1: \begin{proof}

2: 

3: Assuming that $2|\sigma|<\lambda$, we consider the following

4: decomposition of the Hamiltonian:

5: $$

6: H=(\lambda-2|\sigma|)H_0

7: \,+\,

8: \big(2|\sigma|H_0+I\big).

9: $$

10: Using the result 1, we know that the operator $(2|\sigma|H_0+I)$ is positive. Thus we have the operator inequality $0\le (\lambda-2|\sigma|)H_0 \le H$.

11: 

12: Finally, we use the facts that $H$ is essentially self-adjoint for our choice of coupling and that $(\lambda-2|\sigma|)H_0$ has a discrete positive spectrum and we apply the lemma in \cite{lqc_spectrum1} (p.16) to conclude that $H$ has a discrete spectrum.

13: 

14: We do not repeat the rigorous proof of this lemma here. Nevertheless, the intuitive idea is that the subspace generated by eigenvectors of $H$ with eigenvalue inferior to an arbitrary value $\beta_m$ is automatically included in the subspace of of eigenvectors of $(\lambda-2|\sigma|)H_0$ with eigenvalue inferior to the same max value $\beta_m$. Since the latter has a finite dimension, the first also does.

15: 

16: \end{proof}

17: