Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

1-dimensional problem in newtonian gravity- HELP

  1. Apr 11, 2005 #1
    1-dimensional problem in newtonian gravity- HELP!!

    The problem is this:

    Given sun of mass M and a body of mass m (M>>m) a distance r from the sun, find the time for the body to 'fall' into the sun (initially ignoring the radius of the sun).

    Our first equation is therefore [tex] \frac {d^2r}{dt^2} = \ddot{r} = \frac {GM}{r^2} [/tex].

    I am able to integrate this, giving:
    [tex] \dot{r} = - {\sqrt{2GM}}{\sqrt{1/r - 1/R}} [/tex],

    where R is the inital distance of the body from the sun. However, I am unable to integrate this again. I have shoved it into wolfram's integrator for an indicator of what to aim for, but cannot come close.

    Any thoughts would be greatly appreciated.


  2. jcsd
  3. Apr 11, 2005 #2


    User Avatar
    Staff Emeritus
    Science Advisor

    You can think of this as a zero-width elliptical orbit, so you can use Kepler's law relating the orbital period to the length semi-major axis, and then take 1/4 of the orbital period of any orbit with the same semi-major axis.

    If you insist on carrying out the integration, my textbook also suggests the substition

    r = a(1-e*cos(psi))

    in this case I think the eccentricity, e would be 1, so you'd have

    r = a(1 - cos(psi))

    I haven't tried this out, though.
  4. Apr 12, 2005 #3
    Thank you pervect, it's an approach I hadn't considered.


  5. Apr 12, 2005 #4


    User Avatar
    Science Advisor
    Homework Helper
    Gold Member
    Dearly Missed

    Aah, the triple post.
    This has already been solved in Diff-eq.
  6. May 7, 2005 #5
    Here is my try at it. Can you check for errors?
    A rock is dropped from [tex] r_0 [/tex] and falls straight down to R.
    How long does it take. Where R is earth radius and [tex]r_0 >> R [/tex].

    The gravity force equation [tex] F= \frac{GMm}{r^2} [/tex] ,

    integrates to a specific energy equation [tex] W= -\frac{u}{r}[/tex].

    The velocity at some intermediate r is given by:

    [tex] V=\sqrt{V_0^2+\frac{2u}{r} -\frac{2u}{r_0}}[/tex] .

    This is a hard differential equation to solve for r(t). I believe I have a form which resembles your integral.

    Change of symbols:

    [tex] A=2u,\ \ \ \ B = -\frac{2u}{r_0},\ \ \ \ V_0=0[/tex]

    [tex] V(r) = \sqrt{\frac{A}{r}+B}= \frac{dr}{dt}[/tex]

    Now the part which I have never trusted.

    [tex] \frac{dt}{dr}=\frac{1}{\sqrt{\frac{A}{r}+B}}[/tex]

    Then [tex] \int_R^r dt= \int_\frac{A}{R}^\frac{A}{r} \frac{1}{\sqrt{\frac{A}{r}+B}}\ \ dr [/tex]

    So [tex] T = \int_\frac{A}{R}^\frac{A}{r} \frac{1}{\sqrt{\frac{A}{r}+B}}\ \ dr [/tex]

    A change of variables [tex] \frac{A}{r}=x,\ \ r=\frac{A}{x},\ \ dr=\frac{-A}{x^2}dx [/tex]

    T = \int_\frac{A}{R}^\frac{A}{r} \frac{\frac{-A}{x^2}}{\sqrt{x+B}}\ \ dx = -A \left[ \frac{-\sqrt{B+x}}{Bx} -\frac{1}{2B} \left( \frac{2}{\sqrt{-B}} \arctan{ \sqrt{\frac{B+x}{-B}}} \right) \right]_\frac{A}{R}^\frac{A}{r} [/tex]

    [tex] = \left[ \frac{-r_0 \sqrt{x-\frac{2u}{r_0}}}{x} -\frac{2r_0}{\sqrt{\frac{2u}{r_0}}} \arctan{\sqrt{\frac{x-\frac{2u}{r_0}}{\frac{2u}{r_0}}}} \right]_\frac{2u}{R}^\frac{2u}{r_0} =

    [tex] = \frac{-r_0 \sqrt{\frac{2u}{r_0}-\frac{2u}{r_0}}}{\frac{2u}{r_0}} -\sqrt{\frac{2}{u}} r_0^\frac{3}{2} \arctan{\sqrt{\frac{\frac{2u}{r_0}-\frac{2u}{r_0}}{\frac{2u}{r_0}}}} \
    +\ \frac{r_0 \sqrt{\frac{2u}{R}-\frac{2u}{r_0}}}{\frac{2u}{R}} +\sqrt{\frac{2}{u}} r_0^\frac{3}{2} \arctan{\sqrt{\frac{\frac{2u}{R}-\frac{2u}{r_0}}{\frac{2u}{r_0}}}}

    [tex] = \frac{Rr_0}{\sqrt{2u}} \sqrt{\frac{1}{R}-\frac{1}{r_0}}+\sqrt{\frac{2}{u}} r_0^\frac{3}{2} \arctan{\sqrt{\frac{r_0}{R}-1}}= \sqrt{\frac{Rr_0}{2u}} \sqrt{r_0-R} +\sqrt{\frac{2}{u}} r_0^\frac{3}{2} \arctan{\sqrt{\frac{r_0}{R}-1}} [/tex]
    This of course means you have to work the problem backwards usually as an iteration, but I would use a binary search which I found converges rapidly for the Kepler problem. It should be ok here too.
    Last edited: May 7, 2005
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook