I'd like to find all solutions of $u_t$ = $u_{xx}$ of the form

$$u = \left(\frac{1}{\sqrt{t}}\right)v\left(\frac{x}{2\sqrt{t}}\right).$$

I know that this problem reduces to solving a second order ODE for $v(y)$ with one solution $v(y) = e^{-y^2}$. Thus I must seek other linearly independent solutions of the form $v(y) = h(y)e^{-y^2}$.

All help and suggestions are appreciated

Update: Here is the derivation of the ODE, which was asked for in the comments. Suppose: \begin{align} y(x,t) &\equiv \frac{1}{2}xt^{-1/2}\\\ \\ u(x,y) &\equiv t^{-1/2}v(y)\\\ \\ \left. \frac{\partial u}{\partial t} \right|_x &= \left. \frac{\partial^2 u}{\partial x^2} \right|_t\tag{$\ast$}. \end{align} Then: \begin{align} \left. \frac{\partial u}{\partial t} \right|_x &= -\frac{1}{2}t^{-3/2}v(y) - \frac{1}{4}t^{-2}x\frac{dv}{dy}\\ \left. \frac{\partial^2 u}{\partial x^2} \right|_t &= \frac{1}{4}t^{-3/2} \frac{d^2v}{dy^2}. \end{align} So $(\ast)$ becomes: \begin{equation} \frac{1}{8}t^{-3/2} \frac{d^2v}{dy^2} + \frac{1}{8}t^{-2}x\frac{dv}{dy} + \frac{1}{4}t^{-3/2}v=0. \end{equation} Assuming $x\neq 0$, this implies: \begin{equation} \frac{d^2v}{dy^2} + 2y\frac{dv}{dy} + 2v=0,\tag{$\star$} \end{equation} which is the ODE referred to above.

  • 2
    $\begingroup$ What is that second order ODE? $\endgroup$ – Stefan Smith Apr 6 '13 at 1:21
  • $\begingroup$ @Stefan I added the ODE $(\star)$ to the question; my edit is sitting in the queue. rhm52: I hope you don't mind. $\endgroup$ – Douglas B. Staple Apr 6 '13 at 3:17
  • $\begingroup$ Thank you Douglas I should have included that with my original post. $\endgroup$ – rmh52 Apr 6 '13 at 3:23
  • $\begingroup$ Since it's a linear second order ODE, if you have one solution, you should try using variation of parameters to get the other. $\endgroup$ – Ray Yang Apr 6 '13 at 6:23
  • 1
    $\begingroup$ This is homework and I've actually solved it. I'll post the solution when I get the chance. $\endgroup$ – rmh52 Apr 6 '13 at 21:56

If $v(y)=h(y)\mathrm e^{-y^2}$, then $v$ solves the ODE $v''+2yv'+2v=0$ if and only if $h$ solves the ODE $h''-2yh'=0$. Thus, $h'(y)=a\cdot\mathrm e^{y^2}$ for some $a$ and $v=v_{a,b}$ for some $(a,b)$, where $$ v_{a,b}(y)=\mathrm e^{-y^2}\left(a\int_0^y\mathrm e^{z^2}\mathrm dz+b\right). $$ Note that the functions $v_{1,0}$ and $v_{0,1}$ are linearly independent. Hence the set of solutions of the ODE $v''+2yv'+2v=0$ is indeed the two dimensional vector space $\{v_{a,b}\mid(a,b)\in\mathbb R^2\}$. The function $v$ in the question is $v_{0,1}$.

Proof: If $w(y)=\mathrm e^{-y^2}$ then $w'=-2yw$ and $w''=(4y^2-2)w$. Hence, if $v=hw$ then $v'=(h'-2yh)w$, $v''=(h''-4yh'-2h+4y^2h)w$ and $v''+2yv'+2v=(h''-2yh')w$. End of the proof.


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.