directory:
guillimin> cp -a /software/workshop/python/* ./
2) Inspect the file “interactive.pbs”:
guillimin> cat interactive.pbs
#!/bin/bash
#PBS -l nodes=1:ppn=1,walltime=06:00:00
#PBS -N python-interactive
#PBS -V
#PBS -X
#PBS -I
8
Interactive job submission
3) guillimin> qsub interactive.pbs
qsub: waiting for job 28812458.gm-1r16-n04.guillimin.clumeq.ca to start
qsub: job 28812458.gm-1r16-n04.guillimin.clumeq.ca ready
---------------------------------------Begin PBS Prologue Wed Mar 11 12:51:09 EDT 2015 1426092669
Job ID:
28812458.gm-1r16-n04.guillimin.clumeq.ca
Username:
class01
Group:
class
Nodes:
sw-2r08-n24
End PBS Prologue Wed Mar 11 12:51:09 EDT 2015 1426092669
---------------------------------------[classxx@sw-2r08-n24 ~]$
9
Run “Hello World”
4) [classxx@sw-2r08-n24 ~]$ cat hello.py
print("Hello, world!")
[classxx@sw-2r08-n24 ~]$ python3 hello.py
Hello, world!
[classxx@sw-2r08-n24 ~]$ python3
Python 3.5.0 (default, Sep 30 2015, 10:10:22)
[GCC Intel(R) C++ gcc 4.9 mode] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> print("Hello, world!")
Hello, world!
>>> quit()
Note: for Python 2 (unless you use Python 2.6+ with from future
import print function) use print "Hello, world!" instead.
10
Run “Hello World” (IPython)
5) [classxx@sw... ~]$ module load IPython/3.2.3-Python-3.5.0
[classxx@sw... ~]$ ipython
Python 3.5.0 (default, Sep 30 2015, 10:10:22)
Type "copyright", "credits" or "license" for more information.
IPython 3.2.3 -- An enhanced Interactive Python.
?
-> Introduction and overview of IPython’s features.
%quickref -> Quick reference.
help
-> Python’s own help system.
object?
-> Details about ’object’, use ’object??’ for extra details.
In [1]: print("Hello, world!")
Hello, world!
In [2]: %run hello.py
Hello, world!
In [3]: quit
[classxx@sw... ~]$ module load PyQt/4.11.4-Python-3.5.0
[classxx@sw... ~]$ ipython qtconsole
# Brings up a GUI
11
Python script for I/O: examples/io.py
import math
import os
infile = " numbers "
outfile = " f_numbers "
f = open ( infile , ’r ’)
g = open ( outfile , ’w ’)
def func ( y ) :
if y >= 0.0:
return y **5.0* math . exp ( - y )
else :
return 0.0
for line in f :
line = line . split ()
x , y = float ( line [0]) , float ( line [1])
g . write ( " % g %12.5 e \ n " % (x , func ( y ) ) )
f . close ()
g . close ()
Run as cd data; python3 ../examples/io.py
12
Setting up Jupyter
Jupyter (formally known as IPython Notebook) is a web-based interactive
computational environment where execution, text, mathematics, plots
and rich media combine into a single document. We use it for the
remainder of the workshop. Setting up Jupyter:
1. Open https://guillimin2.hpc.mcgill.ca:8443 in a web
browser.
2. In the web page, trust this web site and login using your class
account. Please “Continue” any unresponsive scripts.
Alternatively, you could install a scientific Python distribution such as
Continuum’s Anaconda or Enthought’s Canopy and use Jupyter locally.
In that case you can obtain the workshop files as follows:
yourlaptop> git clone -b mcgill \
https://github.com/calculquebec/cq-formation-intro-python.git
yourlaptop> cd cq-formation-intro-python
13
Further reading:
• Python, NumPy, SciPy, Matplotlib and IPython can all be found at
the .org website, for instance:
http://www.python.org
• Official non-scientific Python tutorial:
http://docs.python.org/3/tutorial
• Two more scientific tutorials:
http://github.com/jrjohansson/scientific-python-lectures
http://python4astronomers.github.io/
• Software Carpentry:
http://swcarpentry.github.io/python-novice-inflammation/
• Followup questions? Contact [email protected].
14