Let $(x,t) \in \mathbb{R}^2$ and consider the following partial differential equation for the real-valued function $U(x,t)$: \begin{equation} \frac{\partial^2 U}{\partial t^2} = - \frac{\hbar^2}{4m^2} \frac{\partial^4 U}{\partial x^4}, \end{equation} where $m$ and $\hbar$ are positive constants. In the following we shall be quite sloppy, and we shall assume that given (smooth enough) initial conditions $U(x,0)$ and $\frac{\partial U}{\partial t}(x,0)$ (lying in some space) there exists a unique (smooth enough) solution $U$ (lying in some space) to this partial differential equation. Let us call the set of solutions $\mathcal{E}$.

Let us define for ease of notation $D_{x}^k F=\left( F, \frac{\partial F}{\partial x}, \dots, \frac{\partial^k F}{\partial x^k} \right)$ for every non-negative integer $k$ and for every smooth enough function $F(x,t)$. I ask whether there exist some non non-negative integer $k$ and some (smooth enough) real-valued functions $p \geq 0$ and $j$, with $p$ non-constant, such that, by setting \begin{equation} P(x,t)=p \left((D_{x}^k U)(x,t), \left(D_{x}^{k} \frac{\partial U}{\partial t}\right)(x,t) \right), \\ J(x,t)=j \left((D_{x}^k U)(x,t), \left(D_{x}^{k} \frac{\partial U}{\partial t}\right)(x,t) \right), \end{equation} the following properties hold:

(i) for every $U \in \mathcal{E}$ the following continuity equation holds \begin{equation} \frac{\partial P}{\partial t} + \frac{\partial J}{\partial x}=0; \end{equation}

(ii) for the special solution $U(x,t)=\cos\left(\sqrt{\frac{2m \omega}{\hbar}}x-\omega t \right)$, we have that $P(x,t)$ is independent of $\omega > 0$.

The answer should be negative, but I have no idea of a possible proof. Obviously, since we have not formulated the problem in a rigorous way, we do not expect to get a rigorous proof, but only some heuristic, convincing argument in this direction.

NOTE (1) Let us explicitly note a trivial consequence of property (ii). Since for the special solution \begin{equation} U(x,t)=\cos\left(\sqrt{\frac{2m \omega}{\hbar}}x-\omega t \right), \end{equation} $P(x,t)$ is independent of $\omega > 0$ for every $(x,t) \in \mathbb{R}^2$, we have that $(x,t) \mapsto P(x,t)$ is a constant function. Indeed, for this special solution, we have that $P(x,t)=F(x-vt)$ for some function $F$, where $v=\sqrt{\frac{\hbar \omega}{2m}}$. If we had $F'(\xi)\neq 0$ for some $\xi \in \mathbb{R}$, we would get for $x-vt=\xi$: \begin{equation} \left. \frac{\partial P}{\partial t}(x,t) \middle/ \frac{\partial P}{\partial x}(x,t) \right. = -v = \sqrt{\frac{\hbar \omega}{2m}}, \end{equation} a contradiction.

NOTE (2) This problem, as the notation shows, has a physical background, and the mathematical formulation of the problem that I give here is my personal interpretation of a physical exposition given by the great XXth century physicist David Bohm in his wonderful treatise Quantum Theory published in 1951. Bohm explicitly states that the problem has a negative answer, without giving any proof or heuristic argument. For all the physical details about this problem see my post Nonexistence of a Probability for Real Wave Functions.

  • 1
    $\begingroup$ If you compute the continuity equation at t=0 you get a relation which depends entirely on the initial conditions for U, which are completely arbitrary. I have the feeling that by playing with suitable choices of U(t=0) and U_t(t=0) you can prove P,J to be constant $\endgroup$ Commented Nov 6, 2018 at 18:13
  • $\begingroup$ Well, if you compute the continuity equation at $t=0$, by using the PDE in order to express $U_{tt}$, you actually get an equation which, given the arbitrariness of $U(t=0)$ and $U_t(t=0)$ and the Borel's Lemma, gives us a PDE for $p$ and $j$. But for sure this PDE has non-constant solutions. Actually, there exist non-constant functions $p \geq 0$ and $j$ satisfying the continuity equation: see the example given by Bohm in my post Nonexistence of a Probability for Real Wave Equations. $\endgroup$ Commented Nov 6, 2018 at 18:42
  • $\begingroup$ We conclude that in any case property (ii) must play some role in the proof of the non-existence result (if it is true). $\endgroup$ Commented Nov 6, 2018 at 18:42

1 Answer 1


This is not an answer, but only a remark too long to be posted as a comment.

