AI reasoning models settle decade-old problem in optimisation theory
On 14 August 2026, researchers Kenan Xu and Xiangfeng Wang published a paper on arXiv (arXiv:2608.14396) demonstrating how frontier artificial intelligence reasoning models resolved a ten-year-old open problem in mathematical optimisation theory. The study detailed how OpenAI Codex, driven by GPT-5.6 Sol, alongside Moonshot AI Kimi Code, powered by Kimi K3, successfully disproved a long-standing conjecture regarding the convergence of the three-block Alternating Direction Method of Multipliers, commonly known as ADMM. Operating under the guidance of human researchers, these advanced reasoning engines constructed formal algebraic counterexamples that proved direct three-block ADMM can fail to converge even under conditions previously believed to guarantee stability.
Why the result matters
ADMM is a widely used algorithm for solving large-scale optimisation problems, with applications across machine learning, signal processing and distributed computing. While the two-block version of the method has well-established convergence guarantees, the behaviour of the three-block extension had remained an open question for a decade. The counterexamples produced in the study close that gap, showing that the conjectured convergence conditions do not hold in general.
Human-guided AI at work
According to the paper, the human researchers set the problem framing and verified each candidate counterexample, while the reasoning models carried out the algebraic construction and iterative refinement. The authors describe the workflow as a collaboration in which the models proposed candidate structures and the researchers checked them against formal criteria, with every result independently verified before publication. The finding adds to a growing list of open mathematical questions resolved with the assistance of AI reasoning systems.
References and related sources
- Primary source: arxiv.org
- vibemathed.com
How iEnvi can help
iEnvi integrates technology and data-driven approaches into environmental consulting. We monitor AI and technology developments that affect how environmental professionals deliver services to clients.
This is an iEnvi Machete news summary. Prepared by iEnvi to summarise the source article for environmental professionals tracking AI, data, and technology developments that affect consulting and project delivery.
Published: 18 Aug 2026
Need advice on this topic? Speak to an iEnvi expert at info@ienvi.com.au or 1300 043 684, or contact us online.
Need advice on this issue? iEnvi provides practical, senior-led environmental consulting across contaminated land, remediation, ecology and environmental risk.
Contaminated land advice Remediation services Discuss your site Talk to iEnvi