Generalized form of Price's theorem and its converse

J. Brown

IEEE Transactions on Information Theory · 1967 · 14 citations · 5 references

Concepts

Abstract

The case of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">n</tex> unity-variance random variables <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">x_{1}, x_{2},\cdots, x_{n}</tex> governed by the joint probability density <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">w(x_{1}, x_{2}, \cdots x_{n})</tex> is considered, where the density depends on the (normalized) cross-covariances <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">\rho_{ij} = E [(x_{i}- \bar{x}_{i})(x_{j} - \bar{x}_{j})]</tex> . It is shown that the condition <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">(^{\ast}) \frac{\delta}{\delta \rho_{ij}}\{E[f(x_{1}, x_{2}, \cdots , x_{n})\} = E|\frac{\delta^{2}}{\delta x_{i} \delta x_{j}}f(x_{1}, x_{2}, \cdots , x_{n})| \mbox{(i\neq j)}</tex> holds for an "arbitrary" function <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">f(x_{1}, x_{2}, \cdots , x_{n})</tex> of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">n</tex> variables if and only if the underlying density <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">w(x_{1}, x_{2}, \cdots , x_{n})</tex> is the usual <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">n</tex> -dimensional Gaussian density for correlated random variables. This result establishes a generalized form of Price's theorem in which: 1) the relevant condition <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">(^{\ast})</tex> subsumes Price's original condition; 2) the proof is accomplished without appeal to Laplace integral expansions; and 3) conditions referring to derivatives with respect to diagonal terms <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">\rho_{ij}</tex> are avoided, so that the unity variance assumption can be retained.

References

5