NAME¶

plfit - fits power-law distributions to empirical data

SYNOPSIS¶

plfit[OPTIONS] [infile ...]

DESCRIPTION¶

Reads data points from each given input file and fits a power-lawdistribution to them, one by one, according to the method ofClauset, Shalizi and Newman. If no input files are given, thestandard input will be processed.

This implementation uses the L-BFGS optimization method to findthe optimal alpha for a given xmin in the discrete case. If youwant to use the legacy brute-force approach originally publishedin the above paper, use the -a switch.

OPTIONS¶

-h

shows this help message

-v

shows version information

-a RANGE

use legacy brute-force search for the optimal alphawhen a discrete power-law distribution is fitted.RANGE must be in MIN:STEP:MAX format, the defaultis 1.5:0.01:3.5.

-b

brief (but easily parseable) output format

-c

force continuous fitting even when every sampleis an integer

-D VALUE

divide each sample in the input data by VALUE to preventunderflows when fitting discrete power-law distribution

-e EPS

try to provide a p-value with a precision of EPS whenthe p-value is calculated using the exact method. Thedefault is 0.01.

-f

use finite-size correction

-m XMIN

use XMIN as the minimum value for x instead of searchingfor the optimal value

-M

print the first four central moments (i.e. mean, variance,skewness and kurtosis) of the input data to helpassessing the shape of the pdf it may have come from.

-p METHOD

use METHOD to calculate the p-value. Must be one ofskip, approximate or exact. Default is skip.

-s SEED

use SEED to seed the random number generator