Skip to main content


in this section

John Nolan

Available on CRAN



Jingjing Zou

An interactive online application to demonstrate methods in the paper “Extreme Value Analysis Without the Largest Values: What Can Be Done? (Zou, Samorodnitsky and Davis, 2017). Users can either choose from existing real data examples (earthquake fatalities, Google+) or upload their own data. Users can artificially remove a number of extreme values from the data and compare estimation results before and after the removal. This application allows real-time computation based on user inputs and visualizations of results with interactive plots.


Phyllis Wan

Simulation of preferential attachment networks

This is an implementation of the generalized scale-free network model described in Section 5 of “Fitting the linear preferential attachment model (Wan et al.)

The two simulation programs, netSim1 and netSim2, correspond to different starting values. netSim1 starts with one node 0, with a self loop, 0 -> 0. netSim2 starts with two nodes 0 and 1, and a connecting edge 0 -> 1. The computation complexity is O(n).


Parameter estimation in a preferential attachment model:
This is an implementation of the parameter estimation methods for linear preferential attachment model described in Fitting the linear preferential attachment model in


The R-scripts netfnsMLE.R and netfnsSnap.R correspond to estimation algorithms for the full network (MLE) and a snapshot of the network, where the details can be found in Section 3 and 4 of Wan et al. respectively.


UMass group (Towsley, Gong, Atwood)


Sidney Resnick

R-functions for univariate and multivariate heavy tail analysis

Compendium of R-functions used in both univariate and multivariate heavy tail analysis. Some are outlined in the book Heavy-Tail Phenomena: Probabilistic and Statistical Modeling. Springer-Verlag, New York, in the appendix starting on p. 364. Use at your own risk.


Zhi-Li Zhang

Software for graph feature discovery.