06.08.2013 Views

Abstract

Abstract

Abstract

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.

CHAPTER 1. OVERVIEW 22<br />

get<br />

<br />

<br />

K(f) <br />

(xm,kj)<br />

⎧<br />

⎪⎨ −<br />

≈<br />

⎪⎩<br />

hkj<br />

2m∗ (−3 fmj +4 fm−1,j − fm−2,j<br />

), if kj > 0<br />

2∆x<br />

− hkj<br />

2m∗ (3 fmj − 4 fm+1,j + fm+2,j<br />

), if kj < 0<br />

2∆x<br />

(1.18)<br />

We use a differencing scheme that depends on the sign of kj to respect the physics of<br />

the problem and thereby keep this approximation from being unstable. If kj > 0, this<br />

corresponds to electrons with positive momentum, moving from the left side of the<br />

RTD to the right side of the RTD. This means information is propogating from the<br />

left when kj > 0, and therefore to compute an approximation to the spatial derivative,<br />

we want to use functions values that are to the left of the current point. Similarly,<br />

when kj < 0, information is propogating from the right side of the RTD to the left<br />

side of the RTD, and we use function values that are to the right of the current point<br />

where we are computing an approximation to the spatial derivative.<br />

The P (f) term is discretized with the composite trapezoidal rule, which is second-<br />

order accurate. So we have the approximation:<br />

<br />

<br />

P (f) <br />

(xm,kj)<br />

≈− 4<br />

Nk <br />

fmj<br />

h<br />

′T (xm,kj − kj ′)wj ′ (1.19)<br />

where wj ′ are the weights of the composite trapezoidal rule, given by:<br />

wj ′ =<br />

⎧<br />

j ′ =1<br />

⎪⎨ ∆k, for j ′ =2, 3,...,Nk − 1<br />

⎪⎩<br />

∆k<br />

2 , for j′ =1,Nk

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

Saved successfully!

Ooh no, something went wrong!