ExercisesLinear Equations > Exercises
Nullspace, rank and range
where , with , and . In the above, the zeroes are in fact matrices of zeroes with appropriate sizes.
Dimension of solution setDetermine the dimension of the sets of solutions to the following linear equations in a vector . Hint: use dyads and matrix notation. Solving linear equations with multiple right-hand sidesOften it is of interest to be able to solve linear equations of the form for many different instances of the output vector . In this problem we assume that we are given such instances , , which are collected as columns of a matrix . A direct approach to this task is encapsulated in the following matlab snippet: n = 500; p = 100; A = randn(n); B = randn(n,p); tic for k=1:p b = B(:,k); x = A\b; end fprintf(’elapsed time = %10.7f\n’, toc) Here is another approach: n = 500; p = 100; A = randn(n); B = randn(n,p); tic [Q,R] = qr(A); for k=1:p b = B(:,k); x = R\(Q’*b); end fprintf(’elapsed time = %10.7f\n’, toc) Which method is faster? Justify your answer. Polynomial interpolationIn this problem, we look at a simple application of the range space for fitting a polynomial through a set of points. We are given points in and we want to find a polynomial of degree such that, for all , we have . That is, the polynomial must go through all the points. A polynomial of degree is parametrized by the vector of its coefficients, that is: We assume that if .
|