24.02.2013 Views

Optimality

Optimality

Optimality

SHOW MORE
SHOW LESS

Create successful ePaper yourself

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

164 D. M. Dabrowska<br />

so that in both cases it is enough to consider the determinants for ordered sequences<br />

s = (s1, . . . ,sm), s1 < s2 < . . . < sm of points in (0, τ] m .<br />

For any such sequence s, the matrix dm(s) has a simple pattern:<br />

⎛<br />

c(s1)<br />

⎜ c(s1)<br />

¯dm(s)<br />

⎜<br />

= ⎜ c(s1)<br />

⎜<br />

⎝ .<br />

c(s1)<br />

c(s2)<br />

c(s2)<br />

c(s1)<br />

c(s2)<br />

c(s3)<br />

. . .<br />

. . .<br />

. . .<br />

c(s1)<br />

c(s2)<br />

c(s3)<br />

.<br />

⎞<br />

⎟ .<br />

⎟<br />

⎠<br />

c(s1) c(s2) c(s3) . . . c(sm)<br />

We have ¯ dm(s) = A T mCm(s)Am where Cm(s) is a diagonal matrix of increments<br />

Cm(s) = diag [c(s1)−c(s0), c(s2)−c(s1), . . . c(sm)−c(sm−1)],<br />

(c(s0) = 0, s0 = 0) and Am is an upper triangular matrix<br />

⎛<br />

1<br />

⎜ 0<br />

⎜<br />

Am = ⎜ .<br />

⎜ .<br />

⎝ 0<br />

1<br />

1<br />

0<br />

. . . 1<br />

. . . 1<br />

. . . 1<br />

1<br />

1<br />

.<br />

1<br />

0 0 . . . 0 1<br />

To see this it is enough to note that Brownian motion forms a process with independent<br />

increments, and the kernel k(s, t) = c(s∧t) is the covariance function of a<br />

time transformed Brownian motion.<br />

Apparently, det Am = 1. Therefore<br />

and<br />

det ¯ dm(s) =<br />

⎞<br />

⎟ .<br />

⎟<br />

⎠<br />

m�<br />

[c(sj)−c(sj−1)]<br />

det ¯ Dm(t, u;s) = det ¯ dm(s)[c(t∧u)−Um(t;s)[ ¯ dm(s)] −1 Vm(s;u)]<br />

j=1<br />

= det ¯ dm(s)[c(t∧u)−Um(t;s)A −1<br />

m C −1<br />

m (s)(A T m) −1 Vm(s;u)].<br />

The inverse A−1 m is given by Jordan matrix<br />

⎛<br />

1 −1 0 . . . 0 0<br />

⎜ 0 1 −1 . . . 0 0<br />

A −1<br />

m =<br />

⎜<br />

⎝<br />

and a straightforward multiplication yields<br />

det ¯ Dm(t, u;s) = c(t∧u)<br />

−<br />

×<br />

.<br />

.<br />

.<br />

.<br />

0 0 0 . . . 1 −1<br />

0 0 0 . . . 0 1<br />

m�<br />

[c(sj)−c(sj−1)]<br />

j=1<br />

⎞<br />

⎟<br />

⎠<br />

m�<br />

[c(t∧si)−c(t∧si−1)][c(u∧si)−c(u∧si−1)]<br />

i=1<br />

m�<br />

j=1,j�=i<br />

[c(sj)−c(sj−1)].

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

Saved successfully!

Ooh no, something went wrong!