28.01.2013 Views

Implicit-Explicit Runge-Kutta schemes for hyperbolic systems ... - utenti

Implicit-Explicit Runge-Kutta schemes for hyperbolic systems ... - utenti

Implicit-Explicit Runge-Kutta schemes for hyperbolic systems ... - utenti

SHOW MORE
SHOW LESS

Create successful ePaper yourself

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

Asymptotic properties of IMEX <strong>schemes</strong><br />

An IMEX scheme <strong>for</strong> an <strong>hyperbolic</strong> system with relaxation has the <strong>for</strong>m<br />

U (i)<br />

i = U0<br />

�i−1<br />

+ h<br />

U1 = U0 + h<br />

j=1<br />

ν�<br />

i=1<br />

ãijF (U (j) ) + h<br />

˜wiF (U (i) ) + h<br />

ν�<br />

j=1<br />

ν�<br />

i=1<br />

1<br />

aij<br />

ε R(U (j) ),<br />

1<br />

wi<br />

ε R(U (i) ).<br />

Definition 1 We say that an IMEX scheme <strong>for</strong> an <strong>hyperbolic</strong> system with relaxation<br />

is asymptotic preserving (AP) if in the limit ɛ → 0 the scheme becomes a consistent<br />

discretization of the limit system of conservation laws. We use the notation APk if the<br />

scheme is of order k in the limit ɛ → 0.<br />

Note that this definition does not imply that the scheme preserves the order of accuracy<br />

in t in the stiff limit ɛ → 0. In the latter case the scheme is said asymptotically accurate.<br />

Examples: Scheme SP(1,1,1) is clearly AP1. Scheme Jin(2,2,2) is AP2, but it is not<br />

uni<strong>for</strong>mly valid in ε. Schemes Midpoint(1,2,2) and CN(2,2,2) are not AP even if both<br />

implicit parts of the <strong>schemes</strong> are A-stable. On the contrary, <strong>schemes</strong> CJR(3,2,2) and<br />

LRR(3,2,2) are AP and uni<strong>for</strong>mly valid in ε, but only scheme LRR(3,2,2) is AP2.<br />

13

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

Saved successfully!

Ooh no, something went wrong!