ENGG2013 Unit 25 Second-order Linear DE Apr, 2011. Yesterday • First-order DE – Method of separating variable – Method of integrating factor • System of first-order DE – Eigenvalues determine the convergence behaviour near a critical point. • Objectives: – Solve the initial value problem • Given the initial value, find the trajectory • Transient-state analysis of electronic circuits – Understand the system behaviour • Does the system converge? • Stable equilibrium point, unstable equilibrium point. kshum 2 Linear Second-order DE • Homogeneous • Non-homogeneous kshum 3 Linear Second-order constant-coeff. DE • Homogeneous • Non-homogeneous kshum 4 Vibrating spring without damping • x(t): vertical displacement • Hooke’s law: Force = k x – k is the spring constant, k > 0 (the constant k is sometime called the spring modulus.) kx m x’’ • Newton’s law: F = m x’’ – m is the mass, m > 0 m x’’ + k x = 0 Assumptions: • Spring has negligible weight • No friction kshum x Second-order, autonomous linear, constant-coefficient and homogeneous 5 • Solutions to undamped springmass model Normalize by m • Direct substitution verifies that are solutions, e.g. kshum and 6 Natural frequency • Let • We know that cos( t) and sin( t) are both solutions. • is called the natural frequency. • Dimension check: The unit of is Hz = s-1. – The spring constant k has unit kg s-2. – k/m has unit s-2. – Square root of k/m has unit s-1. kshum 7 Principle of superposition (aka linearity principle) • For linear and homogenous differential equation, the linear combination of two solutions is also a solution. • For any real numbers a and b, a cos( t) + b cos( t) is a solution to x’’+ 2 x=0. kshum 8 GRAPHICAL METHOD kshum 9 • Graphical illustration of springmass model Define the displacement-velocity vector • Reduction to system of two first-order differential equations. kshum 10 Phase plane for the vibrating spring x'=v v ' = - k x/m k=3 m=1 Sample solution 5 4 3 2 v 1 0 -1 -2 -3 -4 -5 -5 -4 -3 -2 -1 0 1 2 3 4 5 x The trajectory is a ellipse kshum 11 Order reduction technique equivalent Second-order DE with constant coefficients is basically the same as a system of two first-order DE. kshum 12 Vibrating spring with damping • Vibrating spring in honey m x’’ = – k x – d x’ honey Force exerted by the spring Damping due to viscosity Equivalent form kshum Assumption: Magnitude of damping force is directly proportional to x’. d> 0 13 Phase plane for damped spring x'=v v ' = - (k x + d v)/m m=1 d=1 k=3 5 Sample solutions 4 3 2 Convergence to the origin v 1 0 -1 -2 -3 -4 -5 -5 -4 -3 -2 -1 0 1 2 3 4 5 x kshum 14 METHOD OF DIAGONALIZATION kshum 15 Recall: Diagonalization • Definition: an n n matrix M is called diagonalizable if we can find an invertible matrix S, such that is a diagonal matrix, or equivalently, • Example: • Diagonalization is useful in decoupling a linear system for instance. Solution to vibrating spring with damping Characteristic equation Eigenvalues Solution to vibrating spring with damping Eigenvectors have complex components Two eigenvectors are not scalar multiple of each other, because the two eigenvalues are distinct . Hence inverse exists. Concatenate Diagonalize Solution to vibrating spring with damping Substitute by the diagonalized matrix Change of variables An uncoupled system Solution to vibrating spring with damping Solve the uncoupled system K1 and K2 are constants Substitute back (C1, C1, a and b are constants) Solution to vibrating spring with damping 1 x 0.5 0 -0.5 -1 0 2 6 4 time 8 10 A general strategy for linear system Decouple Linear system (Diagonalization) Solve each subsystem separately Solved Piece them together METHOD OF GUESS-AND-VERIFY kshum 23 2nd-order constant-coeff. DE • Homogeneous – b and c are constants. • Non-homogeneous – b and c are constants. – f(t) is a function of independent variable t. kshum 24 The homogeneous case • Idea: try a function in the form ekt as a solution. – k is some constant. • Substitute ekt into x’’+bx’+cx=0 and try to solve for the constant k. • Apply the superposition principle: any linear combination of solutions is also a solution. kshum 25 Examples 1. Solve x’’+3x’+2x=0. General solution: x(t) = c1 e–2t+ c2e–t 2. Solve x’’–4x’+4x=0. General solution: x(t) = c1 e2t+ c2 t e2t 3. Solve x’’+9x= 0. General solution: x(t) = c1 e3i t+ c2 e-3i t = d1 sin(t)+ d2 cos(t) kshum 26 Summary of the three cases Differential equation Characteristic equation Case Roots Basis of solutions General Solution 1 Distinct real and et, et c1et+c2et 2 Repeated root et, tet c1et+c2tet 3 Complex roots =r+i, =r–i e(r+i)t, e(r–i)t er(c1cos t+c2sin t) kshum 27 The non-homogeneous case • Property: If x1(t) and x2(t) are two solutions to then their difference x1(t) – x2(t) is a solution to the homogeneous counterpart kshum 28 Consequence • Suppose that xp(t) is some solution to x’’+bx’+cx=f(t) (given to you by a genie for example) • Any solution of x’’+bx’+cx=f(t) can be written as xh(t) +xp(t) A solution to the homogeneous DE x’’+bx’+c=0 A particular solution kshum 29 Method of trial and error (aka as the method of undetermined coefficient) To solve the non-homogeneous DE x’’+bx’+cx=f(t) 1. Find a particular solution by trial and error (and experience) – Let the particular solution be xp(t). 2. Solve the homogeneous version x’’+bx’+cx=0. – Let the homogeneous solution be xh(t). 3. The general solution is xh(t)+ xp(t). kshum 30 How to guess a particular solution f(t) Choice for xp(t) K xn cnxn+cn-1xn-1+…+c1x1+c0 K eat Ceat K sin(t) c1 sin(t)+c2 cos(t) K cos(t) c1 sin(t)+c2 cos(t) K eat sin(t) eat (c1 sin(t)+c2 cos(t)) K eat cos(t) eat (c1 sin(t)+c2 cos(t)) K, C, a, , c0, c1, c2,… are constants kshum 31 Example • Solve x’’+3x’+2x= e5t. • Try c e5t as a particular solution. (c e5t )’’+3(c e5t )’+(c e5t )= e5t 25c e5t+15c e5t+5c e5t= e5t 25 c + 15c + 5c=1 c = 1/45 Let xp(t) = e5t/45 as a particular solution. General solution is c1e–2t+ c2e–t + e5t/45 Homogeneous solution kshum Particular solution 32 Summary • Graphical method using phase plane. – Reduction to two 1st-order linear DE. • Method of diagonalization – Need to reduce the second-order DE to a system of first-order DE. – Time-consuming but theoretically sound. • Method of undetermined coefficients – Find a solution quickly, but not systematic. – Good for calculation in examination. kshum 33 Final Exam • • • • Date: 6th May (Friday) Venue: NA Gym Time: 9:30~11:30 Coverage: Everything in Lecture Notes and Homeworks • Close-book exam • You may bring a calculator, and a handwritten A4-size and double-sided crib sheet. kshum 34