AxProf

A framework for accuracy profiling of randomized approximate algorithm implementations. See ICSE-2019-Paper.pdf for a full description of AxProf (to appear in ICSE 2019).

---

Directory structure

• AxProf contains the source of AxProf.
• AxProf/checkerGen contains the checker function generator component of AxProf.
• tutorial contains a tutorial script that uses AxProf.
• examples contains example scripts for testing some of the benchmarks from the conference paper.

---

Setup

First, install the required dependencies. Assuming your system is running Ubuntu 20.04 and Python 3.8, run the following commands:

sudo apt update
sudo apt install build-essential default-jdk python3-pip
sudo python3 -m pip install numpy==1.24.3 scipy==1.10.1 matplotlib==3.7.1 minepy==1.2.6

Note that newer versions of Python 3 (e.g., 3.11) may not support AxProf. Use pyenv if necessary to run AxProf in a Python 3.8 environment. Next, run the following commands from the root directory of this repository:

cd AxProf/checkerGen
make

---

Tutorial

A tutorial for using AxProf is available in tutorial/tutorial.py.

---

Example

An example script for testing ekzhu/datasketch is provided in examples/hllEkzhu.py. To run the script, you must first clone the datasketch repository. Run the following commands from the root directory of this repository:

cd examples
git clone https://github.com/ekzhu/datasketch.git

Now you can run examples/hllEkzhu.py to test the library.