I'm trying to solve the following differential equation, that arises from conservation of energy in a physical problem. $R,k$ are constants. $$(1+R^4u^4)u'^2+(u-u_0)^2=k$$ Now, according to my book I should find "approximate solutions" around the point of equilibrium $u_0$. I started my reasoning with simpler equations and I have some questions:

  1. Is there a general good method to solve equations of the form $1/2k_1u'^2+1/2k_2u^2=k_3$? What I usually do is to resort to the physical problem of an harmonic oscillator with mass $k_1$, elastic constant $k_2$ and energy $k_3$ to find a cosinusoidal solution. I can find the amplitude but not the phase difference. In alternative I take the derivative of both sides of the equation and solve the second order linear differential equation that arises. I can find the amplitude of the oscillations by substitution of $u=A\cos t$ in the original equation. Are there more direct approaches?

  2. In the case of an equation of the form $1/2k_1u'^2+1/2k_2(u-u_0)^2=k_3$ I solve them either by inspection or by sobstituting $\xi = u-u_0,\ \xi'=u'$. Is that procedure correct?

  3. Now the original problem $(1+R^4u^4)u'^2+(u-u_0)^2=k$. My book gives what I think is a wrong answer : $u_0+(\sqrt{k}/u_0)\cos(\sqrt{1+R^4u_0^2})$. How to obtain the correct solution? I thought to substitute $u^4 = u_0^4$ in the leftmost term. Am I allowed to do this or it would be too rough an approximation? If I approximate even to first order $u^4$ the differential equation becomes too difficult for me


Request for clarification: is $u_0$ a constant or a function of independent variable? If it's a constant, then it appears that $k$ must be zero; otherwise $u\equiv u_0$ is not a solution.

Re: 1. There is a general approach to autonomous ODE of 1st order. It works for the simpler equations in 1 and 2, but for the original one the computations are probably complicated.

Re: 2. The substitution $u=u_0+\xi$ is reasonable. The equality $\xi'=u'$ is correct as long as $u_0'=0$, see my request for clarification above.

Re: 3. Again, writing $u=u_0+\xi$ is a reasonable place to begin. If you are willing to neglect some "small" terms with $\xi$, the equation may become manageable.

More precisely, in the equation $(1+R^4 (u_0+\xi)^4)(\xi')^2+\xi^2=k$ we may decide to neglect all terms in which $\xi$ and its derivative appear to power higher than 2. This leaves us with $(1+R^4 u_0^4)(\xi')^2+\xi^2=k$, which you know how to solve. Yes, the solution in the book appears to be wrong. I expect a trigonometric function with argument $x/\sqrt{1+R^4 u_0^4}$ where $x$ is the independent variable.

  • $\begingroup$ You are right. $u_0$ is a constant. Physically it is related to the energy of the system, so for "equilibrium solution" I intend the solution when $k$ is zero. I'm interested in the solutions where $k$ is small $\endgroup$ – Ralph Jul 15 '12 at 21:07
  • $\begingroup$ @Ralph I expanded my answer. $\endgroup$ – user31373 Jul 15 '12 at 21:51

You can start with that point:

$$A(u) u'^2 +B(u)=k$$

$$A(u) u'^2 =k-B(u)$$

$$\frac{\sqrt{A(u)}}{\sqrt{k-B(u)}} u' =1 $$

$$\frac{\sqrt{A(u)}}{\sqrt{k-B(u)}} du =dx $$

Did you try that way?

  • $\begingroup$ I think this is the exact solution, but my problem is to find a approximate solution expressed with elementary functions $\endgroup$ – Ralph Jul 15 '12 at 18:44

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.