{[ promptMessage ]}

Bookmark it

{[ promptMessage ]}

# Formulas involving backward di erences may for

This preview shows page 1. Sign up to view the full content.

This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: ution in terms of scaled derivatives at t = tn , i.e., as a n = yn hy0 n h2 y00 : : : hk;1 y(k;1)]T : 2! n (k ; 1)! n (5.7.10) Step size variation is simpli ed with this representation. The transformation between representations may be obtained from the Newton backward-di erence polynomial (5.3.3a) Pl;1(t) = l ;1 X 0 (;1)i ; i+ 1 riyn: i=0 (5.7.11) (The relationship between the order k and the degree of the Newton polynomial l ; 1 will become clear shortly.) Since Pl;1(t) is an approximation of y0(t), di erentiating (5.7.11) and setting = 1 gives a relation between the Nordsieck and backward-di erence representations. A similar linear relationship between the backward-di erence and standard representation (5.7.6b) can be used to relate the Nordsieck and the standard variables. Example 5.7.4. We'll construct a relationship between the Nordsieck and standard representations for the three-step (fourth-order) Adams method (5.7.5b). The Nordsieck vector for this method is 2 3 0 00 000 an = yn hyn h yn h yn ]T : 2 6 52 A second-degree polynomial su ces to get the necessary derivatives thus, setting l = 3 in (5.7.11), we have 0 0 01 P2 = yn + ( ; 1)ryn + 2 ( ; 1) r2yn: Di erentiating while using (5.7.7b) 2 01 0 h dPdt(t) = dPd2( ) = ryn + 2 (2 ; 1)r2yn: 00 Setting = 1 (t = tn ) gives an approximation of yn as 0 00 0 hyn = ryn + 1 r2 yn: 2 Di erentiating again 2 2 P P 0 h2 d dt22(t) = d d 2( ) = r2yn: 2 Setting = 1 000 0 h2 yn = r2yn: Summarizing the results in matrix form 2 32 32 3 yn 10 0 0 yn 6 hyn 7 6 0 1 0 0 7 6 hyn 7 6 2 000 7 = 6 76 0 0 7: 4 h yn=2 5 4 0 0 1=2 1=4 5 4 rhyn 5 000 0 h3 yn =6 0 0 0 1=6 r2hyn Multiplying by the transformation of Example 5.7.3 gives the desired relationship between the Nordsieck and traditional forms as 32 32 3 2 yn 10 00 yn 6 hyn 7 6 0 1 0 0 7 6 hyn 7 : 6 2 000 7 = 6 76 0 0 7 4 h yn=2 5 4 0 3=4 ;1 1=4 5 4 hyn;1 5 000 0 h3yn =6 0 1=6 ;1=3 1=6 hyn;2 Example 5.7.5. Storage was at a premium when the above transformations were developed. Now, with a third-degree interpolating polynomial and l...
View Full Document

{[ snackBarMessage ]}

Ask a homework question - tutors are online