0% found this document useful (0 votes)
20 views

MSM780 2021 StabilityEllipticModelEquation

This document discusses the stability of the elliptic model equation φxx + φyy = 0 using two methods: 1) Von Neumann test for stability. This involves substituting φ = eλt eiax eiby into the model equation and finding the conditions for stability. 2) Modified equation analysis. This involves performing Taylor series expansions in space for the point Jacobi algorithm and identifying the leading order terms, which results in an equation that must match the original differential equation for stability. The Von Neumann test shows the algorithm is unconditionally stable, meaning the time step restriction does not depend on grid spacing. The modified equation analysis identifies the terms that must cancel out to reproduce the original differential equation

Uploaded by

Mitchell Whiting
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
20 views

MSM780 2021 StabilityEllipticModelEquation

This document discusses the stability of the elliptic model equation φxx + φyy = 0 using two methods: 1) Von Neumann test for stability. This involves substituting φ = eλt eiax eiby into the model equation and finding the conditions for stability. 2) Modified equation analysis. This involves performing Taylor series expansions in space for the point Jacobi algorithm and identifying the leading order terms, which results in an equation that must match the original differential equation for stability. The Von Neumann test shows the algorithm is unconditionally stable, meaning the time step restriction does not depend on grid spacing. The modified equation analysis identifies the terms that must cancel out to reproduce the original differential equation

Uploaded by

Mitchell Whiting
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 6

MSM780 Stability – Elliptic Model Equation University of Pretoria

• This slide show accompanies the document:


MSM780_2013_Stability_vonNeumann.pdf
• This slide show accompanies the document:
MSM780_2013_Stability_ModEqn.pdf

1
MSM780 Stability – Elliptic Model Equation University of Pretoria

Elliptic model equation


φ xx + φ yy = 0
• Stability using
– Von Neumann test for stability
– Modified equation analysis

2
MSM780
.
Stability – Elliptic Model Equation University of Pretoria

Von Neumann
1  φ in−1, j + φ in+1, j φ in, j −1 + φ in, j +1 
 
Point Jacobi φ in, +j 1 =
2 ( 1
∆x + ∆y 2
2 ) 
 ∆ x 2
+
∆ y 2 

or
1
∆x 2
φ n
(
i +1, j − 2φ n +1
i, j + φ n
i −1, j +
1
∆y 2
φ n
)
i , j +1 − 2φ i, j(
n +1
+ φ n
i , j −1 = 0 )

n λt iax iby
Substitute φ =e e e
i, j

∆x 2
(
1 ia∆x
e − 2e λ∆t
+ e −ia∆x
+
1 ib∆y
∆y 2
e − 2e)λ∆t
+ e −ib∆y
=0 ( )

But e ia∆x
+e − ia∆x
= 2 cos(a∆x )

Therefore
1
∆x 2
(2 cos a ∆x − 2 e λ∆t
+
1
∆y 2
2 cos )
b ∆y − 2 e λ∆t
(
=0 )
3
MSM780
.
Stability – Elliptic Model Equation University of Pretoria

Von Neumann (cont.)


1
∆x 2
(
2 cos a ∆x − 2e λ∆t
+ )
1
∆y 2
2(cos b ∆y − 2e )
λ∆t
=0

If ∆x=∆y=∆
e λ∆t
= 1
2
(cos a∆x + cos b∆y )
For stability λ∆t
e ≤1

Always true, therefore unconditionally stable

4
MSM780
.
Stability – Elliptic Model Equation University of Pretoria

Modified equation analysis


Differential equation: u xx + u yy = 0
Algorithm:
Point Jacobi
1
∆x 2
u(n
i +1, j − 2u n +1
i, j + u n
)
i −1, j +
1
∆y 2
u(n
i , j +1 − 2u n +1
i, j + u n
)
i , j −1 = 0

Taylor series expansion in space (x-direction):


2 3 4
uin+1, j = uin, j + ∆xu x +

2
x
u xx +

6
x
u xxx +
∆ x
24
u xxxx + O ∆x 5 ( )
2 3 4
n n
ui −1, j = ui , j − ∆xu x +
∆x
2
u xx −
∆x
6
u xxx +
∆ x
24
u xxxx + O ∆x 5
( )
Add equations:
4
uin+1, j + uin−1, j = 2uin + ∆x 2u xx +

12
x
u xxxx + O ∆x 6 ( )
MSM780
.
Stability – Elliptic Model Equation University of Pretoria

Modified equation analysis (cont.)


Point Jacobi
1
∆x 2
u(n
i +1, j − 2u n +1
i, j + u n
)
i −1, j +
1
∆y 2
u(n
i , j +1 − 2u n +1
i, j + u n
)
i , j −1 = 0

Taylor series expansion in space (y-direction):


2 3 4
n n
ui , j +1 = ui , j + ∆yu y +
∆y
2
u yy +
∆y
6
u yyy +
∆y
24
u yyyy + O ∆y 5 ( )
2 3 4
uin, j −1 = uin, j − ∆yu y +

2
y
u yy −

6
y
u yyy +
∆y
24
u yyyy + O ∆y 5 ( )
Add equations: n 4
ui , j +1 + uin, j −1 = 2uin + ∆y 2u yy +
∆y
12
u yyyy + O ∆y 6 ( )
Combine into algorithm:
1  2 ∆x 4
 1  2 ∆y 4

 2ui , j − 2ui , j + ∆x u xx +
2 
u xxxx  + 2  2ui , j − 2ui , j + ∆y u yy + u yyyy  = 0
∆x  12  ∆y  12 
∆x 4 ∆y 4
Modified equation: ∴ u xx + u yy =−
12
u xxxx −
12
(
u yyyy + O ∆x 4 , ∆y 4 )

You might also like