Miscpy (miscellaneous python code)

Mostly self-contained, tested and documented python functions for scientific computing. In case you don't know: you can import a Python file as a module by just placing the file in the same directory as your script.

Files

• argminrelmin.py: index of the minimum local minimum.

• breaklines.py: break a string in lines before or after certain characters.

• cartprod.py: cartesian product of numpy arrays. Useful to prepare an array of parameter values for a grid search on a slow procedure.

• ccdelta.py: compute the cross-correlation of dicrete points with a continuous function.

• clusterargsort.py: filter away values which are close to an higher value in a signal.

• coinc.py: simulates poissonian digital signals and counts coincidences, i.e., how many times it happens they are all 1 simultaneously.

• colormap.py: make a perceptually uniform colormap.

• errorsummary.py: takes a ufloat variable (from the uncertainties module) and separate its error components by tag. Useful to look at different contributions to the error.

• histogram.py: version of numpy.histogram that also computes uncertainties.

• ising.py: diagonalize the 1D quantum Ising hamiltonian.

• maxprominencedip.py: find local minima sorted by higher negative prominence. Fully vectorized.

• mcmc.py: functions for blocking and bootstrapping.

• meanmedian.py: quick less discrete median for ADC waveforms.

• neff.py: compute the effective sample size for an autocorrelated sequence, defined as the asymptotic ratio between the variance and variance of the sample mean.

• npzload.py: superclass to add serialization from/to numpy archives to an object (alternatives are joblib and pickle protocol 5, but they can execute arbitrary code).

• rhat.py: compute the Gelman-Rubin split-$\hat R$ statistics for assessing convergence of Markov chains.

• runsliced.py: minimal batching.

• textbox.py: put a legend-like box with arbitrary text on a matplotlib plot.

• textmatrix.py: table manipulation object. Can print in LaTeX format.

• thin.py: decimate an array almost evenly with randomized disuniformity.

• uevuev.py: compute the variance of the variance.

• uformat.py: format numbers with uncertainties.

• updowncast.py: recursively cast fields of numpy data type to longer/ shorter equivalent types. Useful for saving data with shorter types after checking it's within bounds.

• weighted_mean.py: compute a weighted mean with ufloats (correctly takes into account covariance).