*[These (and further) comments can also be viewed as annotations on the paper via Hypothesis.]*

This is a very interesting paper on the prospects of detecting QCD axion dark matter in terrestrial experiments. In the post-inflationary PQ symmetry breaking scenario, axions form highly overdense clumps (axion miniclusters, MCs) that might contain a fair fraction of the overall dark matter. Tidal interactions with stars in the Milky Way destroy some MCs and produce tidal streams. A previous study (Ref. [2]) showed that while the rate of MCs passing through the Earth is low, tidal streams of disrupted MCs may significantly increase the chances of detecting axions in direct detection experiments.

In this paper, the authors present a significantly improved calculation of the probability for tidal disruption of MCs than [2] including i) the effects of eccentric orbits in different halo potentials (NFW and isothermal) and ii) the contribution of halo and bulge star populations. The key results are:

* The destruction of MCs by the collective disk potential is negligible

* Approximately 2-5 % of all MCs are tidally destroyed and converted into axion streams

* Accounting for non-circular orbits reduces the probability of destruction by disk stars by a factor of 3 (cf. Ref. [2])

* The destruction probability by halo and bulge stars exceeds that of disk stars by a factor of 3-4

* The choice of halo model has an important effect, with the isothermal sphere increasing the destruction probability by a factor of 2-3 compared to an NFW halo.

The authors also estimate the chances of detecting axion streams with gravitational wave detectors.

The article is very concise and clearly written. My only general remark is that some discussion of the uncertainties of the stellar population models that were used would have been helpful. More generally, it is left unclear what the dominant sources of uncertainty are and how they could be reduced.

Some specific remarks:

* Eq. (20) is missing a factor of \(\frac{1}{M}\).

* I have one point of confusion regarding the approximation of the sum by an integral two lines above Eq. (22). In my (perhaps wrong) understanding, the first step should replace the sum by the total number of crossings \(N\) times the integral over the orbital angle from zero to \(\pi\), divided by the number of crossings per orbital angle \(\pi\). The latter is given by \(\pi/\tilde \phi\), hence the final expression should have a prefactor of \(\tilde \pi/\phi\) instead of \(1/\tilde \phi\). If correct, this would further reduce the probability for destruction by disk stars. Since this is already subdominant, the final result is practically unaffected.

* Using the definition of \(U_0\) from below Eq. (11), Eq. (23) and Eq. (30) must be multiplied by \(\sqrt{M}\). Since \(U_0\) is later defined again without the factor \(M\) (below Eq. (34)), this is a mere inconsistency that has no further consequences for the actual calculations if corrected. If I’m not mistaken, Eqs. (23) and (30) are also missing a factor of \(\sqrt{3}\)

*[These comments were shared with us by a member of the cosmology community. They do not necessarily reflect the opinion of the Cosmo Comments team.]*