The Setting
Let be a finite set of coordinates and a measurable set of finite volume. In every coordinate run a stationary Gaussian autoregressive chain with correlation between consecutive steps, the coordinates independent, and ask for the probability that the chain stays in at each of the steps . Time is discrete. A survival rate of is a real number for which some real gives the lower bound , for every , on the probability that the chains stay in at each of the steps . The question is what such a rate says about the Gaussian heat flow on : how much operator norm a Gaussian heat kernel keeps once it is cut down to , and what becomes of this when is dilated.
The Gaussian Chain
The autoregressive row of a real number is the array of reals , indexed by a step and a noise index , defined by recursion on the step, with the Kronecker delta: (1)✦ For , is the weight of the noise in the state of the stationary chain , , driven by independent standard Gaussians .
For correlations , a horizon and a step , the autoregressive evaluation is the linear map from to defined by (2)✦ Under the standard Gaussian measure on noise arrays , the vectors are the steps to of the chains with correlations , one in each coordinate, and the survival probability of up to step is (3)
Kernels and Heat Operators
The product Gaussian density of nonnegative weights is the function on that multiplies the centered Gaussian densities with variances : (4)✦
A bounded kernel on a measurable space is a measurable function on with for a constant , its bound.✦ The difference kernel of a measurable function on a Euclidean space with is the bounded kernel with bound ✦, and the Gaussian product kernel of is the difference kernel of ✦.
The kernel operator of a bounded kernel on , for a finite measure on , is the bounded linear operator on the complex space defined by (5)✦ For a set of finite volume, is the kernel operator of the Gaussian product kernel of for Lebesgue measure on , acting on by . Norms of operators are operator norms on .
For a measurable complex function on a measure space with almost everywhere, the multiplication operator is the bounded linear operator on ✦; the indicator mask of a measurable set is the multiplication operator of its indicator function✦. The Fourier multiplier operator of a measurable real function on a Euclidean space with is on the complex space , with the unitary Fourier transform.✦ The heat operator of a measurable function on is the Fourier multiplier operator of ✦, and the compressed heat operator of a measurable set is ✦.
The Gaussian kernel symbol of weights is the function of a frequency (6)✦ A measurable set satisfies the diagonal heat bound with constant if, for every , its compressed heat operator for satisfies (7)✦
The Claim
Theorem 1. Let , put , and let be a measurable set of finite volume whose survival probabilities satisfy for every , for some reals and . Then (8)✦ and for every such that the dilation has finite volume, (9)✦ Second, let , let be real, and let be a measurable set of finite volume whose dilation is measurable of finite volume. Suppose that for every there is a real with for every , where . Then satisfies the diagonal heat bound with constant : for every , (10)✦
In the paper, the result reads: for a bounded nonempty open convex set , a positive definite diagonal matrix and every , the stationary Ornstein-Uhlenbeck process satisfies , and , where is the diagonal Dirichlet energy of and its Ornstein-Uhlenbeck energy.
The box answers it in discrete time, with the chain of correlations in place of and the diagonal heat bound in place of a bound on , the summand in standing for .
What Is Checked
The claim is stated over the autoregressive rows (1) and evaluations (2), the survival probabilities (3) under the standard Gaussian measure, the kernel operators of the Gaussian product kernel on and , and the diagonal heat bound (7), built from the indicator mask, the heat operator and the symbol .
About these constructions the library proves for the product Gaussian density and for the symbol✦. For a bounded kernel with bound and a finite measure , it shows that the integral (5) of every is again in ✦, additive in ✦, with for the constant function ✦. For a measurable with almost everywhere, the product of every lies in ✦, with ✦.