Polariton condensation into vortex states in the synthetic magnetic field of a strained honeycomb lattice

Dear Editor,

Please find submitted a new version of our manuscript titled “Polariton condensation into vortex states in the synthetic magnetic field of a strained honeycomb lattice”, for publication in SciPost Physics. We have effected anew a series of amendments in the main text.

Reply to the Referee:

We thank the reviewer for his comments and his positive assessment. We have addressed his comments in the revised version of the manuscript.

1. The choice in the direction of the strain is what breaks the symmetry between A and B sites and thus determines the A or B sublattice polarization of the n=0 Landau level (see [29-32] and Appendix A). For positive (negative) strain parameter, $\tau>0$ ($\tau<0$), the $n=0$ wavefunction is localized in the B (A) sublattice. This is mentioned in the paragraph preceding Eq.(3), ``... completely localized in the B sublattice (it would be in the $A$ sublattice if $\tau$ was negative)''. In the revised text, we have added reference to a few works where the interested reader can find a discussion of this fact. A summary of the analytical derivation is available in Appendix A.

2. This is an interesting point. At the end of Sec.2.1, we have added a sentence explaining why we expect that the mode selection mechanism based on spatial overlap with the pump should largely exceed all other mode selection mechanisms.

3. Indeed, as the referee points out, what we have in mind is the homogeneity of the condensate density in real space. Among the available states in which condensation can occur (the degenerate $n=0$ Landau level states), it is favourable to choose a linear combination in a single valley to reduce the interaction energy. When only positive or negative angular momenta states are available, like in rotating cold atoms, the energy is minimized forming a triangular array of vortices or antivortices, which the result of Abrikosov. But what happens if both positive and negative angular momenta are available like in our case? A linear combination of states with both chiralities would entail strong inhomogeneities in real space due to the destructive interference. To give an analytical example, let us imagine for a moment that we are dealing with a continuous space like in rotating BECs. Forming an equally-weighted linear combination of $m=1$ and $m=-1$ single-particle states, would increase, compared to a state with just $m=1$ or $-1$, the interaction energy ($\propto \int |\psi(r)|^4 d^2 r$) by a factor of $\left(\int\limits_0^{2\pi}4\cos^4(\phi)d\phi\right)/\left(\int\limits_0^{2\pi}d\phi\right) = 3/2$. Thus a single orientation is favoured.

We have modified our previous sentence, highlighted by the referee, by ``On the contrary, interactions play instead a key role in the case of wider incoherent pump profiles, where they make condensation in a single Dirac valley---as opposed to in both valleys simultaneously---more likely. In standard BECs at thermal equilibrium, the interaction energy is minimized by choosing among the single-particle ground states the one which has the most homogenous density in real space. Here, a similar mechanism hinders the condensate from occupying both valleys at the same time, which would entail large density inhomogeneities due to destructive interference of states with opposite angular momenta.''

Regarding the the two suggested papers, they are clearly relevant and we now cite them. We have now clarified in the manuscript, though, that we are not dealing with transient phenomena like Kibble-Zurek physics, which is beyond the scope of our work.

1. We have refined the tuning of the sentences in the outlook modifying the last paragraph: ``When applied to strongly-interacting photonic lattices...''.

In view of the above modifications and qualifications, we would be grateful if you could consider our manuscript for publication in SciPost Physics.

Yours sincerely,

C. Lledó, I. Carusotto, M. H. Szymaǹska

1) Added references [44, 45, 47, 50].

2) In the last paragraph of the Introduction, we changed the sentence "...as used in the recent works [27, 40–43]..." for "...as used in the recent experiments [27,40–43] and theoretical works [31,44,45]..."

3) In the paragraph following Eq. (3), we changed "...We will take advantage of this to induce polariton condensation into the n = 0 Landau level." for an extended version "...We will take advantage of this to induce polariton condensation into the n = 0 Landau level [47]: given the full sublattice polarization of this state, we expect that the mode selection mechanism based on spatial overlap with the pumped sublattice will largely exceed other mechanisms that may be active in polariton systems, such as energy relaxation, the exciton-photon fraction of modes, and the amount of localization of the wave function at the pillars."

4) We have corrected a typo at the beginning of Sec. 2.3 : "$P_B >~ \gamma$" has been replaced for $P_B \gtrsim \gamma$.

5) At the end of Sec. 2.2 we have included the new paragraph "The results that we are going to show in the following of the work refer to the long-time limit of the system evolution. A study of the additional features that may occur during the switch-on transient, such as Kibble-Zurek vortex nucleation phenomena [45] go beyond the scope of this work."

6) We have have erased the words "...the y-reflection symmetry is explicitly broken..." from the first paragraph of the Discussion section, as it seemed to obscure the proposal.

7) We have rewritten the last paragraph of the Outlook.

8) We have added a "Data and Code Availability" section before the Acknowledgements.