Is It Possible to Use the Method of Undetermined Coefficients Here?

I'm doing a practice problem I found online, and I get a solution, but I think it should have a sine term in it. I looked up the solution, and most sites say to use variation of parameters, but is it possible to use the method of undetermined coefficients?



The problem is as follows: y'' + 4y = 8cos2t



r^2 + 4 = 0 ----> r^2 = -4 -----> r = +/- 2i



yh = c1*sin2t + c2*cos2t



Particular solution has form At*sin2t + Bt*cos2t (because you add a "t" to both terms so they don't clash with the cos2t or sin2t terms)



First derivative is -2At*sin2t + A*cos2t + 2Bt*cos2t + B*sin2t



Second derivative is -4At*cos2t - 2A*sin2t - 2A*sin2t - 4Bt*sin2t + 2B*cos2t + 2B*cos2t



So then you add that term to the particular solution and get:



-4At*cos2t - 2A*sin2t - 2A*sin2t - 4Bt*sin2t + 2B*cos2t + 2B*cos2t + 4At*sin2t + 4Bt*cos2t



Canceling out terms leaves you with -2A*sin2t - 2A*sin2t + 2B*cos2t + 2B*cos2t = 8*cos2t.



Now the problem is that 8*cos2t is really 8*cos2t + 0*sin2t. Solving for B gives you B = 2, but that means A = 0. But the final solution should have a sine term in it. (And I looked up the solutions, and they do have a sine term)



I thought you could use the Method of Undetermined Coefficients when you were dealing with a sine/cosine function, a regular polynomial, or an exponential function. (I think the fourth one was a function of e^t. I forget off the top of my head.)