Source
Casals & Nho (2025), “Spectral Networks and Betti Lagrangians”, arXiv:2504.08144v1, LIT-017, read in full in NOTE-022. The reader is not a Floer-theory specialist, and NOTE-022’s Limitations mark what could and could not be checked.
What was actually shown
The setting. A Betti surface S is a punctured closed surface with a Legendrian positive-braid link at each puncture. A Betti Lagrangian is a Lagrangian n-fold branched cover of S inside T*S with those asymptotics, locally the D₄⁻ germ at each branch point (Definitions 2.3, 2.6–2.7). The account concerns class (i), the exact ones with conical ends.
Existence (Theorem 1(i)). Every exact Betti Lagrangian with an adapted metric admits a finite Morse spectral network, and the proof produces one with creation vertices only. Gappedness comes from an energy bound on chain-interaction trees (Propositions 3.17–3.18); finiteness comes from exactness. NOTE-022 rates this moderate to strong (C1).
Floer characterisation (Theorem 2). p lies on an active wall iff a broken pseudo-holomorphic strip between T*_pS and the Lagrangian exists in the adiabatic limit; for exact Lagrangians the strip is unbroken. The ”⇐” direction rests on Proposition 4.3, which gives convergence only to trimmed D₄⁻-trees (footnote 13). The ”⇒” direction in the exact case uses creativity of the network (§5.4.1).
Family Floer equals non-abelianization (Theorem 3 = Theorem 5.24). For exact L, the Family Floer functor is well defined, invariant under compactly supported Hamiltonian isotopy, and in the adiabatic limit equal to GMN’s non-abelianization Φ_W. Its proof assumes a finite creative network, which Theorem 1(i) supplies (C5, moderate). Theorem 4 realises Φ_W in the A∞-operations of wrapped and partially wrapped Fukaya categories of T*S.
What could have come out otherwise. A strip could have survived the limit off the network, or the continuation-strip count could have differed from GMN’s detour rule. The paper’s arguments say neither happens; the non-specialist reader could check the combinatorics and elementary flowline analysis, not the analytic convergence.
What this does not say
- That the theorems hold as broadly as they are stated. Theorem 3 says “a spectral network compatible with Σ” but is proved for finite creative networks. The meromorphic (WKB) case of Theorem 2 adds a hypothesis in prose, and Remark 5.10 says the functor is not constructed there.
- That the analysis has been checked. §§4–5 take their technical steps from Ekholm 2007, which has a gap the authors patch in Appendix A (Theorem A.3), and from the second author’s 2024 preprint, cited for about ten lemmas. The paper was unrefereed when read.
- That spectral networks come from Demazure weaves. Theorem 5 is “implied” by Proposition 7.5, which has a one-sentence justification. The Berk–Nevins–Roberts match is a comparison of figures (C9–C10).
- That non-abelianization is injective on objects. That GMN conjecture is argued in two sentences inside a section whose main diagram is itself conjectural (C11).
- Anything about physics beyond the dictionary. BPS states and WKB enter as names for D₄⁻-trees and the meromorphic case. No physical prediction is derived.
Connections
- The record has no technical neighbour. LIT-011 shares the word “spectral” and no mathematics.