Cameron–Martin theorem explained

In mathematics, the Cameron–Martin theorem or Cameron–Martin formula (named after Robert Horton Cameron and W. T. Martin) is a theorem of measure theory that describes how abstract Wiener measure changes under translation by certain elements of the Cameron–Martin Hilbert space.

Motivation

\gamma ^{n} on n -dimensional Euclidean space R ^{n} is not translation-invariant. (In fact, there is a unique translation invariant Radon measure up to scale by Haar's theorem: the n -dimensional Lebesgue measure, denoted here dx .) Instead, a measurable subset A has Gaussian measure\gamma _{n(A)} = (1)/(( 2 \pi) ^{n/2}) \int _{A} \exp\left( - \tfrac 12 \langle x, x\rangle _{R ^{n}} \right)dx.

Here

\langle x,x\rangle _{R ^{n}} refers to the standard Euclidean dot product in R ^{n} . The Gaussian measure of the translation of A by a vector h \in R ^{n} is\begin{align} \gamma _{n(A-h)} & = (1)/(( 2 \pi) ^{n/2}) \int _{A} \exp\left( - \tfrac 12 \langle x - h, x - h\rangle _{R ^{n}} \right)dx\\[ 4 pt] & = (1)/(( 2 \pi) ^{n/2}) \int _{A} \exp\left( (2 \langle x, h\rangle _{R ^{n}} - \langle h, h\rangle _{R ^{n}})/(2) \right)\exp\left( - \tfrac 12 \langle x, x\rangle _{R ^{n}} \right)dx. \end{align}

So under translation through

h , the Gaussian measure scales by the distribution function appearing in the last display:\exp\left( (2 \langle x, h\rangle _{R ^{n}} - \langle h, h\rangle _{R ^{n}})/(2) \right) = \exp\left(\langle x, h\rangle _{R ^{n}} - \tfrac 12 \|h\| ^{2 \right).} _{R ^{n}}

The measure that associates to the set

A the number \gamma _{n(A} - h) is the pushforward measure, denoted (T _{h)} _{*} (\gamma ^{n)} . Here T _{h} : R ^{n} \to R ^{n} refers to the translation map: T _{h(x)} = x + h . The above calculation shows that the Radon–Nikodym derivative of the pushforward measure with respect to the original Gaussian measure is given by(d (T _{h)} _{*} (\gamma ^{n)})/(d \gamma ^{n}) (x) = \exp \left( \left \langle h, x \right \rangle _{R ^{n}} - \tfrac{ 1 }{ 2 } \| h \| ^{2} _{R ^{n}} \right).

The abstract Wiener measure

\gamma on a separable Banach space E , where i : H \to E is an abstract Wiener space, is also a "Gaussian measure" in a suitable sense. How does it change under translation? It turns out that a similar formula to the one above holds if we consider only translations by elements of the dense subspace i(H) \subseteq E .

Statement of the theorem

Let

i : H \to E be an abstract Wiener space with abstract Wiener measure \gamma : \operatorname{Borel}(E) \to [ 0 , 1 ] . For h \in H , define T _{h} : E \to E by T _{h(x)} = x + i(h) . Then (T _{h)} _{*(\gamma)} is equivalent to \gamma with Radon–Nikodym derivative(d (T _{h} ) _{*} (\gamma))/(d \gamma) (x) = \exp \left( \langle h, x \rangle ^{\sim} - \tfrac{ 1 }{ 2 } \| h \| ^{2} _{H} \right),

where

\langle h, x \rangle ^{\sim} = i(h) (x)

denotes the Paley–Wiener integral.

The Cameron–Martin formula is valid only for translations by elements of the dense subspace

i(H) \subseteq E , called Cameron–Martin space, and not by arbitrary elements of E . If the Cameron–Martin formula did hold for arbitrary translations, it would contradict the following result:

If

E is a separable Banach space and \mu is a locally finite Borel measure on E that is equivalent to its own push forward under any translation, then either E has finite dimension or \mu is the trivial (zero) measure. (See quasi-invariant measure.)

In fact,

\gamma is quasi-invariant under translation by an element v if and only if v \in i(H) . Vectors in i(H) are sometimes known as Cameron–Martin directions.

Integration by parts

The Cameron–Martin formula gives rise to an integration by parts formula on

E : if F : E \to R has bounded Fréchet derivative D F : E \to \operatorname{Lin}(E; R ) = E ^{*} , integrating the Cameron–Martin formula with respect to Wiener measure on both sides gives\int _{E} F(x + t i(h)) d \gamma (x) = \int _{E} F(x) \exp \left( t \langle h, x \rangle ^{\sim} - \tfrac{ 1 }{ 2 } t ^{2} \| h \| ^{2} _{H} \right) d \gamma (x)

for any

t \in R . Formally differentiating with respect to t and evaluating at t = 0 gives the integration by parts formula\int _{E} D F(x) (i(h)) d \gamma (x) = \int _{E} F(x) \langle h, x \rangle ^{\sim} d \gamma (x).

Comparison with the divergence theorem of vector calculus suggests

div [V _{h]} (x) = - \langle h, x \rangle ^{\sim,}

where

V _{h} : E \to E is the constant "vector field" V _{h(x)} = i(h) for all x \in E . The wish to consider more general vector fields and to think of stochastic integrals as "divergences" leads to the study of stochastic processes and the Malliavin calculus, and, in particular, the Clark–Ocone theorem and its associated integration by parts formula.

An application

Using Cameron–Martin theorem one may establish (See Liptser and Shiryayev 1977, p. 280) that for a

q x q symmetric non-negative definite matrix H(t) whose elements H _{j,} (t) are continuous and satisfy the condition\int ^{T} _{0} \sum _{j,k=1} ^{q} |H _{j,k} (t)|dt < inf ty,

it holds for a

q −dimensional Wiener process w(t) thatE \left[ \exp \left( - \int ^{T} _{0} w(t) ^{*H(t)w(t)} dt \right) \right] = \exp \left[ \tfrac{ 1 }{ 2 } \int ^{T} _{0} \operatorname{tr} (G(t)) dt \right],

where

G(t) is a q x q nonpositive definite matrix which is a unique solution of the matrix-valued Riccati differential equation(dG(t))/(dt) = 2 H(t) - G ^{2(t)}

with the boundary condition

G(T) = 0 .

In the special case of a one-dimensional Brownian motion where

H(t) =1 / 2 , the unique solution is G(t) = \tanh(t - T) , and we have the original formula as established by Cameron and Martin:$$E\backslash left[\backslash exp\backslash left(-\backslash tfrac12\backslash int\_0^T\; w(t)^2\backslash ,dt\backslash right)\backslash right]\; =\; \backslash frac.$$

References

• Cameron. R. H.. Martin. W. T.. 1944. Transformations of Wiener Integrals under Translations. . 45. 2. 386–396. 1969276. 10.2307/1969276.
• Book: Liptser . R. S. . Shiryayev . A. N. . Statistics of Random Processes I: General Theory . 1977 . Springer-Verlag . 3-540-90226-0 .