Fitting sharply truncated power law...

[alekseygenerozov]

Hi,

I was trying out the powerlaw package. Thanks for publishing it!

I was wondering – is the package capable of fitting a sharply truncated power law? (i.e. one that abruptly cuts off at some point (rather than an exponential turnover). From experiments I had a hard time accomplishing this.

For example, I was trying to fit the attached data which follows an r^-1.5 power law between 0.1 and 10.

Thank you for your help!

pow1.5_example.zip

Aleksey

[keflavich]

The Truncated_Power_Law (

powerlaw/powerlaw.py

Line 1395 in a1abb89

class Truncated_Power_Law(Distribution):

&

powerlaw/powerlaw.py

Line 2787 in a1abb89

def truncated_power_law_likelihoods(data, alpha, Lambda, xmin, xmax=False, discrete=False):

) distribution is likely what you're looking for. You'll have to dig to see how it actually works. There are other forms of truncated power law that implement an exponential cutoff at a fixed x value, like the Schechter function, which aren't implemented.

[jeffalstott]

Truncated_Power_Law is for exponential cutoffs. xmax is for sharp cutoffs. Note that at one point in time 'xmax' had a bug for continuous data (but worked for discrete). I don't know if a fix was implemented:
#89