Perturbation treatment of the anharmonic oscillator Suppose we have a third order non-linear term in the equation of motion of a particular system such as an oscillating pendulum: ψ̈ + ω02 ψ − cψ 3 = 0 g 1 ω02 = , c = ω02 l 6 If the anharmonic term ψ 3 were ignored the solution would be in the form ψ ' sin ω0 t. Since now the equation of motion involves the cube of the displacement we seek a trial solution in an additive form written as a superposition of two different motions, one in sin ωt, other in sin 3ωt. It should be clear that sin 3ωt has been included as suggested by sin3 x = 3 1 sin x − sin 3x. 4 4 In fact, the solution of the unperturbed part is in the form sin ωt the cube of which generates sin 3ωt and the differential equation takes the form A sin ωt + B sin 3ωt = 0 which cannot be satisfied for any t unless A and B are identically zero. Thus, so as to cancel the term sin 3ωt generated by ψ 3 we are forced to add to sin ωt a term such as sin 3ωt. Going further, the new sin 3ωt in the trial solution generates a term in 3 sin 9ωt, and so on. However, if 1 the series converges very rapidly. Therefore, we retain the first order correction and ignore terms like 3 , 9 , etc. We thus write an approximate solution of the form ψ = ψ0 {sin ωt + sin 3ωt} where is a dimensionless constant much less than unity for when | ψ0 | 1. Substituting the trial solution into the equation of motion: ψ̈ = −ω 2 ψ0 sin ωt − 9ω 2 ψ0 sin 3ωt ω02 ψ = ω02 ψ0 sin ωt + ω02 ψ0 sin 3ωt 1 1 1 1 − ω02 ψ 3 = − ω02 ψ03 sin ωt + ω02 ψ03 sin 3ωt − ω02 ψ03 sin2 ωt sin 3ωt 6 8 24 2 Equating the respective coefficients of sin ωt and sin 3ωt we obtain 1 −ω 2 + ω02 − ω02 ψ02 = 0 8 or 1 ω 2 = ω02 {1 − ψ02 } 8 1 ω ' ω0 {1 − ψ02 } 16 and further −9ω 2 + ω02 + 1 2 2 ω ψ =0 24 0 0 1 Setting ω ' ω0 , we obtain = 1 2 ψ 192 0 which is a measure of the fractional contribution of the anharmonic effect to the solution of the unperturbed equation of motion. For ψ0 = 0.3 rad, we have ' 0.001, which indeed is very small. The reason why a term in sin 2ωt were not included in the trial solution is that if a solution of the form ψ = ψ0 {sin ωt + η sin 2ωt} were tried we would find η = 0. The pendulum generates chiefly third harmonics, i.e., terms in sin 3ωt, and no terms in sin 2ωt. The situation would be different for a device for which the equation of motion included a term in ψ 2 . In such a case the solution will have a term in sin 2ωt, and the same technique can be used. 2