31.05.2014 Views

Processus de Lévy en Finance - Laboratoire de Probabilités et ...

Processus de Lévy en Finance - Laboratoire de Probabilités et ...

Processus de Lévy en Finance - Laboratoire de Probabilités et ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

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

3.3. CHOICE OF THE REGULARIZATION PARAMETER 107<br />

From [30, Theorem 2.2],<br />

γ δ (α) ≥ I(Q δ α|Q ∗ ) + I(Q ∗ |P ). (3.16)<br />

Therefore<br />

Using the inequality<br />

I(Q δ α|Q ∗ ) ≤ ‖CQ∗ − C δ M ‖2 w<br />

α<br />

−−−→<br />

α→∞ 0.<br />

|P − Q| ≤ √ 2I(P |Q), (3.17)<br />

where |P − Q| <strong>de</strong>notes the total variation distance (see [30, Equation (2.3)]), this implies that<br />

Q δ α converges to Q ∗ in total variation distance (and therefore also weakly) as α goes to infinity.<br />

The limit relation now follows from Lemma 2.2.<br />

To prove the continuity of ε δ (α), l<strong>et</strong> {α n } be a sequ<strong>en</strong>ce of positive numbers, converging to<br />

α > 0. By the optimality of Q δ α n<br />

, I(Q δ α n<br />

|P ) is boun<strong>de</strong>d and one can choose a subsequ<strong>en</strong>ce of<br />

{Q δ α n<br />

}, converging weakly toward some measure Q ′ , and <strong>de</strong>noted, to simplify notation, again<br />

by {Q δ α n<br />

} n≥1 . We now need to prove that Q ′ is the solution of the calibration problem with<br />

regularization param<strong>et</strong>er α. By weak continuity of the pricing error (Lemma 2.2) and weak<br />

lower semicontinuity of the relative <strong>en</strong>tropy (Lemma 2.11), we have for any other measure Q:<br />

‖C Q′ − C δ M‖ 2 w + αI(Q ′ |P ) ≤ lim inf<br />

n {‖CQδ αn − C<br />

δ<br />

M ‖ 2 w + αI(Q δ α n<br />

|P )}<br />

= lim inf<br />

n {‖CQδ αn − C<br />

δ<br />

M ‖ 2 w + α n I(Q δ α n<br />

|P ) + (α − α n )I(Q δ α n<br />

|P )}<br />

≤ lim inf<br />

n {‖CQ − C δ M‖ 2 w + α n I(Q|P ) + (α − α n )I(Q δ α n<br />

|P )}<br />

= ‖C Q − C δ M‖ 2 w + αI(Q|P ).<br />

Therefore, the sequ<strong>en</strong>ce {ε δ (α n )} converges to one of the possible values of ε δ (α). If this function<br />

is single-valued in α, this means that every subsequ<strong>en</strong>ce of the original sequ<strong>en</strong>ce {ε δ (α n )} has<br />

a further subsequ<strong>en</strong>ce that converges toward ε δ (α), and therefore, the original sequ<strong>en</strong>ce also<br />

converges toward ε δ (α).<br />

The fact that J δ (α) is non<strong>de</strong>creasing as a function of α is trivial. To show the continuity,<br />

observe that for 0 < α 1 < α 2 ,<br />

J δ (α 2 ) − J δ (α 1 ) ≤ ε δ (α 1 ) + α 2 γ δ (α 1 ) − ε δ (α 1 ) − α 1 γ δ (α 1 )<br />

= (α 2 − α 1 )γ δ (α 1 ) ≤ (α 2 − α 1 )I(Q + |P ).

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

Saved successfully!

Ooh no, something went wrong!