01.07.2013 Views

Random Processes in Hyperbolic Spaces Hyperbolic Brownian ...

Random Processes in Hyperbolic Spaces Hyperbolic Brownian ...

Random Processes in Hyperbolic Spaces Hyperbolic Brownian ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

41 <strong>Hyperbolic</strong> <strong>Brownian</strong> Motion<br />

and apply<strong>in</strong>g the <strong>in</strong>verse Fourier transform to both members, we get<br />

∞<br />

−∞<br />

e iλx<br />

2 1 + ξ<br />

f<br />

2<br />

′′ <br />

(ξ) + ν + µ + 3<br />

<br />

ξ f<br />

2<br />

′ <br />

(ξ) + µ + 1<br />

<br />

f(ξ) dξ = 0.<br />

2<br />

S<strong>in</strong>ce, for k = 0, 1, 2, the function xkf (k) (x) is <strong>in</strong>tegrable, we have that the function λku (k) (λ) exists,<br />

is cont<strong>in</strong>uous <strong>in</strong> R and vanishes at <strong>in</strong>f<strong>in</strong>ity. In particular u(λ) is twice cont<strong>in</strong>uously differentiable<br />

outside 0. We have<br />

∞<br />

e<br />

−∞<br />

iλx f (k) (x)dx = (−iλ) k u(λ) = (−1) k u (k) (λ),<br />

∞<br />

−∞<br />

∞<br />

−∞<br />

e iλx xf ′ (x)dx = −i ∂<br />

∞<br />

e<br />

∂λ −∞<br />

iλx f ′ (x)dx = −i ∂<br />

∂λ [−iλu(λ)] = −u(λ) − λu′ (λ),<br />

e iλx x 2 f ′′ (x)dx = − ∂2<br />

∂λ2 ∞<br />

e<br />

−∞<br />

iλx f ′′ (x)dx = − ∂2<br />

∂λ2 [(iλ)2u(λ)] = 2u(λ) + 4λu ′ (λ) + λ 2 u ′′ (λ).<br />

It follows that u(λ) is <strong>in</strong> the kernel of the operator N given by<br />

Nu(λ) = λ2<br />

2 u′′ <br />

(λ) − µ − 1<br />

<br />

λu<br />

2<br />

′ <br />

2 λ<br />

(λ) − + iνλ u(λ).<br />

2<br />

Tak<strong>in</strong>g <strong>in</strong>to account the third condition <strong>in</strong> (3.28) we have u(0) = ∞<br />

−∞ pz(ξ)dξ = 1. With the<br />

change of variable u(λ) = e −λ v(2λ) and ω = 2λ we get<br />

ωv ′′ (ω) + (1 − 2µ − ω)v ′ <br />

1<br />

(ω) − − µ + iν v(ω) = 0.<br />

2<br />

S<strong>in</strong>ce the confluent hypergeometric differential equation ωv ′′ (ω) + (b − ω)v ′ (ω) − av(ω) = 0 has<br />

solution<br />

Φ(a, b; ω) = 21−b Γ(1 − a)e ω<br />

2<br />

π<br />

π<br />

2<br />

0<br />

<br />

ω<br />

<br />

cos tan θ + (2a − b)θ cos<br />

2 −b θdθ<br />

where Φ(a, b; ω) is the confluent hypergeometric function of the second k<strong>in</strong>d with Re{b} < 1 and a<br />

not a positive <strong>in</strong>teger (see Gradshteyn and Ryzhik (16) formula 9.216.1), the characteristic function<br />

u(λ) we are look<strong>in</strong>g for is given by<br />

u(λ) = e −λ v(2λ) = Ke −λ <br />

1<br />

Φ − µ + iν, 1 − 2µ; 2λ .<br />

2<br />

for some real constant K. In particular we note that Φ(a, b; ω) has a f<strong>in</strong>ite non-zero limit for<br />

ω → 0 + ω − and that limω→∞ e 2 Φ(a, b; ω) = 0. To obta<strong>in</strong> the value of the constant K we observe<br />

that<br />

<br />

1<br />

1 = u(0) = KΦ − µ + iν, 1 − 2µ; 0 = K<br />

2 22µ<br />

π Γ<br />

π<br />

1<br />

2<br />

+ µ − iν cos(2iνθ) cos<br />

2 0<br />

2µ−1 θ dθ.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!