Some useful linear algebra Linearly independent vectors • 1v1 2v2 3v3 i vi 0 only for 1 2 i 0 • span(V): span of vector space V is all linear combinations of vectors vi,i.e. v i i i Ax x (I A) x 0 hence (I A) is singular The eigenvalues of A are the roots of the characteristic equation p( ) det( I A) 0 diagonal form of matrix 1 2 S 1 AS . N Eigenvectors of A are columns of S Similarity transform If B M 1 AM then A and B have the same eigenvalues The eigenvector x of A corresponds to the eigenvector M-1x of B Rank and Nullspace A xb m n n 1 m 1 Least Squares Ax b • More equations than unknowns • Look for solution which minimizes ||Ax-b|| = (Ax-b)T(Ax-b) • Solve ( Ax b)T ( Ax b) xi 0 • Same as the solution to • LS solution A Ax A b T T 1 x ( A A) A b T T Properties of SVD si2 are eigenvalues of ATA Columns of U (u1 , u2 , u3 ) are eigenvectors of AAT Columns of V (v1 , v2 , v3 ) are eigenvectors of ATA || A || F ai2, j s i2 i, j Solving A ( At A) 1 At A1 V 1 U with 1 x ( A A) A b t 0 1 t pseudoinverse of A and A V 01 U 1 equal to for all nonzero singular values and zero otherwise Least squares solution of homogeneous equation Ax=0 Minimize || Ax || subject to || x || 1 A UDV T || UDV T x |||| DV T x || and || x |||| VT x || y V T x minimize || DV T x || subject to || VT x || 1 or || Dy || subject to || y || 1 diagonal elements of D in descending order 0 0 y x Vy last column of V . 1 Enforce orthonormality constraints on an estimated rotation matrix R’ R' UDV T replace by R UIV T I is identity matrix Newton iteration X f (P) measurement parameter Find P satisfying X f(P) ε || || Min Start with P0 Assume f is locally linear X f(P1 ) f ( P0 ) J wh ere J P Pi 1 Pi i i is solution t o JΔi i X f ( Pi ) J T J i J T i f( ) is nonlinear Levenberg Marquardt iteration replace NΔ J T J J T i by N with N i,i (1 ) N i ,i Ni,j N i , j for i j Start with 10 3 if error is reduced new 10 if error is increased new 10 1