I am told that equation for a scalar field is:

$$\ddot{\phi}+3 H \dot{\phi}+V^{\prime}(\phi)=0$$

where $H$ is defined as: $H=\dot{R} / R$

For a potential $V(\phi)=V_{0} \exp (-\lambda \phi)$, I am supposed to show that:


and that

$$\phi(t)=\frac{2}{\lambda} \ln \left[\left(\frac{V_{0}}{2\left(6-\lambda^{2}\right)}\right)^{1 / 2} \lambda^{2} t\right]$$

I can verify that these solutions indeed satisfy the equation. How would I go about solving it directly though? (I am happy with complete solutions, but hints are also appreciated.)

Verification of the provided solution


$$\dot{\phi}(t)=\frac{2}{\lambda} t^{-1}$$

$$\ddot{\phi}(t)=-\frac{2}{\lambda} t^{-2}$$

$$\dot{R}(t)=R_{0}\left(\frac{1}{t_{0}}\right)^{\frac{2}{\lambda^{2}}} \frac{2}{\lambda^{2}} t^{\frac{2}{\lambda^{2}}-1}$$

$$\frac{\dot{R}}{R} \dot{\phi}=\frac{R_{0}\left(\frac{1}{t_{0}}\right)^{\frac{2}{\lambda^2}} \frac{2}{\lambda^{2}} t^{\frac{2}{\lambda^{2}}-1}}{R_{0}\left(\frac{t}{t_{0}}\right)^{\frac{2}{\lambda^{2}}}} \frac{2}{\lambda} t^{-1}=\frac{4}{\lambda^{3}} t^{-2}$$

Then we can substitute into original equation:

$$\ddot{\phi}+\frac{3 \dot{R}}{R} \dot{\phi}-\lambda V_{0} e^{-\lambda \phi}=$$ $$=-\frac{2}{\lambda} t^{-2}+3 \frac{4}{\lambda^{3}} t^{-2}-\lambda \operatorname{V_0exp}\left(-\lambda \frac{2}{\lambda} \ln \left[\left(\frac{V_{0}}{2\left(6-\lambda^{2}\right)}\right)^{1 / 2} \lambda^{2} t\right]\right)$$

$$=-\frac{2}{\lambda} t^{-2}+\frac{12}{\lambda^{3}} t^{-2}-\lambda V_{0}\left[\left(\frac{V_{0}}{2\left(6-\lambda^{2}\right)} \right)^{\frac{1}{2}} \lambda^{2} t\right]^{-2}$$

$$=-\frac{2}{\lambda} t^{-2}+\frac{12}{\lambda^{3}} t^{-2}-\lambda V_{0}\left( \frac{V_{0}}{2\left(6-\lambda^{2}\right)} \right)^{-1} \lambda^{-4} t^{-2}$$

$$=-\frac{2}{\lambda} t^{-2}+\frac{2 \lambda^{2}}{\lambda^{3}} t^{-2}=0$$

as wanted.

My thoughts:

  • I think I would be able to solve for $\phi$ if $H$ was constant. Would probably resort to writing $\phi$ as $\phi=\sum a_n x^n$ & then find $a_n$ coefficients, which is probably not the best way but I think it would work.

  • I am not used to solve ODEs where I have two functions to worry about ($\phi$ & $R$). Is this ODE a certain class of ODE which I should look up?

  • 1
    $\begingroup$ Your doubts are justified, there is nothing that binds $R$. Essentially, $R$ via $H$ determines the friction or energy dissipation in the mechanical motion of $ϕ$. Conceptually, there is nothing that gives informations restricting $R$. $\endgroup$ Mar 12 at 11:06
  • $\begingroup$ Thank you. Does that mean that I need something else as well to get to the desired result? $\endgroup$
    – zabop
    Mar 12 at 11:10
  • $\begingroup$ As I read so far, the scalar field is some kind of "magical" background mass density in the evolution of space-time. I think you need some more of that larger picture to make sense of these equations. $\endgroup$ Mar 12 at 12:33
  • $\begingroup$ Yes, I agree. Making sense of them would be nice, but isn't that issue separate from the issue of being able to solve them? $\endgroup$
    – zabop
    Mar 12 at 17:26
  • $\begingroup$ The issue is to have enough equations for the given variables. As there are 2 unknown functions, one would need 2 equations. Of course one could take the equation for $R$ as a second equation, but from your formulation, you have to derive that from some more basic principles. $\endgroup$ Mar 19 at 12:09

You are missing one key equation to close the system.

