LZMD : Lempel-Ziv Montecarlo Diffusion

Join the mailing list View on GitHub

Update:

I got rejected from YCombinator last week, ZFellows in August and got ghosted by VCs inbetween.

I took a gap year to build this. Now it's open-source lol. I'll add fileio and Decoder.c if people are interested.

Disclaimer

Data compression is a solved problem. No ✨ magical algorithm ✨ makes files infinitely small. Also, you can't compress random data.

How It Works

We built a custom diffusion model to simulate the weird change in volume when water and alcohol mix.

It's a navier-stokes solver whose solutions lead to smaller file sizes. This is NOT a data compressor. We don't find probabilities.
\

Technical Overview

A mixture of equal parts water and ethanol presents a volume 10% less than the sum of individual volumes. Ethanol molecules are smaller than water molecules. When mixed, small ethanol molecules occupy the spaces between large water molecules.


We built a mathematical model to simulate this mixture. We :

  • Split a compressed file into two parts - Left and Right analogous to Water and Ethanol.
  • Find large empty spaces in the Left part.
  • Find small integers in the Right part.
  • Place the small integers inside the large spaces.

How to use

This is work in progress. At the moment, Encoder.c displays the total number of bits saved. I'll add fileio and Decoder.c if people are interested

/*
Works best on ZIP, MP4 and JPEG files whose size is greater than 2 MB
Only tested on Ubuntu
Change filename in Encoder.c
*/

Run: gcc Encoder.c -lgmp -lm -o Encoder.o && ./Encoder.o
Output : "N bits saved!"

//
 
Downloads last month

-

Downloads are not tracked for this model. How to track
Inference API
Unable to determine this model's library. Check the docs .