Python for Scientific Computing

Ivan Lima

Description

PythonOverviewScientificToolsAdditionalTools&ResourcesPythonforScientificComputingIvanLimaWoodsHoleOceanographicInstitutionDept. ofMarineChemistry&Goechemistry([e-mail])November14,2006IvanLima([e-mail]) PythonforScientificComputingPythonOverviewScientificToolsAdditionalTools&ResourcesOutlinePythonOverviewScientificToolsArrayObjectModulesGraphics&PlottingScientificPackagesExtendingPythonAdditionalTools&ResourcesIvanLima([e-mail]) PythonforScientificComputingPythonOverviewScientificToolsAdditionalTools&ResourcesWhatisPython?Pythonisanelegantandrobustprogramminglanguagethatcombinesthepowerandflexibilityoftraditionalcompiledlanguageswiththeease of use ofsimplerscriptingandinterpretedlanguages.I HighlevelI InterpretedI ScalableI ExtensibleI PortableI Easytolearn,readandmaintainI RobustI ObjectOrientedI VersatileIvanLima([e-mail]) PythonforScientificComputingPythonOverviewScientificToolsAdditionalTools&ResourcesWhyPython?1. Free&opensource2. Availableonawidevarietyofplatforms3. Bettersupportforarrayswithmorethan2dimensions4. BettersupportforwrappingFORTRAN77/90/95&C/C++code5. Farbettermemorymanagement6. Muchwiderlibrarysupportfornon numericalwork7. FarmoreoptionsforafullfeaturedGUI8. Easiertocreatestand aloneapplicationson anyplatform.9. TrulyObjectOrientedIvanLima([e-mail]) ...

Informations

• Publié par: mtoledan
• Publié le: 11 octobre 2011
• Nombre de lectures: 58
• Langue: English
• Poids de l'ouvrage: 5 Mo

Extrait

Python Overview Scientific Tools Additional Tools & Resources

Python for Scientific Computing

Ivan Lima

Woods Hole Oceanographic Institution Dept. of Marine Chemistry & Goechemistry ([e-mail])

November 14, 2006

Ivan Lima ([e-mail])

Python for Scientific Computing

Outline

Python Overview Scientific Tools Additional Tools & Resources

Python Overview

Scientific Tools Array Object Modules Graphics & Plotting Scientific Packages Extending Python

Additional Tools & Resources

Ivan Lima ([e-mail])

Python for Scientific Computing

Python Overview Scientific Tools Additional Tools & Resources

What is Python?

Python is anelegant and robustprogramming language that combines thepower and flexibilityof traditional compiled languages with the ease-of-useof simpler scripting and interpreted languages. I High level I Interpreted I Scalable I Extensible I Portable I Easy to learn, read and maintain I Robust I Object Oriented I Versatile

Ivan Lima ([e-mail]) Python for Scientific Computing

Why Python?

Python Overview Scientific Tools Additional Tools & Resources

1.Free & open source 2.Available on a wide variety of platforms 3.Better support for arrays with more than 2 dimensions 4.Better support for wrapping FORTRAN 77/90/95 & C/C++ code 5.Far better memory management 6.Much wider library support for non-numerical work 7.Far more options for a full featured GUI 8.Easier to create stand-alone applications onanyplatform. 9.Truly Object Oriented

Ivan Lima ([e-mail] for Scientific Computing) Python

Numeric

Python Overview Scientific Tools Additional Tools & Resources

I Original array module I Very successful I Large code base I Still around I Limited

Ivan Lima ([e-mail])

Array Object Modules Graphics & Plotting Scientific Packages Extending Python

Python for Scientific Computing

Numarray

Python Overview Scientific Tools Additional Tools & Resources

Array Object Modules Graphics & Plotting Scientific Packages Extending Python

I Created at the STSCI I Very powerful & lots of new features I Fast for large arrays but slow for small arrays I Incompatible with existing code base (Numeric/SciPy) I Slow acceptance I Fragmented community

Ivan Lima ([e-mail])

Python for Scientific Computing

NumPy

Python Overview Scientific Tools Additional Tools & Resources

Array Object Modules Graphics & Plotting Scientific Packages Extending Python

I Best of both Numeric & Numarray I Very powerful & flexible I Core array object for SciPy I Include migration tools for Numeric I Numeric development has ceased I Numarray development moving to NumPy

