Chapter 1: System of linear equations Linear system has the general form: a11 x1 a12 x2 a13 x3 ... a1n xn b1 a21 x1 a22 x2 a23 x3 ... a2 n xn b2 . . . an1 x1 an 2 x2 an 3 x3 ... ann xn bn where the a = constant coefficients the b = right hand side value the n = number of equations Linear system can be written in a matrix-vector form as AX = b, where a11 a12 ... a1n a21 a22 ... a2 n A . . a a ... a nn n1 n 2 x1 x2 X . . x 3 b1 b2 b . . b 3 Since the same operations are performed on the matrix A and the vector b, they are combined in the augmented matrix. a11 a12 ... a1n b1 a21 a22 ... a2 n b2 . . a a ... a n1 n 2 nn b3 Faculty of Engineering and Built Environment, SEGi University Kota Damansara Prepared by: Fatin Nur Diana binti Abu Samah 1.1 Gauss Elimination method Gauss Elimination is one of the techniques to solve linear systems. It involves combining equations to eliminate unknowns. The basic Gauss elimination procedure is elementary row operations. Elementary row operation is applied to the augmented matrix to yield an equivalent linear system. The solutions can be obtained by finding the upper-triangular matrix Rules for the elementary row operations: 1. Interchange : The order of two rows can be changed Ri R j 2. Scaling : Multiplying a row by a constant, mR j Ri 3. Replacement : The row can be replaced by the sum/subtract of that row and a nonzero multiple of any row. mR j Ri Ri Steps of elementary row operations: 1. Change the system of linear equations to an augmented matrix. 2. Find the upper triangular matrix. 3. Change back to the system of linear equations to form a back-substitution phase. 4. Solve the equations. Faculty of Engineering and Built Environment, SEGi University Kota Damansara Prepared by: Fatin Nur Diana binti Abu Samah 1.2 Gauss Elimination method with Partial Pivoting Obvious problems may occur when a pivot is zero because it leads to division by zero. 2 x2 3 x3 8 Example: 4 x1 6 x2 7 x3 3 2 x1 x2 6 x3 5 The pivot element means the element locates at the top row. Therefore, it is advantageous to determine the largest available coefficient in the column. After that, the row can be switched to be the pivot element. This is called partial pivoting. Steps of partial pivoting: 1. Change the system of linear equations to an augmented matrix. 2. Form a partial pivoting matrix. 3. Find the upper triangular matrix. 4. Change back to the system of linear equations to form a back-substitution phase. 5. Solve the equations. 1.3 LU Decomposition The combination of lower triangular and upper triangular matrices. Steps of LU decomposition: 1. Change the system of linear equations to an augmented matrix. 2. Find the upper triangular matrix (U) by using elementary row operation. R mR j Ri Ri for m i Rj 3. Find the lower triangular matrix (L) by putting the constant with opposite sign. 4. Solve LY b . 5. Solve UX Y . Faculty of Engineering and Built Environment, SEGi University Kota Damansara Prepared by: Fatin Nur Diana binti Abu Samah 1.4 Gauss-Seidel Iteration method The Gauss-Seidel method is most commonly used the iterative method. It involves with a large number of equations. A square matrix A is said to be strictly diagonally dominant when N akk akj for k 1,2,... N j 1 This means that in each row of the matrix, the magnitude of the main diagonal element must exceed the sum of other magnitudes of elements. 8 x1 x2 x3 8 Example: x1 7 x2 2 x3 4 2 x1 x2 9 x3 12 8 1 1 1 7 2 2 1 9 Steps of Gauss-Seidel method: 1. Make sure matrix A is diagonally dominant. 2. Form the general equations for unknowns. 3. Construct a table. 4. Stop the iteration until the difference xi( k 1) xi( k ) . Faculty of Engineering and Built Environment, SEGi University Kota Damansara Prepared by: Fatin Nur Diana binti Abu Samah