unpingco/python-for-signal-processing

Notebooks for "Python for Signal Processing" book

Jupyter notebooks for Python 2.7 for Signal Processing Book.

This book is available as a blog where you can read the formatted notebooks and comment further. The following are the draft Jupyter notebooks. A subset of the blog and the content here is available in printed form on Amazon.

Notebook Viewer Static Page Views

Signal Processing Reading List

• Sampling Theorem
• Sampling Theorem Part 2 v2
• Fourier Transform
• Frequency Resolution
• More Fourier Transform
• Windowing
• Windowing Part2
• Windowing Part3
• Filtering
• Filtering Part2
• Filtering Part3
• Compressive Sampling

Stochastic Processes Reading List

• Conditional Expectation Gaussian
• Conditional expectation MSE
• Conditional expectation MSE Ex
• Conditional Expectation Projection
• Projection
• Projection Ex
• Projection mdim
• Inverse Projection Constrained Optimization
• Gauss Markov
• Maximum likelihood
• Expectation Maximization
• Markov chains
• Buffons Needle Sim
• Sampling Monte Carlo
• Rectangle Wedge Tail Decomposition

Misc.

• Example CSVs

python-for-signal-processing's Issues

How to apply windowing in time and frequency domain

Hi,
The windowing you are applying to different signals in Windowing.ipynb is in time-domain or frequency domain?
If i have a signal e.g. signal1=numpy.ones((1,50)) and i am applying windowing to it as
wt=numpy.transpose(numpy.kaiser(50,2.5))
signal = signal1*wt.
Is it the correct way of applying windowing?
How to apply windowing in frequency domain?

No overview page

I am looking forward to read this book, but I have no idea how the notebooks are ordered.

There should be an overview page listing the notebooks in order.

case

Signal Processing (Python)

Typo in Frequency_resolution.ipynb?

In Cell [4] shouldn't

t = arange(0,2,1/fs) actually be t = arange(0,1,1/fs) ?

Expectation_Maximization notebook: name 'Lf' is not defined error from like 319

Hello. I was working with this really great notebook, but noticed an error at the very bottom. Line 319 says

z=Lf(xs,mua_step[:,None],mub_step[:,None,None]).sum(axis=2) # numpy broadcasting

So the Lf variable was defined in the scope of the run() function, but not in the global scope. So trying to run the Lf() function is generating the error.

I fixed it by adding this line just above line 319:

Lf=sympy.lambdify((x,mu_a,mu_b), sympy.log(abs(L)),'numpy')

So the total code looks like:

Lf=sympy.lambdify((x,mu_a,mu_b), sympy.log(abs(L)),'numpy')
z=Lf(xs,mua_step[:,None],mub_step[:,None,None]).sum(axis=2) # numpy broadcasting

Hope this helps.

Erroneous math in maximum likelihood estimation document

Hi!

I was reading Maximum Likelihood Estimation (direct link to the github file here) and noticed that there seems to be a tiny error in one of the formula, namely in the "Confidence Intervals" section.

Now here's a link to the exact line I have issue with.

Namely, it seems the denominator in the probability calculations ought be "200", not "100". This is already mended in the nest formula, where we have i.e. the fraction "9999/200".