Thanks to visit codestin.com
Credit goes to github.com

Skip to content
You must be logged in to sponsor sethuiyer

Become a sponsor to Sethu Iyer

@sethuiyer

Sethu Iyer

sethuiyer

I develop physics-informed optimization engines that treat solution spaces as thermodynamic systems.

My flagship project, BAHA (GitHub, Zenodo), detects thermodynamic fractures in optimization landscapes and uses complex-plane branch enumeration (via Lambert-W) to escape local minima—achieving 88% success across 26 hard problems, including solving Ramsey R(5,5,5) at N=52.
If your problem has phase transitions (SAT near threshold, constraint satisfaction, graph coloring), BAHA likely outperforms SA/GA by 2–10×—with zero tuning.

Open-source core. Enterprise-ready extensions available.

Another flagship product is Navokoj, which is powerful enough to be termed as general-purpose engine for finding coherent structure inside astronomically large discrete spaces.

It uses significantly more mathematics and engineering than in the open source versions to solve large scale max-SAT, max-QSAT, XOR-SAT with high degree accuracy. Proprietary solver explores local basin keeping the global geometry in mind.

Your support directly funds the research and development of:

Navokoj: The physics-inspired SAT solver that treats constraints as flowing geometry.
BAHA: High-performance C++/Python engines for discrete combinatorial optimization.

and more. Please refer to https://shunyabar.lovable.app for the open source research. https://navokoj.shunyabar.foo/ for the API. Generous free tier.

@sethuiyer

Reaching this goal would enable me to scale navokoj to more users

Featured work

  1. sethuiyer/casimir-sat-solver

    when boolean logic meets quantum mechanics

    HTML 1
  2. sethuiyer/baha

    Simulated annealing which uses thermodynamics of the landscape to escape local minima and works amazingly on discrete combinatorial problems

    C++ 2
  3. sethuiyer/shunyabar.lua

    A zero-dependency Lua module solving hard combinatorial and physics problems

    Lua 1
  4. sethuiyer/DAPS

    Dimensionally adaptive prime search for discontinuous, non-smooth, and multi-modal landscapes where traditional methods might fail.

    HTML

Learn more about sponsoring developers and organizations.

Report abuse

0% towards $90 per month goal

Be the first to sponsor this goal!

Select a tier

$ one time

Choose a custom amount.

$50 one time

Select

If you used the tools and it saved you 500$!

$200 one time

Select

If you are a researcher or an engineer using my tools to slash compute costs and it saved you more than 2000$ in compute costs!

$1,000 one time

Select

If you used the tools and it saved you 10k USD$!