[schoenan] Reviewing exercise submission

[aaronsch0]

The AI Code The AI tool (in my case, Gemini 3 Pro) created clean and functioning code. It successfully fulfilled my specific requests to write tests, accept a configuration file as input, and generate an output file. I was surprised that it implemented a Python Class with the methods load, solve, and save instead of simply using three functions. I suspect this happened because my prompt stated that "every class, method, etc. needs a proper docstring," but I never actually used the word "function."

Additionally, the methods did not validate the inputs. This likely occurred because the AI also generated the configuration file using valid values that worked for the simulation, and I did not explicitly request input validation in my prompt.

Comparison of the Reviews GitHub Copilot's review was very detailed, focusing mostly on minor issues and error handling. Crucially, Copilot also identified the main issue with the code: the lack of input value validation.

My review was not as detailed as Copilot's. I did not criticize the very minor points that Copilot labeled as "[nitpicking]." However, I did comment on the potential division by zero and suggested a fix.

Reviews of LAMMPS

lammps/lammps#4577
This review is very detailed and also used github copilot as reviewer.

lammps/lammps#4225
This review had a long and interesting discussion.

[aaronsch0]

I forgot the link, its here:
aaronsch0/documentation_exercise#2