Solve the differential equation.
which one?
x''+ax'+bx=c cos(wt)
@experimentX you can read people's minds?
He is skilled in that domain, yes.
x''+ax'+bx = 0 <-- this part gives the complimentary function as usual.
there's no question
for particular integral ... Write this as \[ (D^2 + aD +b)x = c \cos ( \omega t)\] case 1: you simply have no resonance. \[ \omega^2 + a\omega + +b \neq 0\] case 2: you have resonance \[ \omega^2 + a\omega + +b = 0\] let's do for case 1:
the auxiliary equation is \[ r^ 2 +ar + b = 0 \implies r = \frac{-a \pm \sqrt{a^2 - 4b}}{2}\] case 1.1: \( \sqrt{a^2 - 4b } \in \Re\) this case it is simply overdamped case 1.2 \( \sqrt{a^2 - 4b } \in \Im\) this case it is damped. the complimentary functions are 1.1 \( \large x = c_1 e^{\frac{-a + \sqrt{a^2 - 4b}}{2}}+c_2 e^{\frac{-a - \sqrt{a^2 - 4b}}{2}}\) 1.2 \( x = e^{-a \over 2} (c_1 \cos \omega_0t + c_2 \sin \omega_0 t)\) where \( \omega_0 = {\sqrt{ 4b - a^2}\over 2}\) is the frequency of system.
Woops ... worry here \[ \large x = c_1 e^{\left( \frac{-a + \sqrt{a^2 - 4b} }{2}\right)t}+c_2 e^{\left(\frac{-a - \sqrt{a^2 - 4b}}{2}\right)t} \] ^^ Note the above case that \( a> \sqrt{a^2-4b} \) and this always decays with time t. this is characteristics of decay equation. \[ \large x = e^{-a t\over 2} (c_1 \cos \omega_0t + c_2 \sin \omega_0 t) \] ^^ IN above case ... it harmonic but \( e^{-at/2}\) always damps it with passage of time.
For particular integral ... with the input type of Exponential \[ \large f(D) x = e^{at}\] \[ P.I. = \frac{e^{at}}{f(a)} \text{ if f(a) } \neq 0 \\ P.I. = \frac{t e^{at}}{f'(a)} \text{ if f'(a) } \neq 0 \\ ..\\ P.I. = \frac{t^n e^{at}}{f^n(a)} \text{ if } f^n(a)\neq 0 \\ \]
In case 1.2, is c_2 be imaginary?
nope: they are just constants of integration.
if you have complex solution then ... then the linear combination of both real and imaginary part (you have to exclude i) will give the general solution.
\[a^2-4b \le 0\] \[x=c_1e^{-0.5at+0.5i\sqrt{4b-a^2}}+c_2e^{-0.5at-0.5i\sqrt{4b-a^2}}\] \[x=e^{-0.5at}(c_1coswt+ic_1sinwt+c_2coswt-ic_2sinwt)\] \[x=e^{-0.5at}(Acoswt+Bsinwt)\]
you can check that ... \[ f(a + ib) = f(a) + if(b) = 0 \implies f(a) = 0, f(b) = 0 \text{ so both a and b are solution.} \]
\[* ((0.5i \pm \sqrt{4b-a^2})t)\]
\[ \large \cos ( \omega t) = \Re \left( e^{i \omega t}\right)\] you can substitute it to the P.I. above and get it's particular integral.
Is my logic correct for why B should be complex?
okay .. B is complex. do you know what is a linear map?
or linear function?
Yes, I understood your reasons why f(a)=0 and f(b)=0 are both solutions. But as this is to do with an oscillator, I guess you would disregard the complex part.
But I don't understand how you arrived at the earlier form for the PI
In this specific problem, for instance, how would you calculate the PI?
both \( x_1 \) and \( x_2 \) are solution to your system ... from superposition principle ... any linear combination of \( x_1 \) and \( x_2 \) should be solution to your system. so B doesn't necessarily have to be complex.
driven damped harmonic oscillator :)
for this particular problem ... it's quite easy. \[ P.I. = \frac{e^{at}}{f(a)} \text{ if f(a) } \neq 0 \\ P.I. = \frac{t e^{at}}{f'(a)} \text{ if f'(a) } \neq 0 \\ ..\\ P.I. = \frac{t^n e^{at}}{f^n(a)} \text{ if } f^n(a)\neq 0 \\ \] and \[ \large \cos ( \omega t) = \Re \left( e^{i \omega t}\right) \]
and \[ f(D) = D^ 2 + aD + b\] Woops!! should have put 'p' instead of 'a' since we have 'a' already.
So to calculate the PI, we need to find an f(a) such that:\[\frac{f'(a)}{f(a)}=t\]?
\[ P.I. = \Re \left(\frac{c e^{i \omega t}}{f(i \omega)} \right ) \text{ if } f(i\omega) \neq 0 \\ P.I. = \Re \left(\frac{c t e^{i \omega t}}{f'(i \omega)} \right ) \text{ if }f'(i \omega) \neq 0 \\ P.I. = \Re \left(\frac{c t^2 e^{i \omega t}}{f''(i \omega)} \right ) \text{ if } f''(i\omega)\neq 0 \\ \]
lol ... I didn't realize we would have 3 cases of PI
What is f(iw) in this case?
f(D) = D^2 + aD + b you put iw inplace of D and check if it's zero or not.
suppose this is not zero ... then your particular integral is simply \[ P.I. = \Re \left(\frac{c e^{i \omega t}}{(i \omega)^2 + a (i \omega) + b} \right )\]
this is not resonant case. if \( (i \omega)^2 + a (i \omega) + b = 0\), then you have resonance so your PI would be \[ P.I. = \Re \left(\frac{tc e^{i \omega t}}{2(i \omega) + a } \right ) \] this is destructive http://www.wolframalpha.com/input/?i=plot+t+sin%28t%29+from+0+to+8+pi
you can solve for PI with method of undetermined coefficients or reduction of order or variation of parameters the approach i used above is quite unintuitive you can find the proof here it's been discussed here in detail http://ocw.mit.edu/courses/mathematics/18-03sc-differential-equations-fall-2011/unit-ii-second-order-constant-coefficient-linear-equations/exponential-response/ the general proof is on this document. http://ocw.mit.edu/courses/mathematics/18-03sc-differential-equations-fall-2011/unit-ii-second-order-constant-coefficient-linear-equations/linear-operators-linear-time-invariance/MIT18_03SCF11_s17_6text.pdf from this chapter http://ocw.mit.edu/courses/mathematics/18-03sc-differential-equations-fall-2011/unit-ii-second-order-constant-coefficient-linear-equations/linear-operators-linear-time-invariance/ Also
try using try using http://en.wikipedia.org/wiki/Variation_of_parameters http://en.wikipedia.org/wiki/Reduction_of_order http://en.wikipedia.org/wiki/Method_of_undetermined_coefficients they are pretty straightforward and intuitive.
I'll have a look at those, but keep the question open just in case- Thank you
sure ... also try to put the equation on your first post.
Join our real-time social learning platform and learn together with your friends!