Issue tracker for Tutorial 01

[jngrad]

List of things to fix based on feedback from the summer school.

• matplotlib plots
  • should be instantiated with fig = plt.figure(figsize=(10, 8)) instead of fig = plt.figure() to be more readable by taking the entire width of the HTML div
  • should end with plt.show() to avoid printing the return value of the last matplotlib method
• scipy
  • should be imported with import scipy.optimize in the cell where it is used for the first time, i.e. the def autocor() cell
• autocorrelation function
  • the function gives a different result than what is calculated in the core with the Correlator accumulator, because the normalization doesn't consider that for larger tau values, the sample size is smaller and the variance higher
  • numpy and scipy are known to be tricky to use for scripting autocorrelation (de Buyl 2018, DOI:10.21105/joss.00877)
  • the long-term solution here would be to install tidynamics in the docker image
• style
  • apply PEP8 formatting
  • assert don't need round brackets

[RudolfWeeber]

@jgrad @schlaicha

[jngrad]

Also, the tutorial implements the cyclic autocorrelation function:

import numpy as np
import tidynamics
import matplotlib.pyplot as plt

def autocor(x):
    x = np.asarray(x)
    mean = x.mean()
    var = np.var(x)
    xp = x-mean
    corr = np.correlate(xp, xp, 'full')[len(x)-1:]/var/len(x)
    return corr

x = np.arange(256)
y = np.sin(x * np.pi/16) / (x+1)**2

acf_td = tidynamics.acf(y) / np.var(y)
acf_tut = autocor(y) * len(x) / (len(x) - np.arange(len(x)))
print(np.sum(np.abs(acf_td - acf_tut)))

plt.plot(acf_td, label='tidynamics')
plt.plot(acf_tut, label='tutorial')
plt.legend()
plt.show()

[jngrad]

* `assert` don't need round brackets

Adding round brackets can lead to subtle bugs:

delete_precious_data = False
assert(delete_precious_data == True, "I can't delete data")  # tuple evaluates to true
cleanup_data()  # goodbye precious data

In python 3.6+ this construction prints SyntaxWarning: assertion is always true, perhaps remove parentheses? without halting the flow of the script. It also cannot be caught by a try/except without first importing module warning and writing some boilerplate code (see https://stackoverflow.com/a/15934081). In batch jobs submitted to a scheduler, the warning could be hidden deep in the logs.