Quantum Mechanics – How to Determine the Existence of a Ground State in a Finite Quantum Harmonic Oscillator

boundary conditionsground-stateharmonic-oscillatorpotentialquantum mechanics

I am having some problems with a finite, shifted quantum harmonic oscillator potential, and the theorem that states:

Any attractive potential in one dimension must have at least one bound state.

Let's consider the following potential:

$$
\ \begin{cases}
V_o(\frac{x}{a}-1)(\frac{x}{a}+1) & -a \leq x\leq a \\
0 \ \ \ \ \ otherwise
\end{cases}
\
$$

I tried doing it by scaling the Hamiltonian:

$$
\hat{H}=\sqrt{\frac{2V_o}{m}}\cdot\frac{1}{a}\left(\frac{\hat{P}}{2}+\frac{\hat{X}}{2}\right) – V_o
$$

So I suppose it should give shifted Eigenenergies of the Harmonic Oscillator :

$$
E_n =\sqrt{\frac{2V_o}{m}}\cdot\frac{1}{a} \left(\frac{1}{2}+n\right)
$$

And for the bounded states:
$$
E_n < 0 \iff \frac{ma^2V_o}{2} > \left(n+\frac{1}{2}\right)^2
$$

Which, for specified values of $V_o,m$ and $a$ doesn't allow a bounded ground state. But according to the theorem, one should exist. What is the problem here, or with my way of solving it?

Best Answer

  1. The Hamiltonian of your problem is given by $$\begin{align} H &= \frac{P^2}{2m}+ V_0\left(\frac{X}{a} +1\right)\left(\frac{X}{a}-1\right) \theta(X+a) \theta(a-X) \\ &=\frac{P^2}{2m}+\frac{V_0}{a^2}\left(X^2 -a^2\right)\theta(X+a)\theta(a-X),\end{align}$$ describing neither a harmonic oscillator nor being identical to the (obviously wrong) expression for $\hat{H}$ given in your second equation.

  2. If you wish to establish a relation between your system and a harmonic oscillator, you could proceed as follows: as the inequality $x^2-a^2 \ge 0$ holds for $|x|\ge a$, the Hamiltonian $H$ and the operator $$H_{\rm HO} =\frac{P^2}{2m} + \frac{V_0}{a^2}(X^2-a^2)= \frac{P^2}{2m}+\frac{V_0}{a^2} X^2-V_0$$ obey the operator inequality $H \le H_{\rm HO}$. $H_{\rm HO}$ describes a harmonic oscillator with angular frequency $\omega =\frac{1}{a}\sqrt{2 V_0/m}$ and the pure point spectrum $\hbar \omega (n+1/2)-V_0$ (with $n=0, 1, 2, \ldots$). In addition, $H$ fulfills the (trivial) inequality $-V_0 \le H$ and we conclude that the two operator inequalities $$-V_0 \le H \le H_{\rm HO}$$ imply$^\ast$ that the ground-state energy $E_0$ of the Hamiltonian $H$ is bounded from below and above by $$-V_0 \le E_0 \le \frac{\hbar \omega}{2}-V_0, \qquad \omega=\frac{1}{a}\sqrt{\frac{2 V_0}{m}}.$$ In the case of large values of $\omega$ with $\hbar \omega /2 \ge V_0$, the upper bound can be improved to $E_0 \le 0$.

  3. The spectrum of $H$ consists of a point spectrum with eigenvalues $-V_0 \le E_r \le 0$ and a continuous spectrum $[0, \infty)$. There is no reason why the well-known theorem about the existence of at least one bound state in the case of an attractive potential in one spatial dimension should be violated in this specific case.

$^\ast$ $H \le H_{\rm HO}$ means that $\langle \psi | H \psi \rangle \le \langle \psi | H_{\rm HO} \psi \rangle$ for all $\psi \in D(H) \cap D(H_{\rm HO})= D(H_{\rm HO})$. $\inf\limits_{||\psi ||=1} \langle \psi | H \psi \rangle \le \inf\limits_{||\psi||=1} \langle \psi |H_{\rm HO} \psi \rangle$ implies the inequality $E_0 \le \hbar \omega/2 -V_0$ for the ground-state energies of $H$ and $H_{\rm HO}$. Analogously, $ -V_0 \langle \psi |\psi \rangle \le \langle \psi | H \psi \rangle$ for all $\psi \in D(H)$, implies $-V_0 \le E_0$.