EngMaths I: ODEs week 3 EngMaths I: ODEs week 3 Video 4: A case where it goes wrong Alan Champneys & Dalila O’Grady 4. The nasty case of resonance EngMaths I: ODEs week 3 Looking back and looking forward this week so far ▶ given a 2nd-order constant coefficient inhomogeneous ODE a2 dx dx + a1 + a0 x = f (t) dt dt ▶ find solution as xCF (t) + xPI (t). ▶ and we guess xP I (t) by method of undetermined coefficients ▶ there is a table of which ansatz to guess for trig, exponentials, polynomials & sums and products thereof ▶ then we plug into the ODE to find the undetermined coefficients ▶ its a ‘trial and error’ method (if it works it works) this video ▶ a nasty case when the trial and error method doesn’t work ▶ how to get around it 4. The nasty case of resonance EngMaths I: ODEs week 3 A worked example In video 2 we solved x′′ + x = e−t . What if we had instead ẍ − x = e−t Solution in steps: 1. find the complementary function: ẍc − xc = 0 =⇒ xc = eλt =⇒ λ2 − 1 = 0 =⇒ xc = Aet + Be−t 2. make a ansatz for the particular integral xp = αe−t , for some unknown α 3. substitute into the ODE to find α: ẍp − xp = e−t =⇒ αe−t − αe−t = e−t =⇒ 0 = 1. oh poo! the method has failed. I can’t find α it’s because f (t) = e−t is already part of complementary function we call this resonance when f (t) solves the homogeneous ODE. 4. The nasty case of resonance EngMaths I: ODEs week 3 Resonance and how to get around this What we do is rather like the degenerate root case when constructing the complementary function (recall video 5 last week): instead of trying αe−t for our trial function we try a new ansatz xp (t) = αte−t . let’s see if that works: ẋp = α(1 − t)e−t , ẍp = α(t − 2)e−t substitute into the LHS of the ODE: L[xp ] = α(t − 2)e−t − αte−t = −2αe−t hence α = − 12 and the general solution is x(t) = xc (t) + xp (t) = Aet + Be−t − 12 te−t 4. The nasty case of resonance EngMaths I: ODEs week 3 Another example Example 2: Find the general solution to d2 x + 4x = sin(2t) d t2 PAUSE THE VIDEO and have a go, then I will show you how I do it note that we had to choose an ansatz αt sin(2t) + βt cos(2t). If we ignore the cos term, we do not get a solution. general rule: if we write equation as L[x(t)] = f (t) and f (t) solves L[f (t)] = 0 then multiply the trial function you would have used by t. Why does it work? Sorry, you’ll just have to trust me on this one. It works! 4. The nasty case of resonance EngMaths I: ODEs week 3 Updated list for xPI of L[x(t)] = f (t) Function on RHS = f (t) Ansatz eat cos(ωt) sin(ωt) at e cos(ωt) 1 t t2 teat t cos(ωt) Polynomial × Exponential Polynomial × Cos or Sin Poly × Exp × Trig αeat α cos(ωt) + β sin(ωt) α cos(ωt) + β sin(ωt) at αe cos(ωt) + βeat sin(ωt) α αt + β αt2 + βt + γ αteat + βeat αt cos(ωt) + βt sin(ωt) + γ cos(ωt) + δ sin(ωt) Polynomial × Exponential Poly × Cos + Poly × Sin Poly × Exp × Cos + Poly × Exp × Sin Sum of functions above Solves L[x(t)] = 0 Sum of corresponding ansatzes t × ansatz you would otherwise use 4. The nasty case of resonance EngMaths I: ODEs week 3 Summary and next steps Summary: in the method of undetermined coefficients for L[x(t)] = f (t) when we guess an ansatz for particular integral xPI (t) there’s a nasty case when f (t) already solves L[x(t)] = 0 (that is, f (t) is part of the complementary function xCF (t) then we have to try a t× the ansatz we would have chosen. and the algebra can get a bit more fiddly. this is called resonance as we explain in video 6. Confused? Try these videos WARNING, external content! MathTheBeautiful: https://youtu.be/n6FJ86zaM0A?si=GJkaf46rH3T4cTPp Matt Charnley https://youtu.be/RvAz3yZrhy0?si=1Fajxk4HVYDqmX9a Next videos fitting to the initial conditions engineering explanation of resonance Temporary page! LATEX was unable to guess the total number of pages correctly. As there unprocessed data that should have been added to the final page this e has been added to receive it. If you rerun the document (without altering it) this surplus page will go because LATEX now knows how many pages to expect for this documen