Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Yang et. al hope to improve the convergence speed with high correction accuracy for wavefront sensorless adaptive optics. The convergence speed is a key problem of wavefront sensorless AOS. The work is interesting for the applications of the AOS. The concrete comments:

1， As say with the author, there are different types of correction methods for the wavefront sensorless AO. However, in the paper, only different methods of meta-heuristic methods are investigated and compared. The global method, such as the Particle Swarm Optimization, although the search accuracy is high, but the speed is very slowly. Therefore, the other method, such as the SPGD algorithm, should be selected to perform the comparison.

2， Abstract: “which enables quick and efficient correction of small aberrations as well as larger aberrations.”, but in the context, only the turbulence with the strength D/r₀=10 is used to do the simulation. Hence, the correction of turbulence with different strength should be simulated.

3， In the paper, only the point object is used to perform the simulation and experiment. However, in lots of applications, the object is the extended object. So, the author should point out this situation.

4， In section 3.1, “…from order 3 to 104…”，I think it should be the modes of the Zernike polynomial, but not the order. There is the difference between the mode and order.

5， The style of the paper should be revised carefully, for example: Fig.3(a) and Fig.4, the unit should be added; the space should be added between the number and the unit, etc.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

some correction to be applied

1 WaveFront Sensor less (WFS-less), change WFSless in WFS-less in the article

87 explain better as the phase screen are built and add a power spectrum of the mean aberrations of the turbolence

Fig. 1 seem to describe aberrated wavefront coming outisde the DM, add where are created, the lisght source appear as a collimated beam

110 add, see fig. 3, 

Instead of the Strehl ratio, it would be more appropriate to use the encircled energy as a quality parameter

a plot of the rms on the image appling the correction will be interesting

117 in fig. 3  r_0 in subscript, add unit measurement for MR in the axis

128 explain the parameters of the computing, and which language code are you using to do this computation, if simple or parralel computing

143 wil be interesting to see the interferogram of the DM in RUN, PSO, DEA, GA. Explain as the decomposition of Zernike coefficient is computed. Is not clear if is from an iterferogram, or by analyzing the psf, or simple is the result from the algorith computation.

150 is unclear the function of the aperture stop, beacuse of the aperture of the DM is just a stop for the system. I fyou use the AS to different size below the DM aperture please write down in the text.

155 will be interesting, if provided, which is the speed of the cloose loop

180 in time comsuming please ad a row with the time in seconds

190 is better to compare results not only with the peak of the light but also with the encircled energ, because of in case of monocromatic light interference pattern can be present

I found a similar paper, with different algorith used, "Build the Structure of WFSless AO System Through Deep Reinforcement Learning" https://ieeexplore.ieee.org/document/8487017, please consider to citing 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The revision is Ok.