Find all solutions of $$y'''-4y''+5y'-2y=-x^2+5x+2.$$

I know how to find the solutions of the corresponding homogenous differential equation $y'''-4y''+5y'-2y=0$. I've done that in the following way:

The characteristic equation is $$P(x) = x^3 - 4x^2 +5x-2=(x-1)^2(x-2).$$

So if we put $$f_1:x\mapsto e^x\\f_2:x\mapsto xe^x\\f_3:x\mapsto e^{2x}$$ then the solution basis is $$\{f_1, f_2, f_3\}$$ and all solutions are linear combinations of this basis.

However, I am not sure how to deal with the inhomogenous part of the differential equation.

Please share the general approach with me.

  • $\begingroup$ I suppose a typo in the rhs. Please check and fix. $\endgroup$ – Claude Leibovici Aug 25 '14 at 6:55
  • $\begingroup$ Fixed it, my apologies :) $\endgroup$ – rehband Aug 25 '14 at 6:57
  • $\begingroup$ Go and learn some Laplace transform. It should be given by a convolution. $\endgroup$ – Troy Woo Aug 25 '14 at 6:57
  • $\begingroup$ $f_3$ is probably $e^{2x}$ $\endgroup$ – Claude Leibovici Aug 25 '14 at 7:00

Use the method of undetermined coefficients. See second case here. You can try this out.

Essentially, you guess y looks like a polynomial e.g. $y = a_0+a_1x+a_2x^2+...+a_nx^n$ because that is what the RHS is. Then, plug in y.

So the question is what n to use?

Try 5. The highest order derivative is 3 and the degree of the RHS is 2. is y has $a_5x^5$ and is differentiated thrice, it will have an $x^2$. If it doesn't have a $x^5$, the coefficient will turn out $a_5=0$



It may not be a very academic solution but since you found the solution for the homogenous part, let us set, because the rhs is a polynomial, that $$y=c_1 e^x+c_2 x e^{x}+c_3 e^{2x}+P(x)$$ Differentiate three times and substitute. You end with $$P'''(x)-4 P''(x)+5 P'(x)-2 P(x)=-x^2+5x+2$$

Try $P(x)=a+bx+x^2+dx^3$ and identify.

  • $\begingroup$ You sure P(x) of degree 3 is the right guess? math.stackexchange.com/a/908413/140308 $\endgroup$ – BCLC Aug 25 '14 at 7:48
  • $\begingroup$ Not for sure ! It is clear that degree $2$ is clear. $\endgroup$ – Claude Leibovici Aug 25 '14 at 7:52
  • $\begingroup$ What? I mean it looks like the worst case scenario is that the third derivative is of degree 2 so why not try 5...? $\endgroup$ – BCLC Aug 25 '14 at 8:00
  • 1
    $\begingroup$ @BCLC. You could easily show that $P(x)$ must be a polynomial of the same degree of the highest power in the rhs because the lhs is a linear combination of the derivatives. $\endgroup$ – Claude Leibovici Aug 25 '14 at 8:03
  • $\begingroup$ Well Wikipedia seems to agree... en.wikipedia.org/wiki/… $\endgroup$ – BCLC Aug 25 '14 at 9:54

For the particular solution, I usually just guess. For this particular problem, a quadratic polynomial $y(x)=ax^2+bx+c$ seems like a reasonable candidate (edit: because the RHS is also a quadratic polynomial). If you play around with the arithmetic, you will see that $$ y(x)=\frac{1}{2}x^2-3 $$ works.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.