# A predictor-corrector method and stability

1. Apr 14, 2014

### wel

A predictor-corrector method for the approximate solution of $y'=f(t,y)$ uses
y_{n+1}-y_{n}=hf_{n} \tag P

as predictor and
y_{n+1}-y_{n}=\frac{h}{2}(f_{n+1}-f_{n}) \tag C

IF $(P)$ and $(C)$ are used in PECE mode on the vector problem
\frac{du}{dt}=u

\frac{dv}{dt}=-10u-11v+cos(2\pi t)

with $u(0),v(0)$ given, find the largest constant $\gamma >0$ for which the scheme is stable in the sense of Von Neumann (Fourier series stability and frequency) whenever $0<\gamma<0$. Give full details of your argument.

=>
I haven't try very well because its really difficult question for me.
I was thinking
y_{n+1}=y_{n}+hf_{n} \tag P

as predictor and
y_{n+1}=y_{n}+\frac{h}{2}(f_{n+1}-f_{n}) \tag C

iam trying to get transition matrix but these condition
\frac{du}{dt}=u

\frac{dv}{dt}=-10u-11v+cos(2\pi t)

i don't know how and where to use.

2. Apr 15, 2014

### HallsofIvy

Are you saying you do not know how to solve for u and v in those last two equations?
$$\frac{du}{dt}= u$$
should be easy. And once you have that, the second equation if
$$\frac{dv}{dt}+ 11v= -10u+ cos(2\pi t)$$
is a relatively easy "linear equation with constant coefficients".