Numerical Methods Contents - SAM
Numerical Methods Contents - SAM
Numerical Methods Contents - SAM
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
⇒<br />
if y ∈ C 2 ([0, T]), then<br />
y(t k+1 ) − y(t k ) = ẏ(t k )h k + 1 2ÿ(ξ k)h 2 k for some t k ≤ ξ k ≤ t k+1<br />
= f(y(t k ))h k + 1 2ÿ(ξ k)h 2 k ,<br />
Total error arises from accumulation of one-step errors!<br />
since t ↦→ y(t) solves the ODE, which implies<br />
the one-step error from (11.3.4)<br />
τ(h k ,y(t k ))=Ψ h ky(t k ) − y(t k + h k )<br />
ẏ(t k ) = f(y(t k )). This leads to an expression for<br />
(11.3.2)<br />
= y(t k ) + h k f(y(t k )) − y(t k ) − f(y(t k ))h k + 2ÿ(ξ 1 k)h 2 (11.3.6)<br />
k<br />
= 1 2ÿ(ξ k)h 2 k .<br />
error bound = O(h), h := max h l (➤<br />
l<br />
1st-order algebraic convergence)<br />
Error bound grows exponentially with the length T of the integration interval.<br />
Most commonly used single step methods display algebraic convergence of integer order with respect<br />
to the meshwidth h := max k h k . This offers a criterion for gauging their quality.<br />
Sloppily speaking, we observe τ(h k ,y(t k )) = O(h 2 k ) uniformly for h k → 0.<br />
The sequence (y k ) k generated by a<br />
➂ Estimate for the propagated error from (11.3.3)<br />
∥<br />
∥Ψ h ky k − Ψ h ky(t k ) ∥ = ‖y k + h k f(y k ) − y(t k ) − h k f(y(t k ))‖<br />
(11.3.1)<br />
≤ (1 + Lh k ) ‖y k − y(t k )‖ .<br />
(11.3.7)<br />
Ôº ½½º¿<br />
single step method (→ Def. 11.2.1) of order (of consistency) p ∈ N<br />
for ẏ = f(t,y) on a mesh M := {t 0 < t 1 < · · · < t N = T } satisfies<br />
max<br />
k ‖y k − y(t k )‖ ≤ Ch p for h := max<br />
k=1,...,N |t k − t k−1 | → 0 ,<br />
with C > 0 independent of M, provided that f is sufficiently smooth.<br />
Ôº ½½º<br />
➂ Recursion for error norms ǫ k := ‖e k ‖ by △-inequality:<br />
11.4 Runge-Kutta methods<br />
ǫ k+1 ≤ (1 + h k L)ǫ k + ρ k , ρ k := 2 1h2 k max ‖ÿ(τ)‖ . (11.3.8)<br />
t k ≤τ≤t k+1<br />
Taking into account ǫ 0 = 0 this leads to<br />
Use the elementary estimate<br />
ǫ k ≤<br />
k∑ l−1 ∏<br />
(1 + Lh j ) ρ l , k = 1, ...,N . (11.3.9)<br />
l=1 j=1<br />
(1 + Lh j ) ≤ exp(Lh j ) (by convexity of exponential function):<br />
So far we only know first order methods, the explicit and implicit Euler method (11.2.1) and (11.2.4),<br />
respectively.<br />
Now we will build a class of methods that achieve orders > 1. The starting point is a simple integral<br />
equation satisfied by solutions of initial value problems:<br />
Note:<br />
(11.3.9) ⇒ ǫ k ≤<br />
l−1 ∑<br />
h j ≤ T for final time T<br />
j=1<br />
k∑ l−1 ∏<br />
exp(Lh j ) · ρ l =<br />
l=1 j=1<br />
k∑<br />
l=1<br />
exp(L ∑ l−1<br />
j=1 h j)ρ l .<br />
IVP:<br />
∫<br />
ẏ(t) = f(t,y(t)) ,<br />
t1<br />
⇒ y(t 1 ) = y 0 + f(τ,y(τ)) dτ<br />
y(t 0 ) = y 0 t 0<br />
k∑<br />
ρ<br />
k∑<br />
ǫ k ≤ exp(LT) ρ l ≤ exp(LT) max k<br />
h<br />
k h l<br />
l=1<br />
k<br />
l=1<br />
≤ T exp(LT) max h l · max ‖ÿ(τ)‖ .<br />
l=1,...,k t 0 ≤τ≤t k<br />
‖y k − y(t k )‖ ≤ T exp(LT) max h l · max ‖ÿ(τ)‖ .<br />
l=1,...,k t 0 ≤τ≤t k<br />
Ôº ½½º¿<br />
Idea: approximate integral by means of s-point quadrature formula (→ Sect. 10.1,<br />
defined on reference interval [0, 1]) with nodes c 1 ,...,c s , weights b 1 ,...,b s .<br />
s∑<br />
y(t 1 ) ≈ y 1 = y 0 + h b i f(t 0 + c i h, y(t 0 + c i h) ) , h := t 1 − t 0 . (11.4.1)<br />
i=1<br />
Obtain these values by bootstrapping<br />
Ôº¼ ½½º