100 * (R - 16/7)^2 > G > I
(Finite area restriction)

I discovered that the primary was a quadratic function of rate fairly early on, but made the fatal assumption that all this math was just meant to disguise the metric, so I happily optimized a min rate solve. Then a demon in my brain told me to look at the graph of the function more closely, and staring back at me was the true minimum of the function at 16/7 R.

If I had more time this week I would redesign from scratch with 16/7 R in mind, but I don't, so you're getting my hastily modified 1R solution.