07.05.2013 Views

Métodos miméticos de pasos fraccionarios - E-Archivo - Universidad ...

Métodos miméticos de pasos fraccionarios - E-Archivo - Universidad ...

Métodos miméticos de pasos fraccionarios - E-Archivo - Universidad ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

§2.3 spatial semidiscretization: a cell-no<strong>de</strong> mfd method 21<br />

• • •<br />

(i, j + 1)<br />

(i + 1, j + 1)<br />

⊗<br />

⊗<br />

• •⊙<br />

⊗<br />

(i, j)<br />

⊗<br />

(i + 1, j)<br />

• •<br />

•<br />

Figure 2.6: Stencil for the discrete operator A located at the internal cell center<br />

(i + 1/2, j + 1/2). Symbols: • : Ψ h ·,·; ⊗ : (G X Ψ h )·,·, (G Y Ψ h )·,·, K XX<br />

·,· , K XY Y Y<br />

·,· , K·,· ;<br />

•⊙ : (A Ψ h )i+1/2,j+1/2.<br />

hand, the components of G Ψ h at the internal (i, j)-no<strong>de</strong> reduce to:<br />

(G XΨh )i,j = (Ψh i−1/2,j−1/2 + Ψh i−1/2,j+1/2 ) − (Ψh i+1/2,j−1/2 + Ψh i+1/2,j+1/2 )<br />

2hX ,<br />

2hY ,<br />

(2.31)<br />

for i = 2, 3, . . . , Nx − 1 and j = 2, 3, . . . , Ny − 1. The obtained formulae give an<br />

approximation to the partial <strong>de</strong>rivatives ∂xψ and ∂yψ, respectively. By contrast,<br />

the expressions for the discrete gradient on the boundary no<strong>de</strong>s involve one-si<strong>de</strong>d<br />

differences. For instance, the x-component of G Ψh on the bottom boundary is<br />

<strong>de</strong>fined as:<br />

(G Y Ψ h )i,j = (Ψh i−1/2,j−1/2 + Ψh i+1/2,j−1/2 ) − (Ψh i−1/2,j+1/2 + Ψh i+1/2,j+1/2 )<br />

(G X Ψ h )i,1 = Ψh i−1/2,3/2 − Ψh i+1/2,3/2<br />

hX ,<br />

for i = 2, 3, . . . , Nx−1. Similar equations can be <strong>de</strong>rived for the other boundaries.<br />

Finally, at the left bottom corner, we get:<br />

(G XΨh )1,1 = − Ψh 3/2,3/2<br />

1 .<br />

2<br />

hX<br />

Formulae for the remaining corner no<strong>de</strong>s are <strong>de</strong>duced analogously.<br />

In or<strong>de</strong>r to obtain the formula for the discrete diffusion operator A = DK G ,<br />

we shall apply the discrete divergence (2.30) to a vector W h whose components<br />

at the (i, j)-no<strong>de</strong> are given by:<br />

W X i,j<br />

W Y i,j<br />

= KXX i,j (G XΨh )i,j + KXY i,j (G Y Ψh )i,j<br />

= KXY i,j (G XΨh Y Y<br />

)i,j + Ki,j (G Y Ψh )i,j,<br />

•<br />

(2.32)

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

Saved successfully!

Ooh no, something went wrong!