This is a Python implementation of a radix packing compression technique which we refer to as Bignum. We show that Bignum is similar to and at times better than prior published state of the art integer compression techniques in terms of bits per integer.

This code is written for Python 2.7, and tested only on Unix-like operating systems, specifically OS X, and Ubuntu.

The basic command to run a comparitive study of different compression techniques:

./main.py -i $target_name.csv 

Since the current work is focused on polygons from DHS, the format of the csv file is based on the data file found here WEA polygons.

Aggregated results after the execution of main.py can be found under the results folder in csv files.

Plots can be generated post running the main.py by:

./comparison_plots.py

All plots would be stored under the results folder.

For comparison purposes, we considered TurboPFor (cloned on October 25, 2015) executed using:

./icbench $input_file -f1

Since the above takes a long time to run, you can view the results for each of polygon under the folder results/delta_min_state_of_art and results/delta_state_of_art. Each filename corresponds to an identification between 1 & 11,370.

Separate file for each polygon as input to icbench can be taken from folders delta_min_polygons and delta_polygons. A file for each polygon is generated after running main.py.

• Poster
• Paper official
• Paper arXiv