---
title: "UMD Team Solves 35-Year Math Puzzle That Could Unlock Faster, Cooler AI Chips"
date: 2025-05-12T05:30:00-04:00
author: University of Maryland
canonical_url: "https://today.umd.edu/umd-team-solves-35-year-math-puzzle-that-could-unlock-faster-cooler-ai-chips"
section: Articles
---
# UMD Team Solves 35-Year Math Puzzle That Could Unlock Faster, Cooler AI Chips

*May 12, 2025* — by [Aleena Haroon M.P.P. ’25](/author/aleena-haroon-m-p-p-25)


> Steiner Forest Problem Stumped Scientists for Nearly Four Decades

*Math hero 1920x1080 — The solution to a longstanding math problem by a UMD team could allow for more powerful and efficient computer processors.*

A University of Maryland faculty-student team has cracked a mathematical quandary that stumped scientists for more than 35 years, but it’s more than just a clever solution to a puzzle: The discovery could boost the power and energy efficiency of the chips used to run computers and power AI systems.

Led by Mohammad T. Hajiaghayi, the Jack and Rita G. Minker Professor of Computer Science, the group recently published [a](https://obj.umiacs.umd.edu/126-proof/2504.11398v1.pdf)[ 126-page paper](https://obj.umiacs.umd.edu/126-proof/2504.11398v1.pdf) advancing the solution to the Steiner Forest problem—a network optimization challenge seeking a minimum-cost subgraph that connects a set of terminal pairs in each graph—with a recent *Journal of the ACM* [paper ](https://obj.umiacs.umd.edu/steiner-forest/3722551.pdf)on a similar subject serving as a first step toward this result.

So what does that mean? Imagine you want to set up several lights in your living room, but to do so, you need to figure out the best way to connect the lights using electric wire, possibly through some shared junctions. What if there was a mathematical formula to determine the minimum amount of wire you needed to buy, so you could avoid a mess of cables while also saving cash?

This is where the Steiner Forest problem comes in. In mathematics and computer science, the use of network optimization algorithms can often determine the shortest way to connect a set of points in a network. By applying a Steiner Forest solution, you can plan your lighting setup to be cost-effective and clean.

But the Steiner Forest problem has many more applications than helping with home décor. Modern computer and AI chips contain millions of tiny transistors that need to be connected. Hajiaghayi and his team’s new Steiner Forest algorithm can help tech giants like Nvidia and Google design the shortest wiring paths between these components, slashing costs and energy usage. In the era of artificial intelligence’s vast and growing hunger for computing power, this can allow these and other companies to create faster models that consume less.


*group of four gathers around whiteboard — Team members include, from left, Ali Ahmadi, Mohammad Mahdavi, Mohammad T. Hajiaghayi and Peyman Jabbarzade. Not pictured, Iman Gholami.*




**Topics:** [Research](https://today.umd.edu/tags/research)


**Tags:** [Artificial Intelligence](https://today.umd.edu/topic/artificial-intelligence), [Computer Science](https://today.umd.edu/topic/computer-science), [Mathematics](https://today.umd.edu/topic/mathematics), [Research](https://today.umd.edu/topic/research), [Student Experience](https://today.umd.edu/topic/student-experience)


**Units:** [College of Computer, Mathematical, and Natural Sciences](https://today.umd.edu/topic/college-computer-mathematical-and-natural-sciences), [University of Maryland Institute for Advanced Computer Studies](https://today.umd.edu/topic/university-of-maryland-institute-for-advanced-computer-studies)