RANDOM

R-based Analyzer for Numerical Domains

Last version: 0.3.1 - 19/6/2012

Authors: Gianluca Amato, Maurizio Parton and Francesca Scozzari

Random is a static analyzer of programs, based on the theory of abstract interpretation. It can analyze simple imperative R programs (with the standard constructs while, for, if-then-else) with 4 abstract domains: intervals, template parallelotopes, a combination of intervals and template parallelotopes and, finally, general (non-template) parallelotopes. Template parallelotopes are a family of new abstract domains recently introduced in “Deriving numerical abstract domains via Principal Component Analysis”, SAS 2010 and “Discovering invariants via Simple Component Analysis ”, JSC 2011.

For a brief presentation of the tool Random you may read

1. “A tool which mines partial execution traces to improve static analysis”, RV 2010.
2. “Random: R-based Analyzer for Numerical DOMains”, LPAR 2012.

Download the analyzer: random-0.3.1.tgz.

The analyzer has been tested on Ubuntu Linux, but it should work on any Linux distribution.

You need:

• R - the interactive environment with the additional packages RGtk2 and CairoDevice

  (on Ubuntu install the packages r-recommended, r-cran-rgtk2, r-cran-cairodevice)

• GNU Multiple Precision Arithmetic Library with the development library (on Ubuntu install libgmp, libgmp-dev)

• gcc – or another equivalent C compiler

How to start:

• Download random-0.3.1.tgz into a folder and unpack it

• run 'R' from the folder where Random has been installed

• from the R prompt, type:

  • install.packages("gmp")
  • install.packages("ieeeround")
  • install.packages("fastICA")
  • install.packages("proto")
  This is only needed the first time you execute Random, in order to install required packages. Please note that the you need at least version 0.5.1 of the gmp package.

• from the R prompt, type source("random.R").

Random is a prototype, which is currently under development at University of Chieti-Pescara, Italy. For any information, you may contact g.amato@unich.it and scozzari@sci.unich.it

Disclaimer: the analyzer is distribuited without any warrant of any kind, under the terms of the GNU GPLv3.

Older versions of Random are also available.