STAB ==== Simulation and estimation of stable scalar variables. TUTORIAL -------- # install !pip install pystab # import import stab # generate one million normal samples x = stab.stables(2, 0, 10**6) # estimate the four parameters from the vector x stab.fit_quantiles(x) > (1.9972697816715559, 0.018033003626273775, 0.9993768437879497, -0.000605309795968803) # generate one million Lévy samples x = stab.stables(0.5, 1, 10**6) # estimate the parameters # (notice that the shift parameter is imprecise due to heavy tails) stab.fit_quantiles(x) > (0.5000056240198412, 0.9998393548232919, 0.9901156603367183, 1.0144504595708657) DOCUMENTATION ------------- # core functions stable(α, β): # Return a S(α,β,1,0) pseudorandom sample, α∈[2,0[, β∈[-1,1] stables(α, β): # Return a vector with n stable samples i.i.d. S(α,β,1,0) fit_quantiles(x): # Return a tuple (α,β,σ,μ) with the estimated 4 parameters # convenience functions uniform(): # Return a U(0,1) pseudorandom sample normal(): # Return a N(0,1) pseudorandom sample stable4(α,β,c,μ): # Return a S(α,β,c,μ) pseudorandom sample uniforms(n): # Return vector with n pseudorandom samples i.i.d. U(0,1) normals(n): # Return vector with n pseudorandom samples i.i.d. N(0,1) stables4(α,β,c,μ,n): # Return vector with n stable samples i.i.d. S(α,β,c,μ) ACKNOWLEDGEMENT --------------- Reference: https://en.wikipedia.org/wiki/Stable_distribution The simulation algorithm is that of Chambers-Mallows-Struck and Weron The estimation algorithm is McCulloch's quantile method, using the GPL implementation in C by Royuela. - McCulloch's original implemantation in GAUSS: https://www.asc.ohio-state.edu/mcculloch.2/programs/STAB - Royuela's website: http://www.lpi.tel.uva.es/~jroyval/ RELEASING --------- To update the pypi hub, you need an account on pypi, and then run the following commands: cd pypi # update version number on file setup.py # update version number on file stab.py python setup.py sdist bdist_wheel python -m twine upload dist/pystab-X.tar.gz rm -rf build dist pystab.egg-info