Axplorer, the AI that tries to bring mathematical research from the supercomputer to the laptop
The tool developed by Axiom Math promises to cut time and costs in solving complex graph theory problems
Called Axplorer (but the initial A is pronounced like an E), it is an artificial intelligence-assisted mathematical research tool and promises to make it possible to obtain results on a normal laptop that, until now, required the use of a supercomputer.
To explain how this is possible, Axiom Math, the company that developed it, gives a practical example, relating to Turán's 4-loop problem. In a nutshell, it involves answering this question: if you want to connect the points of a network together, what is the maximum number of connections you can make without drawing a square?
The question, which falls under the branch of mathematics that is graph theory, is not only theoretical. It is in fact applied to analysing complex networks, such as social networks. In a paper published in November 2024, a group of researchers proposed tackling the problem through a methodology called PatternBoost. In practice, a series of random solutions were generated, artificial intelligence was used to select the best ones, on the basis of which the pattern was trained in a process of continuous refinement.
Whereas PatternBoost used to describe each graph by listing all possible node pairs, including unconnected ones, Axplorer describes each graph by listing only those links that actually exist. A change in approach that considerably reduces the time and computing power required: whereas previously 116 million attempts were required to arrive at an optimal solution, with the tool developed by Axiom Math, 2.6 million are sufficient and the process is completed in two and a half hours. And, the company points out, it only takes a single GPU and the whole operation costs three dollars. Put another way, where before you needed a supercomputer, now all you need is a laptop and some change.
Axplorer promises in other words to reduce the costs of mathematical research. In announcing the launch of this tool, they even go so far as to speak of democratisation. A process in which an Italian mind is also present: in fact, Alberto Alfarano, who trained at the Milan Polytechnic, heads the company's technical staff.
