Information Technology Reference
In-Depth Information
Fig. 4.2
L
2
(J
;
L
2
(G))
convergence rate for
θ
=
1
/
2
using uniform time steps
(
top
) and graded time steps
(
bottom
)
both with uniform time steps and with graded time steps. The exact values are ob-
tained using the analytic formulas [21]. We use the algebraic grading factor
β
=
3
for the call option and the extreme grading factor
β
25 for the digital option.
It can be seen in Fig.
4.2
that for the call and the digital option we only obtain
the optimal convergence rate using a graded time mesh.
=
Remark 4.4.2
Note that the theory from [146, Chap. 3.3] suggests that choosing
β
10 is sufficient in the case of a digital option in the Black-Scholes market to
obtain full convergence order of the given discretization. But the analysis in [146,
Chap. 3.3] is performed in a semi-discrete setting, i.e. there is no discretization error
in the space domain. This explains why in our situation, discretizing in space and
time, a larger grading factor has to be chosen.
=
Search WWH ::
Custom Search