Elementary Matrices An elementary matrix is a certain simple type of invertible matrix that differs only slightly from the identity matrix. There are three generic types of elementary matrices. We describe these and list all the 3 x 3 elementary matrices. Elementary Matrices of Type I (Permutation matrices). These are simply the identity matrix with the columns interchanged. Example. The 3 × 3 type I elementary matrices are the following : 100 100 010 001 001 010 010 , 001 , 100 , 100 , 010 , 000 . 001 101 010 001 010 100 Elementary Matrices of Type II. These are simply the identity matrix with one of the diagonal entries replaced by a nonzero constant a. Example. The 3 × 3 type II elementary matrices are the following: a00 100 100 010 , 0a0 , 010 001 001 00a Elementary Matrices of Type III (Shear Matrices). These are simply the identity matrix with at most one nonzero number a in an off diagonal position. Example. The 3 × 3 type III elementary matrices are the following: 1a0 10a 100 100 100 100 010 , 010 , a10 , 01a , 010 , 010 001 001 001 001 a01 0a1 for a ! 0. Inverses of Elementary Matrices All the elementary matrices are invertible. In fact, the inverses of elementary matrices can be obtained by inspection. Inverse of a Type I Elementary Matrix. The inverse of such a matrix is simply its transpose. We can verify this using the dot product form of matrix multiplication. Here an elementary matrix T of type I has the form [ei(1), … , ei(n)] , where {i(1), … , i(n)} = {1, … ,n}. Here the vectors ej are the Elementary Matrices page 1 column vectors T T e1 = (1, 0, … , 0) , … , en = (0, 0, … , 1) . So the matrix product is T [ei(1), … , ei(n)] [ei(1), … , ei(n)] = (ei(s)"ei(t))s,t = (# st)s,t = I. where #s,t is the Kronecker delta given by #s , t = 1 if s = t . 0 if s ! t 010 001 Example. The inverse of 0 0 1 is 1 0 0 . 100 010 Inverse of a Type II Elementary Matrix. If the type II matrix is the identity matrix except for a ! 0 in the i,i term, then the inverse of the matrix is the identity matrix except for 1/a in the i,i term. Example. The inverse of 100 1 0 a 0 is 0 1/a 0 . 001 0 0 0 0 1 Inverse of a Type III Elementary Matrix. If the type III matrix is the identity matrix except for a ! 0 in the i,j term for i ! j, then the inverse of the matrix is the identity matrix except for - a in the i,j term. Example. 1a0 010 001 -1 1 -a 0 = 0 1 0 . 0 0 1 Multiplication on the Left by Elementary Matrices Multiplication of an n x p matrix A on the left by an elementary matrix E causes certain changes in the rows of the matrix A. We describe the specific changes below. Multiplication on the left by an elementary matrix E of type I. Let E by the elementary matrix E =[e1, e2, … , ei-1, ej, ei+1, … , ej-1, ei, ej+1, … en] Elementary Matrices page 2 T (the identity matrix with the i and j column interchanged); then EA is the matrix A except that the i and j row of A have been interchanged. Example. a11 a12 a13 a14 100 a11 a12 a13 a14 001 a21 a22 a23 a24 = a31 a32 a33 a34 . 010 a31 a32 a33 a34 a21 a22 a23 a24 Other examples are similar but depend on the introduction of the concept of permutation. We defer this to later when we discuss determinants. Multiplication on the left by an elementary matrix E of type II. Let E by the elementary matrix equal to the identity matrix except for a ! 0 in the i,i place. Then EA is equal to the matrix A except that the ith row of A has been multiplied by A. Example. a11 a12 a13 a14 100 a11 a12 a13 a14 0a0 a21 a22 a23 a24 = aa21 aa22 aa23 aa24 . 001 a31 a32 a33 a34 a31 a32 a33 a34 Multiplication on the left by an elementary matrix E of type III. Let E by the elementary matrix equal to the identity matrix except for a in the i,j place where i ! j. Then EA is equal to the matrix A except that A has a new jth row equal to the jth row of A plus a times the ith row of A. Example. a11 +aa21 a12 + aa22 a13 +aa23 a14 + aa24 1a0 a11 a12 a13 a14 010 a21 a22 a23 a24 = a21 a22 a23 a24 001 a31 a32 a33 a34 a31 a32 a33 a34 . Here a is in the 1,2 place and so there is a new first row equal to the original first row of A added to a times the second row of A. Elementary Matrices page 3