next up previous
Next: h_compute Up: h_compute Previous: h_compute

6. Matrix with given null space

Problem. Find a matrix $ M$ whose null space is the span of
$ \left[\begin{array}{r}1\\  2\\  1\\  2\\  3\end{array}\right]$, $ \left[\begin{array}{r}2\\  4\\  3\\  7\\  8\end{array}\right]$, and $ \left[\begin{array}{r}1\\  2\\  3\\  8\\  8\end{array}\right]$.



Discussion. $ M$ should have five columns; we don't yet know how many rows. Concentrate on one row of $ M$; since it's unknown, call it x$ = \left[\begin{array}{rrrrr}x&y&z&s&t\end{array}\right]$. When we multiply it by any of the three given column vectors we are supposed to get 0. Since x times a matrix equals x times each column of the matrix, if we let $ A$ be the matrix with the given vectors as columns we have x$ A =
\left[\begin{array}{rrr}0&0&0\end{array}\right]$. This equation describes the space of all row vectors eligible to be a row of $ M$. So we should choose $ M$ to have rows that are a basis for this space. How? Just transpose to get a more familiar kind of problem, a matrix times an unknown column vector: $ A^t$   x$ ^t =$   0. (Here we used the idea $ (BC)^t = C^t B^t$.) Now we know what to do:

Method. Make $ A$ with the given vectors as columns, find a basis for the null space of $ A^t$, and use those basis vectors as the rows of $ M$. (In doing that we're transposing the basis vectors to make them row vectors.)




next up previous
Next: h_compute Up: h_compute Previous: h_compute
Kirby A. Baker 2001-10-17