Python Rocks Introducing the Python programming language to the SMEDG community By Rupert Osborn H&S Consultants Presentation notes

Pages on a blue background are additions to the presentation for those who didnt attend the talk They give a summary of the points made in the talk where necessary The second part of the talk was presented using the IPython Notebook. This should also be available for download from the SMEDG website The last slide gives advice on how to download Python use a distribution like Anaconda! http://continuum.io/downloads What is Python? Python is a general purpose, high-level, object oriented, programming language

Free and open-source Readable and intuitive What is Python? Python is an easy to use programming language Many geologists feel intimidated by programming languages Programming languages can help the exploration geologist by automating repetitive tasks and enable quicker, deeper data analysis

History of Python Started in 1989 by Guido van Rossum Named after Monty Pythons Flying Circus History of Releases Python 1.0 released in 1994 Python 2.0 released in 2000 Python 3.0 released in 2008 Python 2.7 is currently the recommended version??? Python 3 is the future of the language Micromine uses Python 3.3 (so thats what I

use) Why Use Python? Fast to write Easy to read Continually developing Thriving ecosystem of third-party libraries Huge number of simple and advanced functions Loads of examples and help online Great IDEs (integrated development environments) Multi-platform (Windows, Apple, Linux, etc)

Why Use Python? Reads like English Third-party libraries (e.g. NumPy) add huge functionality and are growing Was considered as the glue between languages but is quickly becoming the Swiss army knife Basically, IDEs are the interface that you type into. The ones available free for Python are great. They mark up mistakes (syntax errors) and include tab completion which is a bit like predictive text

Languages used at CSIRO Python Matlab IDL Perl Fortran C C++ C# Java VB.net R + 5-10 others

Source: Python Charmers presentation Different language for each task? Languages used at CSIRO Python is developing to be able to do all the tasks done by a large selection of specialist languages Python has improved massively over the last five years because more and more people are enjoying the ease of use and therefore more people are putting an effort

into producing more functions and thirdparty libraries Computing Speed for $ Year Approximate cost Platform providing the lowest cost per per GFLOPS (US$) GFLOPS 1961 1984 $1,100,000,000,000 About 17 million IBM 1620 units costing $64,000 each $18,750,000 Cray X-MP/48 1997 2000

2003 $30,000 $1,000 $82 Beowulf clusters with Pentium Pro microprocessors Bunyip Beowulf cluster KASY0 2007 2011 $48 $1.80

Microwulf HPU4Science 2013 2015 $0.22 $0.08 Sony PlayStation 4 Celeron G1830 R9 295x2 System Computing Speed for $

GFLOPS means Giga Floating point Operations per Second Fortran was designed when programmers time was cheaper than processing time Fortran and C are 10x slower to write than Python (and that is not including the use of ready made functions available in Pythons third-party libraries)

The computer takes longer to process Python scripts but optimised libraries like NumPy make it much quicker (similar to C++) Python Successes United Space Agency - NASA Google: Maps, Gmail, Groups, News YouTube, Reddit, BitTorrent Civilisation IV Financial analysis Research: Universities worldwide for a variety of disciplines Python Successes

Because Python is easy and quick to read and write, and because it is so versatile, people are using it more and more for a variety of applications Python Libraries For Scientific Computing Python Libraries For Scientific Computing So far I have tried to explain why Python is a good choice of languages to learn But how will a programming language

benefit geologists? Python has a load of third-party libraries, some of which are shown on the slide Specialist geological libraries are lacking but there are a few geophysical and hydrogeological libraries available or in development NumPy: Numerical Python NumPy is the fundamental package for scientific computing with Python. It adds a

fast and sophisticated array facility to the Python language NumPy The NumPy library forms the basis of Pythons scientific

computing ability It includes routines for mathematical, logical, shape manipulation, sorting, selecting, discrete Fourier transforms, basic linear algebra, basic statistical operations, random simulation and much more Python uses loops to iterate over data. This is slow because it needs to convert commands to machine code and execute them one by one NumPy is built to operate efficiently on arrays of numbers by sending batches of data to optimized C and Fortran code The first part of the IPython Notebook that accompanies this presentation explains a little more about why NumPy is so useful Pandas

DataFrame object for data manipulation with integrated indexing Label-based slicing, fancy indexing, and subsetting of large data sets

Tools for reading and writing data to different file formats Basic but handy quick plotting Group by engine allowing split-apply-combine operations on data sets Reshaping and pivoting of data sets Time series-functionality, merging and joining, integrated handling of missing data, Data structure column insertion and deletion, Hierarchical axis indexing to work with highdimensional data in a lower-dimensional data structure Pandas Pandas uses a DataFrame object which can be thought of as a table of data (although it can be more complicated) It was built by the finance sector to aid with data manipulation and data analysis

It has loads of brilliant functions to really dig into your data It has useful functions for reading and writing to file types such as csv and Excel Matplotlib Plotting library for graphs Designed to closely resemble MATLAB Simple to publication quality graphs Great gallery with examples Plenty of tutorials and help (Stack Overflow) Matplotlib

Matplotlib is now the recommended plotting library to make graphs etc. The Matplotlib figures can be easily customised and produce publication quality plots Using the Matplotlib, NumPy and Pandas libraries together make data analysis much easier and reproducible than in Excel SciPy: Scientific Python

Gathers a variety of high level science and engineering modules together: stats: statistical functions spatial: KD-trees, nearest neighbors, distance functions

interpolate: interpolation tools e.g. IDW, RBF ndimage: various functions for multi-dimensional image processing optimize: optimization algorithms including linear programming constants: physical constants and conversion factors fftpack: Discrete Fourier Transform algorithms integrate: numerical integration routines linalg: linear algebra routines misc: miscellaneous utilities (e.g. image reading/writing) signal: signal processing tools special: special functions SciPy: Scientific Python The SciPy library has loads of more

advanced functions enabling, for example, high-level statistical functions to be called using a single line of code Scikits: SciPy Toolkits scikit-learn: various classification, regression and clustering algorithms including support vector machines, random forests, gradient boosting, k-means and DBSCAN scikit-image: algorithms for segmentation, geometric transformations, colour space manipulation, analysis, filtering, morphology, feature detection, and more.

Scikits: SciPy Toolkits Scikits offer a few more specialised libraries e.g. Scikits-Learn which provides a load of machine learning capabilities. Supervised and unsupervised classification processes are available The accompanying IPython Notebook shows how easy it is to conduct dimensionality reduction (Factor Analysis) and unsupervised classification (k-means clustering) on ICP data to help differentiate rock types

IDE (Integrated Development Environment) IDEs are basically the interface into which you type the code Great packages like Spyder, PyCharm and the Ipython Notebook Auto-complete and tab-complete options Syntax and spelling errors are highlighted automatically Download and Install There are a couple of Python distributions that are easy to download and install These include the most popular Python

libraries Anaconda is recommended for science, math, engineering, data analysis. The link will only work if viewing this presentation in Slide Show. Alternatively just search for Download Anaconda Ipython Notebook Example For those at the presentation were going to go through the IPython Notebook part of the presentation A version of this will be available for download but unfortunately I wont be able to provide the data with it

Install Anaconda then you can try it with your own data