Non-Linear Systems By : Dr. Atul R. Phadke Associate Professor in Electrical Engineering College of Engineering Pune (Maharashtra) OBJECTIVES: After studying this unit, you will be able to understand: ✓ Successive Substitution Method for solution of non-linear systems. ✓ Newton-Raphson method for solution of non-liear systems. ✓ MATLAB fsolve function. 2 SUCCESSIVE SUBSTITUTION METHOD: System of non-linear equations can be expressed as: 𝑓1 𝑥1 , 𝑥2 , ⋯ , 𝑥𝑛 = 0 𝑓2 𝑥1 , 𝑥2 , ⋯ , 𝑥𝑛 = 0 ⋯ 𝑓𝑛 𝑥1 , 𝑥2 , ⋯ , 𝑥𝑛 = 0 Therefore, the solution is the values of 𝑥 ′ 𝑠 for that make the equations equal to zero. For simplicity, let us consider a system of two non-linear simultaneous equations with two unknowns 𝑥 and 𝑦: 𝑓1 𝑥, 𝑦 = 0 𝑓2 𝑥, 𝑦 = 0 3 SUCCESSIVE SUBSTITUTION METHOD: 𝑓1 𝑥, 𝑦 = 0 𝑓2 𝑥, 𝑦 = 0 As in the case of single equation, these equations may be written in the form: 𝑥1 = 𝑔1 𝑥, 𝑦 𝑥2 = 𝑔2 𝑥, 𝑦 If 𝑥0 , 𝑦0 is an initial approximation of the root, then the modified guess can be determined as: 𝑥1 = 𝑔1 𝑥0 , 𝑦0 𝑦1 = 𝑔2 𝑥0 , 𝑦0 𝑥2 = 𝑔1 𝑥1 , 𝑦1 𝑦2 = 𝑔2 𝑥1 , 𝑦1 For faster convergence, recently computed value of 𝑥𝑖 may be used while evaluating 𝑦𝑖 4 SUCCESSIVE SUBSTITUTION METHOD: In general 𝑥𝑛+1 = 𝑔1 𝑥𝑛 , 𝑦𝑛 𝑦𝑛+1 = 𝑔2 𝑥𝑛 , 𝑦𝑛 For faster convergence, recently computed value of 𝑥𝑖 may be used while evaluating 𝑦𝑖 Try it: Find a root of the equations – 𝑦 2 − 5𝑦 + 4 = 0 3y𝑥 2 − 10𝑥 + 7 = 0 To apply the iteration method, we rewrite the equations as – 1 𝑥= 3y𝑥 2 + 7 10 1 2 𝑦 = 𝑦 +4 5 Let 𝑥0 = 0.5 and 𝑦0 = 0.5 5 SUCCESSIVE SUBSTITUTION METHOD: 1 1 2 2 𝑥𝑛+1 = 3y𝑛 𝑥𝑛 + 7 𝑦𝑛+1 = 𝑦𝑛 + 4 10 5 1 𝑥1 = 3 × 0.5 × 0.52 + 7 = 0.7375 10 1 𝑦1 = 0.52 + 4 = 0.85 5 1 𝑥2 = 3 × 0.85 × 0.73752 + 7 = 0.8387 10 1 𝑦2 = 0.852 + 4 = 0.9445 5 1 𝑥3 = 3 × 0.9445 × 0.83872 + 7 = 0.8993 10 1 𝑦3 = 0.0.94452 + 4 = 0.9784 5 Solution converges at 𝑥 = 1 and 𝑦 = 1 6 NEWTON RAPHSON METHOD: 𝑓1 𝑥, 𝑦 = 0 Where, 𝑓2 𝑥, 𝑦 = 0 𝑓1,0 = 𝑓1 𝑥0 , 𝑦0 Let the initial approximation to the root is 𝑥0 , 𝑦0 . 