28.05.2013 Views

Etude des marchés d'assurance non-vie à l'aide d'équilibres de ...

Etude des marchés d'assurance non-vie à l'aide d'équilibres de ...

Etude des marchés d'assurance non-vie à l'aide d'équilibres de ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

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

tel-00703797, version 2 - 7 Jun 2012<br />

Chapitre 4. Asymptotiques <strong>de</strong> la probabilité <strong>de</strong> ruine<br />

This result is in line with Theorem 7 of Chaudry and Zubair (1994) or Theorem 3.1 of Chaudry<br />

and Zubair (2002). It is closely related to the generalized error function<br />

√<br />

π<br />

erfc(x; b) =<br />

4 e2√ <br />

b<br />

e 2√ √ <br />

b b<br />

erfc x + + e<br />

x<br />

−2√ √ <br />

b b<br />

erfc x − .<br />

x<br />

If x = 0, we get<br />

J(a, b, x) =<br />

√ π<br />

2 √ a e−2√ ab = J(a, b).<br />

If b = −1, then it is equivalent as changing the occurence of √ b by the imaginary number i.<br />

We get<br />

√<br />

π<br />

J(a, −1, x) =<br />

4 √ <br />

e<br />

a<br />

2i√ <br />

a √ax i<br />

erfc + + e<br />

x<br />

−2i√ <br />

a √ax i<br />

erfc − .<br />

x<br />

This number is of type z1z2 + ¯z1¯z2, where z1 = e2i√a and z2 = erfc √ ax + i<br />

<br />

x . It is easy to<br />

check that<br />

z1z2 + ¯z1¯z2 = 2|z1z2| cos(arg(z1) + arg(z2)) ∈ R.<br />

Thus, we extend the notation J(a, b, x) for b = −1 by the above expression.<br />

4.6.5 Analysis<br />

Integration by parts<br />

We give in this subsection the integration by part theorem for the Lebesgues and the Stieltjes<br />

integrals from Gordon (1994)’s graduate book. Other reference books inclu<strong>de</strong> Hil<strong>de</strong>brandt<br />

(1971); Stroock (1994).<br />

Lebesgues integral<br />

Theorem (Theorem 12.5 of Gordon (1994)). Let f be Lebesgues-integrable on [a, b] and<br />

F (x) = x<br />

a f for each x ∈ [a, b]. If G is absolutely continuous on [a, b], then fG is Lebesguesintegrable<br />

and we have<br />

b<br />

a<br />

fG = F (b)G(b) −<br />

b<br />

a<br />

F G ′ .<br />

If the integral limits as b tends to infinity exist, one can consi<strong>de</strong>r the in<strong>de</strong>finite integral<br />

version +∞<br />

+∞<br />

fG = lim F (b)G(b) − F G<br />

b→+∞ ′ .<br />

Stieltjes integral<br />

a<br />

Theorem (Theorem 12.14 of Gordon (1994)). Let f, g be boun<strong>de</strong>d functions on a closed interval<br />

[a, b]. If f is Stieltjes-integrable with respect to g on [a, b], then g is also Stieltjes-integrable<br />

with respect to g on [a, b] and we have<br />

210<br />

b<br />

g(t)df(t) = [g(t)f(t)]<br />

a<br />

b b<br />

a −<br />

a<br />

a<br />

f(t)dg(t).

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

Saved successfully!

Ooh no, something went wrong!