Ivan Lima ([e-mail] for Scientific Computing) Python

Python Overview Scientific Tools Additional Tools & Resources

Python/NumPy Example

>>> import numpy >>> a = numpy.arange(12) >>> print a [ 0 1 2 3 4 5 6 7 8 9 10 11] >>> print a.size, a.ndim, a.shape 12 1 (12,) >>> a = numpy.reshape(a,(3,4)) >>> print a [[ 0 1 2 3] [ 4 5 6 7] [ 8 9 10 11]] >>> print a.size, a.ndim, a.shape 12 2 (3, 4) >>> print a.min(), a.max(), a.mean(), a.std() 0 11 5.5 3.45205252953 >>> print a[:,0] [0 4 8] >>> print a[2,:] [ 8 9 10 11] >>> b = numpy.arange(3) >>> c = numpy.arange(4) >>> b * c ValueError: shape mismatch: objects cannot be broadcast to a single shape >>> print b, c [0 1 2] [0 1 2 3]

Ivan Lima ([e-mail])

Array Object Modules Graphics & Plotting Scientific Packages Extending Python

>>> print numpy.reshape(b,(3,1)) [[0] [1] [2]] >>> print numpy.reshape(b,(3,1)) * c [[0 0 0 0] [0 1 2 3] [0 2 4 6]] >>> print b[:,numpy.newaxis] * c [[0 0 0 0] [0 1 2 3] [0 2 4 6]] >>> print a.shape, c.shape (3, 4) (4,) >>> print a * c [[ 0 1 4 9] [ 0 5 12 21] [ 0 9 20 33]] >>> print a.shape, b.shape (3, 4) (3,) >>> print a * b[:,numpy.newaxis] [[ 0 0 0 0] [ 4 5 6 7] [16 18 20 22]] >>> print a.shape, b[:,numpy.newaxis].shape (3, 4) (3,1)

Python for Scientific Computing

Python Overview Scientific Tools Additional Tools & Resources

Matplotlib/Basemap

Array Object Modules Graphics & Plotting Scientific Packages Extending Python

I Emulates Matlab syntax (pylab interface) I Good for interactive sessions I Object Oriented I Excellent 2-D graphics & some 3-D capability I Map toolkit (Basemap) I Works with Numeric, Numarray & NumPy

Ivan Lima ([e-mail]) Python for Scientific Computing

Array Object Modules Py S t c h i o e n nti O fi v c er T v o i o e l w s Graphics & Plotting Additional Tools & Reso Scientific Packages urces Extending Python

Matplotlib Example: Simple Plot

>>> import numpy as N >>> import pylab as pl >>> x = N.arange(0,1,0.01) * 5 * N.pi >>> pl.plot(x,N.sin(x)) >>> pl.xlabel(’x’) >>> pl.ylabel(’sin(x)’) >>> pl.title(’Plot Title’)

Ivan Lima ([e-mail])

Python for Scientific Computing

Python Overview Array Object Modules Scientific Tools  Graphics & Plotting Additional Tools & Resources Scientific Packages Extending Python

Matplotlib Example: 2-D Plot

>>> import numpy as N >>> import pylab as pl >>> def z_func(x,y): ... return (1-x+x**3+y**5)*N.exp(-(x**2+y**2)) ... >>> x = N.arange(-3.0,3.0,0.025) >>> y = N.arange(-3.0,3.0,0.025) >>> X,Y = pl.meshgrid(x, y) >>> Z = z_func(X, Y) >>> im = pl.imshow(Z,interpolation=’bilinear’,/ ... cmap=pl.cm.Spectral) >>> cset = pl.contour(Z,N.arange(-1.2,1.6,0.2),/ ... linewidths=2,cmap=pl.cm.hot) >>> pl.clabel(cset,inline=True,fmt=’%1.1f’,/ ... fontsize=10) >>> pl.colorbar(im) >>> pl.axis(’off’) >>> pl.title(’$z=(1-x+xˆ3+yˆ5) eˆ-(xˆ2+yˆ2)$’)

Ivan Lima ([e-mail])

Python for Scientific Computing