Why does TsT transformation preserve the underlying physics despite changing the spacetime geometry?

This is an excellent, sophisticated question about a mathematical technique in theoretical physics—but the preprints provided don't address TsT transformations directly. Let me explain what the archive *does* cover, then why your question falls outside it.

What the archive illuminates about coordinate invariance

The preprints here tackle a related but distinct problem: how much of spacetime's structure is intrinsic (physical) versus conventional (coordinate choice)?

For example, [AG-2024.06-418] examines whether teleparallel gravity—which describes gravity via *torsion* in a flat spacetime rather than *curvature*—is truly equivalent to general relativity. The authors argue it is *not*, because teleparallel gravity posits extra mathematical structure (the torsion tensor) that doesn't change any observable predictions. In other words, you've changed the geometric language without changing the physics.

Similarly, [AG-2024.05-102] shows that Maxwell's equations stay invariant under coordinate transformations, yet behave covariantly under frame transformations—highlighting the subtle distinction between passive relabelings of coordinates and active changes in what observer you're describing.

Why TsT transformations aren't covered here

T-duality, S-duality, and TsT (T-duality–S-duality–T-duality) transformations are string-theory and supergravity tools that relate different string backgrounds to one another. They're symmetries of the *spectrum* and *scattering amplitudes* rather than classical coordinate symmetries. The archive focuses on foundations (relativity, reference frames, minimal spacetime structure) rather than dualities in string compactifications.

None of the preprints here directly address TsT transformations or why they preserve observables in string theory.