We can make the formulation of the problem more explicit in the following way. Let $y=(y_0,y_1,\dots,y_k) \in \mathbb{R}^{k+1}$ and $z=(z_0,z_1,\dots,z_k) \in \mathbb{R}^{k+1}$. Then to say that the functions $(y,z) \mapsto p(y,z)$ and $(y,z) \mapsto j(y,z)$ satisfy property (i) means that for every $U \in \mathcal{E}$ \begin{equation} \nabla_{y} p \cdot D_{x}^{k} \frac{\partial U}{\partial t} + \nabla_{z} p \cdot D_{x}^{k} \frac{\partial^2 U}{\partial t^2} + \nabla_{y} j \cdot D_{x}^{k} \frac{\partial U}{\partial x} + \nabla_{z} j \cdot D_{x}^{k} \frac{\partial^2 U}{\partial x \partial t}=0, \end{equation} which is equivalent to say that \begin{equation} \nabla_{y} p \cdot D_{x}^{k} \frac{\partial U}{\partial t} - \frac{\hbar^2}{4 m^2} \nabla_{z} p \cdot D_{x}^{k} \frac{\partial^4 U}{\partial x^4} + \nabla_{y} j \cdot D_{x}^{k} \frac{\partial U}{\partial x} + \nabla_{z} j \cdot D_{x}^{k} \frac{\partial^2 U}{\partial x \partial t}=0. \end{equation} Now Borel Theorem tells us that for every real sequence $(a_{n})_{n=0}^{\infty}$ and every fixed $x_0 \in \mathbb{R}$ there exists a smooth function $f:\mathbb{R} \rightarrow \mathbb{R}$ such that \begin{equation} \frac{d^m f}{d x^m}(x_0)=a_{n} \quad (n=0,1,2,\dots), \end{equation} so that from the arbitrariness of the initial data $x \mapsto U(x,0)$ and $x \mapsto \frac{\partial U}{\partial t}(x,0)$, by replacing $\frac{\partial^m U}{\partial x^m}$ by $y_m \in \mathbb{R}$ and $\frac{\partial^{m+1} U}{\partial^m x \partial t}$ by $z_m \in \mathbb{R}$ for every non-negative integer $m$, we conclude that $p$ and $j$ satisfy the following equation:

\begin{equation} \sum_{m=0}^{k}\frac{\partial p}{\partial y_m} z_{m} - \frac{\hbar^2}{4 m^2}\sum_{m=0}^{k} \frac{\partial p}{\partial z_m} y_{m+4} + \sum_{m=0}^{k} \frac{\partial j}{\partial y_m} y_{m+1} + \sum_{m=0}^{k} \frac{\partial j}{\partial z_m} z_{m+1} =0 \qquad (I). \end{equation} From this equation we immediately deduce that $p$ does not depend on $z_{k-2}, z_{k-1}, z_k$ and that $j$ does not depend on $z_k$. Moreoever, we also get that \begin{equation} - \frac{\hbar^2}{4m^2} \frac{\partial p}{\partial z_{k-3}}+ \frac{\partial j}{\partial y_k} = 0, \end{equation} and since $p$ and $j$ do not depend on $z_k$ it also holds \begin{equation} \frac{\partial p}{\partial y_k} + \frac{\partial j}{\partial z_{k-1}}=0. \end{equation}

There are no other simple consequences of Equation (I). Now, the issue in the post amounts to see whether there exist solutions $p \geq 0$ and $j$ of Equation (I), with $p$ non-constant and satisfying property (ii). This latter request makes all the matter quite involved.

Let us explicitly note there are simple non-constant functions $p(y,z) \geq 0$ which satisfy (ii). Consider e.g. $p(y,z)=(y_1 z_1 - z_0 y_2)^2$. It is easy to see that in this case there is no $k$ and no function $j$ such that $p$ and $j$ satisfy equation (I). Indeed, equation (I) would imply in our case that $j$ depends only on $y_0,\dots, y_4, z_0,z_1$. Moreover it would also imply that

\begin{equation} \frac{\partial j}{\partial y_4}= \frac{\hbar^2}{2 m^2}(y_1 z_1 - z_0 y_2) y_1, \\ \frac{\partial j}{\partial z_1}= 2(y_1 z_1 - z_0 y_2) z_0. \end{equation}

From the last equation we would get for some function $g$: \begin{equation} j=2(y_1 z_1 - z_0 y_2) z_0 z_1 + g(y_0,\dots,y_4,z_0), \end{equation} so that we should have \begin{equation} \frac{\partial g}{\partial y_4} = \frac{\partial j}{\partial y_4} = \frac{\hbar^2}{2 m^2}(y_1 z_1 - z_0 y_2) y_1, \end{equation}

which is a contradiction since $g$ does not depend on $z_1$.

The difficulty of the problem is to show in general that conditions (i) and (ii) are not compatible, and I have no idea for now of how to tackle the problem.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.