Differential Equations Solutions 138

Differential Equations Solutions 138 - . Then the Jacobian...

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

View Full Document Right Arrow Icon
148 Chapter 24. Solutions: Nonlinear Systems It is silly to refactor the matrix B each time, when it is just a rank-1 update of the previous matrix. Instead, update a decomposition or (less desirable) update the inverse using the techniques of Chapter 7. CHALLENGE 24.3. For some vector r , we need to form ( A ZV T ) 1 r = A 1 r + A 1 Z ( I V T A 1 Z ) 1 V T A 1 r , where A = B ( k ) , Z = y B ( k ) s ,and V = s / ( s T s ). Thus we need to form t = A 1 r and u = A 1 Z ,atacosto f2 p multiplications, form α =1 V T u atacosto f n multiplications, form w =(( V T t ) ) u atacosto f2 n multiplications and 1 division add t and w . The total number of multiplications is 2 p +3 n + 1. Again, updating a matrix decomposition has many advantages over this approach! CHALLENGE 24.4. (Partial Solution.) (a) We compute the partial of ρ a ( λ, x ) with respect to λ : s = · x 2 y 3 + xy 2 ( x a 1 ) 2 xy 2 + x 2 y + xy ( y a
Background image of page 1
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: . Then the Jacobian of a is the 2 3 matrix b J ( , x ) = s , (1 ) I + J ( x ) , where J ( x ) is the matrix from Problem 1. (b) In order for the function to be transversal to zero, the matrix b J ( , x ) must be full rank (i.e., rank-2) at every point [0 , 1), x, y ( , ). The matrix J ( x ) has two eigenvalues call them 1 and 2 . The matrix K = (1 ) I + J ( x ) has eigenvalues (1 )+ i , so it is singular only if = 1 / (1 1 ) or = 1 / (1 2 ). Even if that happens, it is likely that the vector s will point in a dierent direction, making the rank of J ( , x ) equal to 2....
View Full Document

Ask a homework question - tutors are online