quiz1_sols_18085

# quiz1_sols_18085 - MIT OpenCourseWare http/ocw.mit.edu...

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: MIT OpenCourseWare http://ocw.mit.edu 18.085 Computational Science and Engineering I Fall 2008 For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms. 18.085 Quiz 1 October 5, 2007 Professor Strang Grading 1 2 3 Your PRINTED name is: SOLUTIONS 1) (39 pts.) With h = 1 3 there are 4 meshp oints 0, 1 , 2 , 1 and displacements u0 , u1 , u2 , u3 . 33 a) Write down the matrices A0 , A1 , A2 with three rows that produce the ﬁrst diﬀerences ui − ui−1 : A0 has 0 boundary conditions on u A1 has 1 boundary condition u0 = 0 (left end ﬁxed) A2 has 2 boundary conditions u0 = u3 = 0. b) Write down all three matrices AT A0 , AT A1 , AT A2 . 0 1 2 CROSS OUT IF FALSE / GIVE REASON BASED ON COLUMNS OF A ! K0 = AT A0 is (singular) (invertible) (p ositive deﬁnite) 0 Reason: K1 = AT A1 is (singular) (invertible) (p ositive deﬁnite) 1 Reason: c) Find all solutions w = (w1 , w2 , w3 ) to each of these equations: AT w = 0 0 AT w = 0 1 AT w = 0 2 1 Solution. a) � � A0 = � � b) � 1 −1 ⎡ � ⎡ � −1 1 −1 1 −1 1 ⎢ ⎢ ⎢ ⎣ ⎡ � � A1 = � −1 � � 1 1 −1 1 ⎢ ⎢ ⎢ ⎣ ⎡ � ⎢ � ⎢ A2 = � −1 ⎢ 1 � ⎣ 0 −1 � 1 0 ⎡ � ⎢ � ⎢ � −1 ⎢ 2 −1 T ⎢ A0 A0 = � � ⎢ � −1 2 −1 ⎢ � ⎣ −1 1 � singular c) � � AT A1 = � −1 1 � 2 −1 ⎢ ⎢ 2 −1 ⎢ ⎣ −1 1 AT A2 = � 2 � 2 −1 −1 2 ⎡ ⎣ � (invertible) (positive deﬁnite) �⎡ � ⎢w � ⎢� 1⎢ � 1 −1 ⎢� ⎢ T ⎢�w ⎢ A0 w = � 2 � ⎢ � 1 −1 ⎢ � ⎣ � ⎣ w3 1 AT w = 0 1 −� � −1 ⎡ −� w=0 w=0 �⎡ � ⎡ w1 �⎡ ⎢ 1 −1 0� ⎢ 0 � ⎣ � w2 ⎢ = � ⎣ AT w = � 2 0 1 −1 � ⎣ 0 w3 −� �⎡ 1 �⎢ �⎢ w = c�1⎢ �⎣ 1 2 2) (33 pts.) a) Find the eigenvalues �1 , �2 , �3 and unit eigenvectors y1 , y2 , y3 of B . � Hint: one eigenvector is (1, 0, −1)/ 2. � ⎡ 1 −1 0 ⎢ � � ⎢ B = � −1 2 −1 ⎢ . � ⎣ 0 −1 1 b) Factor B into Q�QT with Q−1 = QT . Draw a graph of the energy function f (u1 , u2 , u3 ) = 1T u Bu. 2 This is a surface in 4-dimensional u1 , u2 , u3 , f space so your graph may not be perfect—OK to describe it in 1 sentence. c) What diﬀerential equation with what boundary conditions on y (x) at x = 0 and 1 is the continuous analog of B y = �y ? What are the eigenfunctions y (x) and eigenvalues � in this diﬀerential equation ? At which x’s would you sample the ﬁrst three eigenfunctions to get the three eigenvectors in part (a) ? 3 Solution. �⎡ 1 �⎢ �⎢ a) y1 = � 1 ⎢ has B y1 = 0 so �1 = 0 (check: trace =4) �⎣ 1 � 3 � ⎡ 1 � ⎢ � ⎢ given vector y2 = � 0 ⎢ has B y2 = y2 so �2 = 1 � ⎣ −1 � 2 � ⎡ 1 � ⎢ ⎢ � y3 is orthogonal to y1 , y2 y3 = � −2 ⎢ with �3 = 3 � ⎣ 1 � 6 b) The orthonormal eigenvectors are the columns of Q (orthonormal gives Q T Q = I ). � ⎡ � ⎡ 1 1 1 � ⎢ 0 � ⎢ ⎢ � �1 ⎢ 0 −2 ⎢ � Then B = Q�QT with Q = � �=� 1 ⎢ ⎣ ⎣ � 1 −1 1 3 � � � 3 2 6 1 The graph of f = 2 uT Bu has a valley along the line of eigenvectors u = (c, c, c). The surface goes up the orthogonal directions. f �⎡ 1 �⎢ �⎢ �1⎢ �⎣ 1 4 c) B is free-free so the equation is −y �� = �y with y � = 0 at x = 0, 1. yk = cos k�x , k = 0, 1, 2, . . . ⎤1 , 3, 5 666 ⎥ to get (a multiple of) Sample at the points x = the discrete eigenvectors y1 , y2 , y3 . 5 3) (28 pts.) The ﬁxed-ﬁxed ﬁgure shows n = 2 masses and m = 4 springs. Displacements u1 , u2 . c1 m1 c2 m2 c4 c3 a) Write down the stretching-displacement matrix A in e = Au. b) What is the stiﬀness matrix K = AT CA for this system ? c) Theory question about any AT CA. C is symmetric positive deﬁnite. What condition on A assures that uT AT CAu > 0 for every → vector u = 0 ? Explain why this is greater than zero and where you use your condition on A. 6 Solution. a) � e1 �⎢� ⎢� ⎡ �⎢� ⎢ � e2 ⎢ � −1 1 ⎢ u1 ⎢ � ⎣ = Au � ⎢=� �⎢� ⎢ � e3 ⎢ � 0 1 ⎢ u2 �⎣� ⎣ 0 −1 e4 b) � c1 c2 c3 ⎡� 1 0 ⎡ ⎡ � 1 0 ⎡ AT CA = � � 1 −1 0 1 ⎡� � 0 0� ⎣� � 1 −1 � � ⎢� ⎢� ⎡ ⎢� ⎢ ⎢ � −1 c + c2 −c2 1⎢ ⎢� ⎢=� 1 ⎣ ⎢� ⎢ ⎢� 0 1⎢ −c2 c2 + c 3 + c 4 ⎣� ⎣ c4 0 −1 c) Write uT AT CAu = (Au)T C (Au) = eT Ce Since C is p ositive deﬁnite, this is positive unless e = 0. Condition on A: Independent columns. Then e = Au is zero only if u = 0. So AT CA is positive deﬁnite. 7 ...
View Full Document

• Fall '10
• staff
• Boundary value problem, Eigenvalue, eigenvector and eigenspace, Orthogonal matrix, Boundary conditions, ositive deﬁnite

{[ snackBarMessage ]}

Ask a homework question - tutors are online