# Solving a fifth order non-homogeneous differential equation

Tags:
1. Mar 17, 2017

### JMFL

1. The problem statement, all variables and given/known data
Find the general solution of $y^{(5)}-y(1)=x$

3. The attempt at a solution
I found the complementary function by substitution of the solution form $y=e^{kx}$ giving $k=0,1,-1,i,-i$, so $y_{cf}=a_0+a_1e^x+a_2e^{-x}+a_3e^{ix}+a_4e^{-ix}$

Now for the particular integral, the general trial solution form of a forcing term of x on the right is $y=b_0+b_1x$

However if I plug this in, I get $-b_1=x$ !

I admit that I am not too familiar with dealing with differential equations of order greater than two. It seems that the general form of particular integrals when you have higher order ODEs has to be different? Or is the problem the fact that I already have a constant term in my complementary function sot here will not be a constant term in the particular integral too?

Last edited by a moderator: Mar 17, 2017
2. Mar 17, 2017

### Staff: Mentor

I assume you mean $y^{(5)} - y' = x$
For the last two, it's easier to use multiples of $\sin(x)$ and $\cos(x)$.
Your particular solution won't work, as $y_p' = b_1$. Instead, try $y_p = b_0 + b_1x + b_2x^2$
BTW, you are not using the tex tags correctly. You have the opening tag right, but the closing tag is [/itex]. You are omitting the slash character. I fixed them in your original post, but didn't in what I copied from your post.

Last edited: Mar 18, 2017
3. Mar 17, 2017

### Ray Vickson

You actually have a 4th order differential equation in the variable $z(x) = y'(x) = dy(x)/dx$. Then $y(x) = C+\int_0^x z(t) \, dt.$

4. Mar 18, 2017

### haruspex

So try adding a b2x2 term.

5. Mar 18, 2017

### JMFL

Hi haruspex,

I was considering doing this, but I wanted to know why this form of the particular integral is not working. I did not wish to be blindly guessing particular integral forms or adding more x's without knowing why I was doing what I was doing!

Why does the particular integral form fail? Are the PI forms that I am used to only for second order ODEs?

6. Mar 18, 2017

### haruspex

I don't know where your rule of just using a linear term comes from. I've always regarded finding PIs as a combination of experience and imagination. If you want a general rule it would be more like plugging in an entire power series with unknown coefficients and seeing if anything useful results. Either a simple equation or a recurrence relation.

7. Mar 18, 2017

### pasmith

The suggestion of $y_p(x) = ax + b$ in the case of $y'' + py' + qy = x$ has the property that $y''_p = 0$, so you need only worry about satisfying $py'_p + qy_p = x$.

Generalizing that, here you should probably look for something whose fifth derivative is zero, ie. a fourth order polynomial, so you need only worry about satisfying $-y_p' = x$.

8. Mar 18, 2017

### Staff: Mentor

The technique of annihilators provides a more direct way of dealing with nonhomogeneous linear differential equations. I wrote a two part Insights series on using the annihilator method. You can find them by searching amongst the Insights articles, under Tutorials.

There is also this short article: http://jekyll.math.byuh.edu/courses/m334/handouts/annihilator.pdf

9. Mar 18, 2017

### Ray Vickson

If you had paid attention to post #3 you would have seen the solution immediately: the equation for $z = y'$ is $z^{(4)} - z = x$. Isn't $z = -x$ an obvious solution? Now integrate to get $y$.