Feldman–Hájek theorem

Theory in probability theory

In probability theory, the Feldman–Hájek theorem or Feldman–Hájek dichotomy is a fundamental result in the theory of Gaussian measures. It states that two Gaussian measures μ {\displaystyle \mu } and ν {\displaystyle \nu } on a locally convex space X {\displaystyle X} are either equivalent measures or else mutually singular:[1] there is no possibility of an intermediate situation in which, for example, μ {\displaystyle \mu } has a density with respect to ν {\displaystyle \nu } but not vice versa. In the special case that X {\displaystyle X} is a Hilbert space, it is possible to give an explicit description of the circumstances under which μ {\displaystyle \mu } and ν {\displaystyle \nu } are equivalent: writing m μ {\displaystyle m_{\mu }} and m ν {\displaystyle m_{\nu }} for the means of μ {\displaystyle \mu } and ν , {\displaystyle \nu ,} and C μ {\displaystyle C_{\mu }} and C ν {\displaystyle C_{\nu }} for their covariance operators, equivalence of μ {\displaystyle \mu } and ν {\displaystyle \nu } holds if and only if[2]

  • μ {\displaystyle \mu } and ν {\displaystyle \nu } have the same Cameron–Martin space H = C μ 1 / 2 ( X ) = C ν 1 / 2 ( X ) {\displaystyle H=C_{\mu }^{1/2}(X)=C_{\nu }^{1/2}(X)} ;
  • the difference in their means lies in this common Cameron–Martin space, i.e. m μ m ν H {\displaystyle m_{\mu }-m_{\nu }\in H} ; and
  • the operator ( C μ 1 / 2 C ν 1 / 2 ) ( C μ 1 / 2 C ν 1 / 2 ) I {\displaystyle (C_{\mu }^{-1/2}C_{\nu }^{1/2})(C_{\mu }^{-1/2}C_{\nu }^{1/2})^{\ast }-I} is a Hilbert–Schmidt operator on H ¯ . {\displaystyle {\bar {H}}.}

A simple consequence of the Feldman–Hájek theorem is that dilating a Gaussian measure on an infinite-dimensional Hilbert space X {\displaystyle X} (i.e. taking C ν = s C μ {\displaystyle C_{\nu }=sC_{\mu }} for some scale factor s 0 {\displaystyle s\geq 0} ) always yields two mutually singular Gaussian measures, except for the trivial dilation with s = 1 , {\displaystyle s=1,} since ( s 2 1 ) I {\displaystyle (s^{2}-1)I} is Hilbert–Schmidt only when s = 1. {\displaystyle s=1.}

See also

References

  1. ^ Bogachev, Vladimir I. (1998). Gaussian Measures. Mathematical Surveys and Monographs. Vol. 62. Providence, RI: American Mathematical Society. doi:10.1090/surv/062. ISBN 0-8218-1054-5. (See Theorem 2.7.2)
  2. ^ Da Prato, Giuseppe; Zabczyk, Jerzy (2014). Stochastic Equations in Infinite Dimensions. Encyclopedia of Mathematics and its Applications. Vol. 152 (Second ed.). Cambridge: Cambridge University Press. doi:10.1017/CBO9781107295513. ISBN 978-1-107-05584-1. (See Theorem 2.25)
  • v
  • t
  • e
Measure theory
Basic concepts
  • Absolute continuity of measures
  • Lebesgue integration
  • Lp spaces
  • Measure
  • Measure space
    • Probability space
  • Measurable space/function
SetsTypes of MeasuresParticular measuresMapsMain resultsOther results
For Lebesgue measure
Applications & related
  • v
  • t
  • e
Basic concepts
Derivatives
Measurability
Integrals
Results
Related
Functional calculus
Applications
  • v
  • t
  • e
Basic concepts
Main results
Maps
Types of sets
Set operations
Types of TVSs
  • Category