𝑓2,0 = 𝑓2 𝑥0 , 𝑦0 If 𝑥0 + ℎ, 𝑦0 + 𝑘 is the root of the system then – 𝑓1 𝑥0 + ℎ, 𝑦0 + ℎ = 0 𝑓2 𝑥0 + ℎ, 𝑦0 + ℎ = 0 Expanding above equations by Taylor series and neglecting higher order terms, 𝜕𝑓1 𝜕𝑓1 𝑓1 𝑥0 + ℎ, 𝑦0 + 𝑘 = 𝑓1,0 + ℎ +𝑘 =0 𝜕𝑥0 𝜕𝑦0 𝜕𝑓2 𝜕𝑓2 𝑓2 𝑥0 + ℎ, 𝑦0 + 𝑘 = 𝑓2,0 + ℎ +𝑘 =0 𝜕𝑥0 𝜕𝑦0 𝜕𝑓1 𝜕𝑓1 = 𝜕𝑥0 𝜕𝑥 𝜕𝑓1 𝜕𝑓1 = 𝜕𝑦0 𝜕𝑥 𝜕𝑓2 𝜕𝑓2 = 𝜕𝑥0 𝜕𝑥 𝜕𝑓2 𝜕𝑓2 = 𝜕𝑦0 𝜕𝑥 𝑥=𝑥0 , 𝑦=𝑦0 𝑥=𝑥0 , 𝑦=𝑦0 𝑥=𝑥0 , 𝑦=𝑦0 𝑥=𝑥0 , 𝑦=𝑦0 7 NEWTON RAPHSON METHOD: 𝜕𝑓1 𝜕𝑓1 𝑓1,0 + ℎ +𝑘 =0 𝜕𝑥0 𝜕𝑦0 𝜕𝑓2 𝜕𝑓2 𝑓2,0 + ℎ +𝑘 =0 𝜕𝑥0 𝜕𝑦0 Therefore, Therefore, 𝜕𝑓1 ℎ 𝜕𝑥0 𝑘 𝜕𝑓2 𝜕𝑥0 𝜕𝑓1 𝜕𝑥0 𝜕𝑓2 𝜕𝑥0 𝜕𝑓1 𝑓1,0 𝜕𝑦0 =− 𝜕𝑓2 𝑓2,0 𝜕𝑦0 𝜕𝑓1 𝜕𝑦0 𝜕𝑓2 𝜕𝑦0 𝜕𝑓1 𝜕𝑓1 ℎ +𝑘 = −𝑓1,0 𝜕𝑥0 𝜕𝑦0 Where, 𝜕𝑓2 𝜕𝑓2 ℎ +𝑘 = −𝑓2,0 𝜕𝑥0 𝜕𝑦0 the Jacobian matrix 𝐽. is called as Unique solution for this system exists if 𝑑𝑒𝑡 𝐽 ≠ 0 8 NEWTON RAPHSON METHOD: Therefore, 𝜕𝑓1 ℎ 𝜕𝑥0 𝑘 𝜕𝑓2 𝜕𝑥0 Therefore, 𝜕𝑓1 𝑓1,0 𝜕𝑦0 =− 𝜕𝑓2 𝑓2,0 𝜕𝑦0 𝑥1 − 𝑥0 𝑦1 − 𝑦0 = − 𝐽 −1 𝑓1,0 𝑓2,0 𝑥1 𝑥0 𝑦1 = 𝑦0 − 𝐽 −1 𝑓1,0 𝑓2,0 Therefore, ℎ 𝑘 𝑓1,0 𝐽 =− 𝑓2,0 Multiplying both sides by 𝐽−1 ℎ =− 𝐽 𝑘 −1 𝑓1,0 𝑓2,0 9 NEWTON RAPHSON METHOD: In general for a system of non-linear equations: 𝑓1 𝑥1 , 𝑥2 , ⋯ , 𝑥𝑛 = 0 𝑓2 𝑥1 , 𝑥2 , ⋯ , 𝑥𝑛 = 0 ⋯ 𝑓𝑛 𝑥1 , 𝑥2 , ⋯ , 𝑥𝑛 = 0 𝑥1, 𝑖+1 𝑥1, 𝑖 𝑥2, 𝑖+1 𝑥2, 𝑖 = . .. . 𝑥𝑛, 𝑖+1 𝑥𝑛, 𝑖 𝜕𝑓1,𝑖 𝜕𝑓1,𝑖 𝜕𝑥1 𝜕𝑥2 𝜕𝑓2,𝑖 𝜕𝑓2,𝑖 − 𝜕𝑥1 .𝜕𝑥2. . . 𝜕𝑓𝑛,𝑖 𝜕𝑓𝑛,𝑖 𝜕𝑥1 𝜕𝑥2 𝜕𝑓1,𝑖 .. 𝜕𝑥𝑛 𝜕𝑓2,𝑖 . . 𝜕𝑥𝑛 . . . . 𝜕𝑓𝑛,𝑖 . . 𝜕𝑥𝑛 −1 𝑓1, 𝑖 𝑓2, 𝑖 .. 𝑓𝑛, 𝑖 10 NEWTON RAPHSON METHOD: Solve the system 𝑥 2 + 𝑦 2 = 1 and 𝑦 = 𝑥 2 by Newton-Raphson method. 𝑥1 𝑥0 𝑦1 = 𝑦0 − 𝐽 −1 𝑓1,0 𝑓2,0 Let 𝑓1 = 𝑥 2 + 𝑦 2 − 1 = 0 and 𝑓2 = 𝑦 − 𝑥2 = 0 𝜕𝑓1 𝜕𝑥 𝐽= 𝜕𝑓2 𝜕𝑥 𝜕𝑓1 𝜕𝑦 2𝑥 = 𝜕𝑓2 −2𝑥 𝜕𝑦 2𝑦 1 Let the initial guess is 𝑥0 = 𝑦0 = 0.5 𝑓1,0 = 0.52 + 0.52 − 1 = −0.5 𝑓2,0 = 0.5 − 0.52 = 0.25 𝐽= 2𝑥 −2𝑥 2𝑦 1 1 = −1 1 1 𝑥1 1 1 −1 −0.5 0.5 𝑦1 = 0.5 − −1 1 0.25 𝑥1 0.8750 = 𝑦1 0.6250 Next iteration – 𝑥2 0.8750 1.75 1.25 = − 𝑦2 0.6250 −1.75 1 𝑥2 0.7909 = 𝑦2 0.6178 Solution converges at – 𝑥 0.7862 = 𝑦 0.6180 −1 0.15625 −0.14 11 MATLAB CODE FOR NEWTON RAPHSON METHOD: 12 MATLAB FUNCTION FSOLVE: Let, 𝑓1 = 𝑥 2 + 𝑦 2 − 1 = 0 and 𝑓2 = 𝑦 − 𝑥 2 = 0 First write a function to hold the equations Use function fsolve to determine the roots of equations. 13