The Setting
At zero temperature, the Parisi formula for the Sherrington-Kirkpatrick model gives the ground-state energy as the minimum of the Parisi functional over order parameters . The functional is the value at the origin of a stochastic control problem, less a term linear in ; the gradient of the value function, weighted by , is the drift of a diffusion . Take a minimizer . What does minimality say about the second moment of that gradient along , at the times in the support of the Stieltjes measure of , and about the integral, from a time to , of the excess of that second moment over the time itself?
Order Parameters and the Parisi Functional
An order parameter is a function on the time interval that is nonnegative, nondecreasing, right-continuous and Lebesgue integrable; outside it is extended by zero.✦ A Stieltjes measure of is a measure on with for every .✦
A Brownian system on a measurable space is a probability measure on together with a real process , measurable at each time, that is a standard Brownian motion under ; is the expectation under .✦ For a time , a bounded control from is a real process with at every time and outcome, progressively measurable for the filtration generated by the increments , .✦
Fix a Brownian system . The value function of an order parameter is the function of time and position given by the supremum, over bounded controls from , of the expected control payoff: (1)✦ Its derivatives in are the value gradient ✦ and the value curvature .✦
The Parisi functional at is (2)✦ and a Parisi minimizer is an order parameter with for every order parameter .✦
The Diffusion and the Contact Potential
A bounded Lipschitz drift is a jointly measurable function of time and position for which there are constants with and for all , , .✦ Given a process , the Picard solution of is ✦, where is the limit as of the iterates✦ (3)✦
The driving Brownian motion of is the process ✦, where the progressive version of is the limit as of , the value of at the last point not after of the grid of mesh .✦
For the horizon is ✦, and the localized diffusion is the Picard solution, driven by , for the drift with its time clamped to : (4)✦ With ✦, the diffusion of is (5)✦ where is taken to be for ✦.
Along , the second moment of the value gradient is (6)✦ that of the value curvature is ✦, and the contact potential of is the function of a real number (7)✦
The Claim
Theorem 1. Let be a Brownian system on a measurable space , let be a Parisi minimizer for , and let be a Stieltjes measure of . At every time in the support of , (8)✦ and at every time the contact potential is nonnegative: (9)✦
In the paper, the result reads: for a minimizer , the contact potential is nonnegative and vanishes on the support of , and at every point of the support, and .
What Is Checked
The claim is stated over a Brownian system , an order parameter that minimizes the Parisi functional (2), a Stieltjes measure of and the times in its support, and the second moments of the value gradient and the value curvature along the diffusion (5), with the contact potential (7), all built from the value function (1). About these constructions, the library proves that each horizon lies in ✦ and that the clamped drift in (4) is a bounded Lipschitz drift with bound .✦