Comments for Codebright's Blog

Comment on Python regular expression surprise by Nagesh

Nagesh — Thu, 20 Sep 2012 11:54:29 +0000

Thanks, was looking out for this!

Comment on SendKeys in Linux by Henry Charles

Henry Charles — Fri, 09 Mar 2012 13:54:57 +0000

Found this, works great: https://sourceforge.net/projects/x11guitest/

Comment on Reading gzip files in Python – fast! by me

me — Mon, 30 Jan 2012 17:42:23 +0000

I’d:

- avoid reading the entire file into memory using subprocess.communicate() and cStringIO
- initialize the cc, lc and sz outside of the internal loop

Comment on Reading gzip files in Python – fast! by insider

insider — Wed, 18 Jan 2012 10:32:20 +0000

cool, i’m doing some researches in parsing syslog logs, do you know any syslog parse module like in Perl Parse::Syslog?

Comment on Linear Algebra Review and numpy by codebright

codebright — Sun, 09 Oct 2011 20:45:28 +0000

Many thanks for all the hints and tips. I appreciate the input.
Paul

Comment on Linear Algebra Review and numpy by dietrich

dietrich — Sat, 08 Oct 2011 11:57:10 +0000

I used to work with Octave before switching to python. You’ll find a good cheat sheet at http://www.scipy.org/NumPy_for_Matlab_Users
It helped me a lot.
Cheers, dietrich

Comment on Linear Algebra Review and numpy by Ludovico Fischer

Ludovico Fischer — Sat, 08 Oct 2011 09:28:49 +0000

There is no need to use numpy.matrix. You can obtain matrix multiplication on an numpy.array with the .dot method: arrayA.dot(arrayA) gives the same result as matrixA * matrixA.

Comment on Linear Algebra Review and numpy by Gael Varoquaux

Gael Varoquaux — Sat, 08 Oct 2011 09:19:49 +0000

You can also write the transpose as: ‘A.T’.

On a side note, I suggest strongly that you do not use the numpy matrix object but stick to arrays, and use the ‘np.dot’ function to do matrix multiplication. The reason is that it looks very much like an array object, but behaves ever so slightly differently. When you start having a large codebase, the mixture of the two will make your code hard to follow. This is the prevailing view in the numy community.

Finally, to do machine learning in Python, check out the scikit-learn. Hopefully it should be a good companion to the Stanford courses .

Comment on Linear Algebra Review and numpy by bc

bc — Sat, 08 Oct 2011 09:07:02 +0000

Note, numpy matrices support the following abbreviations:
>>> import numpy
>>> m = numpy.matrix([[1,2],[3,4]],dtype=numpy.complex128) + 0.0+0.5j
>>> m
matrix([[ 1.+0.5j, 2.+0.5j],
[ 3.+0.5j, 4.+0.5j]])
>>> m.T #transpose of m
matrix([[ 1.+0.5j, 3.+0.5j],
[ 2.+0.5j, 4.+0.5j]])
>>> m.H #Herminian conjugate
matrix([[ 1.-0.5j, 3.-0.5j],
[ 2.-0.5j, 4.-0.5j]])
>>> m.I #Inverse of m
matrix([[-2.0-0.25j, 1.0+0.25j],
[ 1.5+0.25j, -0.5-0.25j]])
>>> m.A #convert to an array
array([[ 1.+0.5j, 2.+0.5j],
[ 3.+0.5j, 4.+0.5j]])
>>> m.A1 #convert to a flattened array
array([ 1.+0.5j, 2.+0.5j, 3.+0.5j, 4.+0.5j])