PDA

View Full Version : solve x'' = cox(x)


peacex
Feb10-07, 07:14 AM
im new to 2nd order differential eqns and im getting myself all confused.
ive tried so hard but i just cant solve it.
can someone please help me with tips to solve/solutions?

x'' = cos(x)

AlphaNumeric
Feb10-07, 07:20 AM
Try using \ddot{x} = \dot{x}\frac{d\dot{x}}{dx}

That will allow you to turn it into a first order equation at the very least.

peacex
Feb10-07, 07:29 AM
thankyou!! i didnt think of that!

dextercioby
Feb12-07, 01:06 AM
Try using \ddot{x} = \dot{x}\frac{d\dot{x}}{dx}

That will allow you to turn it into a first order equation at the very least.

I don't think that's correct.
You should have written

\ddot{x}=\frac{1}{\dot{x}}\frac{d}{dt}\left(\frac{ \dot{x}^{2}}{2}\right)

HallsofIvy
Feb12-07, 06:57 AM
No, Dextercioby, the first response was correct. If you let u= dx/dt, then du/dt= d2x/dt2. But we can apply the chain rule to the left side: du/dt= (du/dx)(dx/dt)= u du/dx since u= dx/dt. d2u/dt2= u du/dx.

This method is called "quadrature" (because the integral involves u2) and can be used to reduce the order of a differential equation when the independent variable, t, does not appear explicitely in the equation.

x.users
Mar30-07, 06:40 PM
x" = cosx

its x-missing

v = x'

v' = x" = v dv/dx

v dv/dx = cosx

vdv = cosx dx

v^2 = 2 sinx

(x')^2 = 2 sinx

x' = ( 2 sinx )^1/2

dx / (sinx )^1/2 = 2^1/2 dy

Q = 2^1/2 y + C


such that Q = http://integrals.wolfram.com/index.jsp

Matthew Rodman
Mar30-07, 06:47 PM
You cannot neglect the constant of integration.

if

y^{\prime \prime} = \cos{y}

multiply by y^{\prime}

gives you

y^{\prime} y^{\prime \prime} = y^{\prime} \cos{y}

integrate, and you get

\frac{1}{2}y^{\prime 2} = \sin{y} + C

Note the constant of integration which must not be ignored.

Therefore, you have

y^{\prime} = \sqrt{2 \sin{y} + C}

which isn't easily soluble -- the only way to get an answer is to use elliptic integrals -- have a look at this Wolfram page. (http://mathworld.wolfram.com/EllipticIntegraloftheFirstKind.html)

dextercioby
Apr4-07, 05:08 AM
No, Dextercioby, the first response was correct.

It can't be, since \frac{d\dot{x}}{dx}=0 .

HallsofIvy
Apr4-07, 05:26 AM
Where in the world did you get that idea?

If x(t)= et, then x'= et= x. In other words,
\frac{dx'}{dx}= 1
not 0.