Skip to content

OpenAI

GPT-6 Astra co-built a candidate lattice Standard Model

4 min read

Written by an AI model from the linked sources and checked against them by a second model. How we work

All about: OpenAI models and products →
Image: openai.com

Physicist Nathaniel Craig says OpenAI’s GPT-6 Astra took the lead in proving his October proposal for a lattice definition of the full Standard Model, one that does not rely on treating the theory’s interactions as small corrections, he wrote in a November 15 post on OpenAI’s site. It is a candidate that holds on a finite grid and under a bound on field strengths, and Craig himself calls it “far more an answer in principle than an answer in practice,” so nothing here is ready for lattice simulations yet, and other experts still have to decide whether it has the properties the field needs.

The obstacle has always been the weak force. It treats left- and right-handed fermions differently, which makes the Standard Model a chiral theory. Put a left-handed fermion on a lattice (a grid standing in for spacetime) with the most natural rules, and a right-handed partner nobody asked for shows up. A 1981 theorem by Holger Bech Nielsen and Masao Ninomiya shows that this is built in: under general assumptions such as locality and translational symmetry, any lattice formulation of chiral fermions ends up with equal numbers of left- and right-handed ones. The strong force and electromagnetism treat both hands alike, so lattice methods could work out the proton’s mass from first principles while a complete definition of the full Standard Model stayed missing.

The missing phase rule

Paul Ginsparg and Nobel laureate Ken Wilson worked out the relevant lattice rules decades ago. Rajamani Narayanan and Herbert Neuberger then found overlap fermions, a lattice definition of fermions that keeps only the wanted hand and drops the doublers. Martin Lüscher showed how to tell left- from right-handed overlap fermions, and Yoshio Kikukawa and collaborators explored using them for the weak interactions.

What stayed open was a phase. Vary the force fields on the grid in small steps until they return to where they started, and the left-handed fermions can come back with a different phase than they left with. A consistent theory needs a rule that compensates for that shift around every loop, and the shift depends on what the forces are doing at every grid point. Lüscher wrote the rule for simpler theories such as a chiral version of electromagnetism. For the full Standard Model it was open for more than 25 years.

Craig’s strategy carries the force fields from the grid where the theory is defined onto a succession of ever finer grids. Comparing the allowed fermion states on two neighboring grids gives one piece of the phase correction, and adding the pieces up gives the rule on the original grid. The proof shows that the rule gives the right correction around every allowed loop, that the correction for a change at one spot depends mainly on nearby fields, and that different mathematical descriptions of the same fields give the same answer. All of this holds at finite lattice spacing, under a bound on field strengths at the scale of the grid.

Who did which part

Craig says he had the strategy but had not been able to prove it covered every case it needed to. Over several rounds of work, Astra suggested how to carry a given set of force fields onto the finer grids, showed that the comparisons between grids converge to a definite answer and took the lead in developing the detailed proof. Lattice field theorists Maarten Golterman, Andreas Kronfeld and Yigal Shamir looked over Astra’s proof and asked probing questions that, Craig says, helped him “build confidence in the result.”

He still writes that “scientific dialogue among experts is required to decide if this prescription has all the properties we desire.”

Where the lattice stops

Defining the interacting theory in continuous spacetime is a separate question, and the construction leaves it open. Craig points to OpenAI’s recent proof that Yang-Mills theory exists (the strong and weak forces are examples of such theories) as the missing piece. He says it let OpenAI prove the existence of the simplest unified Yang-Mills theory that includes the Standard Model’s gauge fields and fermions, starting from overlap fermions on the lattice and taking the continuum limit. That continuum step leans on an OpenAI physics result that, in late October, the company was still keeping unreleased.

Simulations are further off. Craig says much work lies ahead before the method is useful for realistic lattice calculations. A separate arXiv paper submitted November 4 notes that no chiral gauge theory has yet been simulated nonperturbatively.

Craig’s next check is reflection positivity, a condition that connects calculations in the lattice model to a quantum theory in which probabilities are conserved. “Right now I’m thinking a lot about whether this construction satisfies reflection positivity at finite lattice spacing,” he writes.

No comments yet. Be the first.

Join the conversation

Sign in to leave a comment. It appears after moderation.