Solving Ax = 0: pivot variables, special solutions We have a deﬁnition for the column space and the nullspace of a matrix, but how do we compute these subspaces? Computing the nullspace The nullspace of a matrix A is made up of the vectors x for which Ax = 0. Suppose: ⎤ ⎡ 1 2 2 2 8 ⎦. A=⎣ 2 4 6 3 6 8 10 (Note that the columns of this matrix A are not independent.) Our algorithm for computing the nullspace of this matrix uses the method of elimination, de spite the fact that A is not invertible. We don’t need to use an augmented matrix because the right side (the vector b) is 0 in this computation. The row operations used in the method of elimination don’t change the solution to Ax = b so they don’t change the nullspace. (They do affect the column space.) The ﬁrst step of elimination gives us: ⎡ ⎤ ⎡ ⎤ 1 2 2 2 1 2 2 2 8 ⎦ −→ ⎣ 0 0 2 4 ⎦ . A=⎣ 2 4 6 3 6 8 10 0 0 2 4 We don’t ﬁnd a pivot in the second column, so our third column of the second row: ⎡ ⎤ ⎡ 1 2 2 2 1 2 2 ⎣ 0 0 2 4 ⎦ −→ ⎣ 0 0 2 0 0 2 4 0 0 0 next pivot is the 2 in the ⎤ 2 4 ⎦=U 0 The matrix U is in echelon (staircase) form. The third row is zero because row 3 was a linear combination of rows 1 and 2; it was eliminated. The rank of a matrix A equals the number of pivots it has. In this example, the rank of A (and of U) is 2. Special solutions Once we’ve found U we can use back-substitution to ﬁnd the solutions x to the equation Ux = 0. In our example, columns 1 and 3 are pivot columns containing pivots, and columns 2 and 4 are free columns. We can assign any value to x2 and x4 ; we call these free variables. Suppose x2 = 1 and x4 = 0. Then: =⇒ 2x3 + 4x4 = 0 1 x3 = 0 and: x1 + 2x2 + 2x3 + 2x4 = 0 ⎡ ⎤ ⎢ So one solution is x = ⎣ =⇒ x1 = −2. −2 1 ⎥ (because the second column is just twice the 0 ⎦ 0 ﬁrst column). Any multiple of this vector is in the nullspace. Letting a different free variable equal 1 and setting the other free variables equal to zero gives us other vectors in the nullspace. For example: ⎡ ⎤ 2 ⎢ 0 ⎥ x=⎣ −2 ⎦ 1 has x4 = 1 and x2 = 0. The nullspace of A is the collection of all linear combi nations of these “special solution” vectors. The rank r of A equals the number of pivot columns, so the number of free columns is n − r: the number of columns (variables) minus the number of pivot columns. This equals the number of special solution vectors and the dimension of the nullspace. Reduced row echelon form By continuing to use the method of elimination we can convert U to a matrix R in reduced row echelon form (rref form), with pivots equal to 1 and zeros above and below the pivots. ⎡ ⎤ ⎡ ⎤ ⎡ ⎤ 1 2 0 −2 1 2 0 −2 1 2 2 2 4 ⎦ → ⎣ 0 0 1 2 ⎦ = R. U=⎣ 0 0 2 4 ⎦ → ⎣ 0 0 2 0 0 0 0 0 0 0 0 0 0 0 0 By exchanging some columns, R can be rewritten with a copy of the identity matrix in the upper left corner, possibly followed by some free columns on the right. If some rows of A are linearly dependent, the lower rows of the matrix R will be ﬁlled with zeros: � � I F R= . 0 0 (Here I is an r by r square matrix.) � � −F If N is the nullspace matrix N = then RN = 0. (Here I is an n − r by I n − r square matrix and 0 is an m by n − r matrix.) The columns of N are the special solutions. 2