A scalar field carries energy and its this energy that determines how fast the Universe expands. For a canonical scalar field the kinetic energy is $\frac{1}{2}\partial_\mu \phi\partial^\mu\phi$ which is just $\frac{1}{2}\dot{\phi}^2$ in a smooth Friedmann universe. The potential energy is $V(\phi)$ so the Friedmann equation (in Planck units: $8\pi G = 1$) is $$H^2 = \left(\frac{d\log R}{dt}\right)^2 = \frac{1}{3} \rho_\phi = \frac{1}{3}\left(\frac{1}{2}\dot{\phi}^2 + V(\phi)\right)$$ when the scalar field is the only source of energy. Thus the full system reads $$H^2 = \frac{1}{3} \rho_\phi = \frac{1}{3}\left(\frac{1}{2}\dot{\phi}^2 + V(\phi)\right)$$ $$\ddot{\phi} + 3H\dot{\phi} + V'(\phi) = 0$$ Only for very special classes of potentials do we have analytical solutions. A general method of studying such systems is to write them as a coupled set of first order equations - a so-called dynamical system.

If we introduce $x_1 = \frac{\dot{\phi}}{\sqrt{6}H}$ and $x_2 = \frac{\sqrt{V}}{\sqrt{3}H}$ then the Hubble equation says that $x_1^2 + x_2^2 = 1$. We further find with $x = \log R$ as our time-coordinate: $$\frac{d x_1}{dx} = -3x_1 + \frac{\sqrt{6}}{2}\lambda_\phi x_2^2 - x_1 \frac{d\log H}{dx}$$ $$\frac{d x_2}{dx} = -\frac{\sqrt{6}}{2}\lambda_\phi x_1x_2 - x_2 \frac{d\log H}{dx}$$ $$\frac{d\lambda_\phi}{dx} = -\sqrt{6}\lambda_\phi^2(\Gamma-1)x_1$$ where $$\frac{d\log H}{dx} = -\frac{3}{2}(1+x_1^2 - x_2^2)$$ $$\lambda_\phi = -\frac{V'}{V},\,\,\,\Gamma = \frac{V''V}{V'^2}$$ Now for an exponential potential $\lambda_\phi = \lambda$ stays constant and $\Gamma = 1$ and the system simplifies greatly down to $$\frac{d x_1}{dx} = -3x_2^2[x_1 - \frac{\lambda}{\sqrt{6}}]$$ $$\frac{d x_2}{dx} = 3x_1x_2[x_1 - \frac{\lambda}{\sqrt{6}}]$$ or simply $$\frac{d x_1}{dx} = -3(1-x_1^2)[x_1 - \frac{\lambda}{\sqrt{6}}]$$ This is possible to integrate though its hard to invert the resulting function, but we see right away that if $x_1 \equiv \frac{\lambda}{\sqrt{6}}$ then we have one solution. This is a so-called fixpoint of this system (put derivatives to zero and solve the resulting algebraic system). Finding and classifying such fixpoints (i.e. are they attractive or repellent) is super useful (much more than trying to find analytical solutions that might not exist) for understanding the dynamics of the system and the given analytical solution is such a fixpoint. We can also see that the fixpoint is attractive so even if we did not start with this initial value of $x_1$ we would be driven towards it and the system would settle close to it.

If you now unwrap this solution you will find the given solution. For example $\frac{d\log H}{dx} = -3x_1^2 = -\frac{\lambda^2}{2}$ gives us $H \propto R^{-\frac{\lambda^2}{2}}$ and $H = \frac{d\log R}{dt}$ gives us $R(t)$ and from the definition of $x_1$ we get $\dot{\phi} \propto H \implies \phi \propto \log R$.

  • $\begingroup$ Great answer, thank you! $\endgroup$
    – zabop
    Mar 30 at 16:40

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.