3
$\begingroup$

I've got the equation which I am supposed to use laplace transforms to solve:

$$ y'' + y = e^t + tsin(t), ~~~~~y(0) = 3/2,~~~~~y'(0) = 5/2 $$

Using laplace transforms, I come up with the equation:

$$ Y(s) = {1 \over (s-1)(s^2+1)} + {2s \over (s^2+1)^3} + {3s+5 \over 2(s^2+1)}$$

I can use partial fraction and inverse laplace to solve most if it:

$$ y(t) = \frac{1}{2}e^t +cos(t) + 2sin(t) + L^{-1} \left[\frac{2s}{(s^2+1)^3} \right]$$

I've been trying to use convolution:

$$ L^{-1} \left[ \frac{s}{(s^2+1)^2} \right] = \frac{t}{2}sin(t) ~~~~~L^{-1} \left[ \frac{2}{(s^2+1)} \right] = 2sin(t) $$

$$ \left(2sin(t) * \frac{t}{2} sin(t)\right) = \int_0^t \frac{\tau}{2}sin(\tau) \cdot 2sin(t-\tau) d\tau$$

I just can't seem to get the two terms I'm missing. Is there a better way to approach the laplace or convolution? Or do I just need to make sure I'm doing the integral correctly?

Symbolab answer to check for correctness:

$$y=-\frac{1}{4}t^2\cos \left(t\right)+2\sin \left(t\right)+\frac{e^t}{2}+\frac{1}{4}t\sin \left(t\right)+\cos \left(t\right)$$

$\endgroup$
1
  • $\begingroup$ I didn't show the complete partial fractions, but that is what I got as well. If you take the inverse laplace of the first 3 terms here you will get the same as my third equation. The problem area is the term $ \frac{2s}{(s^2 +1)^3}$ $\endgroup$ – Arod529 Nov 19 '16 at 23:19
2
$\begingroup$

What if we write:

$$q(s) = \dfrac{2s}{(s^2+1)^3} = -\dfrac{1}{2} \dfrac{d}{ds} \dfrac{1}{(s^2+1)^2}$$

Now, we ca write:

$$\mathcal{L}^{-1} (q(s)) = \mathcal{L}^{-1} \left(-\dfrac{1}{2} \dfrac{d}{ds} \dfrac{1}{(s^2+1)^2} \right)$$

Let's do it with a Laplace Table first, items $11$ and $30$:

$$ \mathcal{L}^{-1} \left(-\dfrac{1}{2} \dfrac{d}{ds} \dfrac{1}{(s^2+1)^2} \right) = \dfrac{1}{4} t\left(\sin t - t \cos t\right)$$

We can also use the convolution method that you tried.

$$ \mathcal{L}^{-1} \{ F(s) G(s) \} = (f * g)(t)$$

We set (try other variants):

$F(s) = \dfrac{2}{(s^2+1)^2} \implies \mathcal{L}^{-1}(F(s)) = f(t) = \sin t -t \cos t \\ G(s) = \dfrac{s}{s^2+1} \implies \mathcal{L}^{-1}(G(s)) = g(t) = \cos t$

We now have:

$$\displaystyle \int_0^t f(t-v) g(v)~dv = \int_0^t f(v) g(t-v)~dv = \dfrac{1}{4} t( \sin t- t \cos t)$$

Looks like you know how to do the rest of the problem.

Update

For the integral, we have $f(v) g(t-v) = \cos (t-v) (\sin v -v \cos v)$. When we integrate, we get:

$$\displaystyle \int \cos (t-v) (\sin v -v \cos v)~dv = \dfrac{1}{8} \left(-2 v^2 \cos t+2 v (\sin (t-2 v)+2 \sin t)-3 \cos (t-2 v)\right)$$

Evaluating this at $v = 0$ and accounting for subtraction, we get:

$$\dfrac{3}{8} \cos t$$

Evaluating this at $v = t$ and accounting for subtraction, we get:

$$\dfrac{1}{8} \left(-2 t^2 \cos t+2 t \sin t-3 \cos t\right) $$

Adding those two, gives the result above.

$\endgroup$
2
  • $\begingroup$ Apparently the the equivalent to item 30 in my book is not written in general form... Using the tables is much more convenient than convolution. I'm making a mistake somewhere amongst all the trig identities and integration by parts; I'll just have to keep working through it. $\endgroup$ – Arod529 Nov 20 '16 at 0:53
  • $\begingroup$ After doing the integration $\int_0^t f(v)g(t-v) dv = cost~cosv~sinv+sint~sin^2v-v~cost~cos^2v-v~sint~sinv~cosv~dv$ I am left with two terms I can't seem to cancel: $-\frac{t~sin2t}{4}$ and $\frac{t~sint~cos2t}{4}$. They are part of the result of the 3rd and 4th inintegration terms respectively. $\endgroup$ – Arod529 Nov 20 '16 at 2:54

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.