Draft feedback 08/28

[ckoerber]

This issue addresses questions and adds comments regarding the ILP draft send out by @cchang5 08/28.

I tried to infer questions potential readers might have.
Obviously, this is subjective, so feel free to point out if some thoughts are not helpful.

Btw. I am not sure how much this helps, but I did some refactoring a few months ago summarized in PR #11

General remarks

• To some degree, I feel we present two papers (ILP & simulation), making it a little bit more challenging to have a coherent message.
• I think we should put more emphasis early on that we cross-check experiment with a simulation (which we label theory) and should mention at least a bit of details about the simulation. E.g., what is our model, which assumptions and approximations do we make, and how many parameters are present and estimated. This should, in my opinion, happen before we present the first scaling plots.
• I find the organization of sections confusing. E.g., in results, we present the mapping equations while the first plots are shown in discussion. I think it would be beneficial to first mention the blocks we intend to present (1. ILP, 2. Experiment scaling, 3. Simulation to understand scaling) (also different order possible). We should also conclude with observations made for individual blocks (e.g., ILP obtains the right result for individual graphs we analyzed, simulation (non-trivially) scaling agrees with D-Wave scaling, conclusion about which effects are essential to describe the experiment).
• I think we should not emphasize this, but, at least to a small degree, it would be nice to see or describe what happens if we change simulation details. E.g., what happens if we run simulations with just local damping or full counting. By how much do (e.g., temperature) parameter changes change this outcome.
• This might be a stupid question. If we want to address quantumness, would this correspond to changing the initial density (eq. 4.5) to H(0) -> H_final and run the simulation without the [H, rho] part (which is somehow not hbar ->0)?

Specific

• II Section B.1

minimum independent set
Should be Minimum Dominating Set, I guess.

• Embedding scaling (paragraph below 2.21)

the embedding at worse scale
Is that statement correct? E.g., don't we refer to the logical qubits, and depending on the connectivity, the scaling can be worse?

• Why do we investigate different offsets (offsets are first mentioned in paragraph 2 of section A.)?

• Why don't we show Fig. 3 not on a log scale?

• Linear offset extrapolations (Sec III A. 1.)

We extrapolate their values by a linear extrapolation
I guess that's for the simulation, only

• Qubit grouping due to large and small field:

max|{h}|−min|{h}|
Do you mean absolute value of h_i here (to me, the |.| suggests that)

• IIIB.

resample the DWave
What does resample mean in this case? Repeat computation (if, yes, why not say so?)?

• Fig 4.: Why +/- in the legend? I think it might be helpful to have the same offset color coding as in the other figures.

• Fig. 6 (maybe somewhere else). What was the prescription for estimating temperatures? Would it make sense to mark "fitted" points in this figure?

• How do bars change if we change parameters?

• Section IV A: Which decoherence model is needed for what?

• Description of eq. 4.3. Can we be more precise on how the j label relates to the nu mu labels?

• Above 4.5, why do we use this initial state?

• Fig 7 looks like just a single red line to me 😅

• Fig.8 Can we use this as a proof of quantumnessiness? E.g., it seems like -0.05 has a larger slope after the transition. Thus I would expect P(gs) to be larger if it was not delayed early on. How does this curve look like if we just used thermal annealing? Or could we still find different behavior by changing parameters?

[cchang5]

I quickly skimmed the comments and I agree with most of your comments and concerns.
Especially with the reordering of the sections though, we should talk about it more.
I wrote it this way because it was how it is ordered in my head, and I wanted to push out a first draft asap.
I also want to send this to either Sci Rep or Nature QIS, so they have guidelines on how the paper should be structured as well (at least Nat. QIS does). So I can explain to you a bit how this paper is put together with respect to that as well.

In any case, let's find a time to chat about this for sure.

[cchang5]

Done