You are on page 1of 2

Lecture#1: The geometry of linear equations

 Row picture of system of equations. It shows two lines meeting at a single point (solution)
 Column picture of system of equations. It combines column vectors on the left side to produce
the column vector on the right side.
 b is a linear combination of columns of matrix A with x being weights of the combination.
 Columns of matrix A determine which b’s are achievable.
 Row picture leads to dot product. Row * col = scalar

Lecture#2: Elimination with matrices

 Gauss elimination produces upper triangular matrix U.


 Pivots are the diagonal entries of the upper triangular matrix U.
 If every column has a pivot, matrix is all fine.

Lecture#3: Matrix operations and inverses

 For matrix multiplication AB, matrices must be of reasonable shape. For example, if A is m x n, B
must be of n x p i.e. number of columns of A must be equal to number of rows of B.
 4 ways to perform matrix multiplication:
o An entry at a time. An entry at ABij comes from ith row of A * jth column of B.
o ith row of matrix A linearly combines rows of matrix B to form ith row of AB.
o Ith column of matrix B linearly combines columns of matrix A to form ith column of AB.
o Block multiplication.
 Inverse of a matrix. Gause-Jordan elimination. Continue where Gauss has left us and eliminate in
the upward direction until A goes to I. Matrix E will be the inverse A.

Lecture#4: LU and LDU factorization

 Elimination gives us A = LU. L is a lower triangular matrix containing multipliers at (i,j) locations
used by matrix E
 A = LDU. L contains the multipliers. D contains the pivots. U has 1’s on the diagonal now.

Lecture#5: Transposes and permutations, spaces R^n

 Transposes follow reverse ordering rule just like inverses do.


 The dot product of Ax and y is equal to the dot product of x and ATy
 To get good pivots, re-order rows of A using permutation matrix. Then, PA = LU.

Lecture#6: Column space and nullspace

 A subspace containing v and w must also contain all of their linear combinations.
 To solve Ax=b, b must lie in the column space of A i.e. we must be able to get b using linear
combination of columns of A.
 The column space C(A) describes all the attainable right sides b.
Lecture#7: Solving Ax = 0: pivot variables, special solutions

You might also like