How does subcell limiting decide when to activate without degrading accuracy in marginally-steep regions?

I appreciate the question, but I need to be direct: none of these preprints address subcell limiting in numerical methods for PDEs, which is what your question asks about.

The archive you've provided contains papers on quantum machine learning, neural network theory, and high-energy physics—not computational fluid dynamics or finite-volume/finite-difference schemes where subcell limiting is a standard technique.

Subcell limiting (used in methods like MUSCL reconstruction or slope limiters in shock-capturing schemes) is a computational tool for controlling oscillations near steep gradients while preserving accuracy in smooth regions. That topic doesn't appear in these results.

If you have a question about frontier physics—propulsion, relativity, quantum algorithms, or mathematical foundations of modern physics—I'm well-equipped to help. Otherwise, you may need to consult a computational methods archive or textbook on numerical PDEs.