Scribd
Upload
113 views4 pages

Stability Analysis Euler PDF

The document analyzes and compares the stability of the forward Euler and backward Euler methods for solving differential equations numerically. The forward Euler method is conditionally stable, with the stability condition being that the time step h must be less than -2/λ, where λ is a property of the differential equation. The backward Euler method is unconditionally stable, meaning the time step h can be any positive value and stability is guaranteed. Implicit methods like backward Euler are preferred for stiff differential equations that require very small time steps for explicit methods to remain stable.

Uploaded by

Jack Cavaluzzi
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
113 views4 pages

Stability Analysis Euler PDF

The document analyzes and compares the stability of the forward Euler and backward Euler methods for solving differential equations numerically. The forward Euler method is conditionally stable, with the stability condition being that the time step h must be less than -2/λ, where λ is a property of the differential equation. The backward Euler method is unconditionally stable, meaning the time step h can be any positive value and stability is guaranteed. Implicit methods like backward Euler are preferred for stiff differential equations that require very small time steps for explicit methods to remain stable.

Uploaded by

Jack Cavaluzzi
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/ 4

Stability Analysis

Siyang Wang
September 26, 2013

We analyze the stability condition of forward Euler method and backward


Euler method by using the so called test equation.

Forward Euler method


The test equation reads

y 0 = y (1)
y(0) = y, (2)

where is a complex number. We also assume y 6= 0, otherwise we get the


trivial zero solution. The forward Euler method reads

yk+1 = yk + hf (tk , yk ).

Substituting ODE to the above formula, we get

yk+1 = yk + hyk
= (1 + h)yk .

Therefore, by induction we have

yk = (1 + h)k y. (3)

The exact solution of the test equation is

y(t) = yet .

(To see this, we take derivative with respect to t on both sides of the above
equation, we have y 0 (t) = yet = y, which is the same as the ODE.) If we
restrict such that its real part is negative, i.e. Re() < 0, then the exact
solution eventually decays to 0 as t goes to infinite (Do you know why?).
Numerically, this corresponds to the case that yk goes to 0 when k goes to
infinite in Equation (3). To satisfy this, we have to require

|1 + h| < 1. (4)

Inequality (4) is the stability condition for forward Euler method. We can
also plot the stability condition to visualize the stability region as shown in
Figure 1.

Figure 1: Stability region (shadowed circle) for forward Euler method

If we restrict that is a real number (which is often the case in practice),


then we can simplify the stability condition. From inequality (4), we obtain

1 < 1 + h < 1
2 < h < 0
0 < h < 2/

h is the step size, so it is always positive. Then the stability condition is


simplified to
h < 2/ (5)

This is also what you have seen in the lab session.


Backward Euler method
We apply backward Euler method to the test equation (1) with the initial
condition (2),

yk+1 = yk + hf (tk+1 , yk+1 ) = yk + hyk+1 .

Put yk+1 terms on one side,

(1 h)yk+1 = yk

which gives
1
yk+1 = yk .
(1 h)
Therefore, by induction, we get
 k
1
yk = y (6)
1 h

Again, we restrict to have negative real part so that the exact solution
decays to 0 when t goes to infinite. Correspondingly, the numerical solution
should have the same behaviour. Thus, we require
1
| |<1 (7)
1 h
Before we claim the stability condition of backward Euler method, we can
visualize inequality (7) in Figure 2.
Remember that we restrict that the real part of is negative. This corre-
sponds to the left half plane, which is covered by the stability region. This
means that backward Euler method is unconditionally stable.
If we restrict that is a real number (which is often the case in practice),
then we can see the unconditional stability from inequality (7):

|1 h| > 1
1 h > 1 or 1 h < 1
h < 0 or h > 2

Since h > 0 (h is the step size, it must be a positive number) and < 0,
we always have h < 0. Thus, backward Euler method is unconditionally
stable.
Figure 2: Stability region (shadowed) for backward Euler method

Remark
Implicit methods are in gerenal unconditionally stable. For a stiff problem,
explicit methods need to take very small time steps. And implicit methods
could be preferred in this case.
In MATLAB, we have ODE solvers suitable for stiff problems, for example
ode15s.

You might also like