General comments

This manuscript proposes the application of the stochastic gradient descent (SGD) method from deep learning to the wind farm optimization problem, which maximizes annual energy production (AEP) with respect to wind turbine location. It is claimed that this work is the first such application, and it shows results demonstrating significant reduction in computation time for a small set of test cases. The manuscript is generally well-written, and the writing is concise.

Specific comments

Results are presented for three wind farms of different sizes, with 20 optimizations each with randomized initial designs. The authors are also asked to consider different wind farm shapes. Since there are some tuning parameters in the algorithm (and some tuning was required to achieve favorable results), a concern is whether significant problem-specific tuning is required. Demonstrating consistent results with different wind farm shapes (beyond just the one rectangular domain) would be very compelling.

There is clearly a cost to the reduction in computation time since constraints are not exactly satisfied. Could the authors comment on the degree of constraint violation and its significance, considering a range of different problems? Perhaps it would be helpful to visualize a Pareto front of computation time versus constraint violation, varying eta_T.

Technical corrections:

Line 28-29: Please rephrase - sentence is confusing

Line 44 and onwards: "SciPy SLSQP" should be changed to "SLSQP" - SciPy is a just a package that wraps an implementation of SLSQP

Line 77: for completeness, it should be stated at this point where the 8760 comes from, or at least mention that the units are hours

Line 107: I suggest separating the method from the implementation. For example, here, the choice to use the finite-difference method and algorithmic differentiation are aspects specific to the implementation, not aspects of the general method

Line 128: "intial" - typo

Figure 2: why do these curves have these shapes? Please comment in the manuscript on the trends and whether they agree with your intuition

Line 200: again, please comment on *why* the gains with SGD are higher with wind farms with more wind turbines

Line 214: "constrain" - typo

