How does the x-y duality constraint differ from symplectic duality when predicting BPS invariants?

This is a sophisticated question about the relationship between two types of geometric dualities in topological recursion and their role in computing invariants of quantum systems.

The Core Distinction

The key difference lies in what gets swapped and what remains invariant [AG-2024.05-860]:

x-y duality is a symmetry operation that literally exchanges the roles of two coordinates in a spectral curve—the mathematical object encoding quantum energy levels. When you swap x and y, you're performing a kind of geometric reflection.

Symplectic duality, by contrast, is a broader transformation that respects the underlying symplectic geometry (the structure that encodes how phase space is organized). Remarkably, symplectic duality can be *decomposed* as a composition of multiple x-y dualities in a more general framework [AG-2024.05-860]. In other words, x-y duality is more elementary; symplectic duality is built from it.

Why This Matters for BPS Invariants

BPS invariants are special quantum numbers counting certain bound states of particles—they're "protected" by supersymmetry, meaning they don't change under small perturbations. The reason duality matters is that it gives you multiple ways to compute the same invariant [AG-2025.04-1162]. If two dual descriptions compute different answers, something is wrong; agreement confirms the calculation.

However—and this is crucial—the free energy (a key quantity in these calculations) is not generally invariant under x-y swapping alone [AG-2025.02-1029]. The difference $F_g - F_g^\vee$ between the original and x-y-swapped free energy is nonzero. But this difference can now be calculated explicitly as a residue at singularities, which actually *helps* when one dual side vanishes [AG-2025.02-1029]. This shows that x-y duality and symplectic invariance are different notions—understanding their interplay refines how we predict invariants.

The preprints don't directly compare how these two dualities *differently predict* BPS invariants in a concrete example; you may want a paper that walks through a specific gauge theory or spectral curve.