[Paper Review] Hierarchies in relative Picard-Lefschetz theory

Marko Berghoff, Erik Panzer|arXiv (Cornell University)|Dec 12, 2022

Homotopy and Cohomology in Algebraic Topology4 citations

TL;DR

This paper develops a relative Picard-Lefschetz theory for families of complex manifolds with transverse divisors, introducing a hierarchy of iterated variations in relative homology that constrain the monodromy of parameter integrals. By resolving singularities via blowups to achieve simple type configurations, the authors derive new constraints that refine classical S-matrix hierarchy principles and explain analytic structures in Aomoto polylogarithms and Feynman integrals, including degenerate cases like the massless triangle and ice cream cone diagrams.

ABSTRACT

We prove a relative version of the Picard-Lefschetz theorem, describing the variation of relative homology groups $H_d(Y_t \setminus A_t,B_t\setminus A_t)$ in the fibers of a smooth fiber bundle $Y o T$ of complex manifolds with $A\cup B \subset Y$ transverse. From this we derive the vanishing of certain iterated variations, a system of constraints dubbed "hierarchy". As applications, we rederive the known analytic structure of Aomoto polylogarithms and massive one loop Feynman integrals. Moreover, we introduce the "simple type" to prove hierarchy constraints in degenerate cases where the Picard-Lefschetz formula does not apply, e.g. the massless triangle or the ice cream cone Feynman diagram. We compare our findings with a "classical" hierarchy of iterated variations (from 1960's $S$-matrix theory) and show how our setup not only explains, but also refines the latter. In order to do so, we need to further resolve the geometry of Feynman motives: We boldly blow up what no one has blown up before.

Motivation & Objective

• To establish a relative version of the Picard-Lefschetz theorem for families of complex manifolds with transverse divisors A and B.
• To derive a hierarchy of iterated monodromy variations in relative homology, explaining analytic constraints in parameter integrals.
• To resolve degenerate Feynman diagram singularities (e.g., massless triangle, ice cream cone) using a new blowup construction to access 'simple type' configurations.
• To refine and explain the classical hierarchy from 1960s S-matrix theory using modern geometric and homological tools.
• To demonstrate that the hierarchy constraints are preserved under blowups of Feynman motives, even when standard Picard-Lefschetz theory fails.

Proposed method

• Introduce a relative homology group $ H_d(Y_t \setminus A_t, B_t \setminus A_t) $ for families $ Y \to T $ with transverse divisors $ A \cup B \subset Y $, and define the Landau variety $ L \subset T $ as the singular locus of the monodromy representation.
• Define the variation operator $ \operatorname{Var}_\gamma = \gamma_* - \operatorname{id} $ on relative homology, and use Leray's residue and boundary maps to analyze local monodromy around critical values.
• Introduce the concept of 'simple pinch' components in the Landau variety, where the vanishing cycle is an iterated coboundary, enabling explicit computation of $ \operatorname{Var}_{\ell} $.
• Define a preorder on simple pinch components to organize the hierarchy of iterated variations, leading to constraints like $ \operatorname{Var}_{\ell} \circ \operatorname{Var}_{\ell_{A_{12}}} = 0 $ for $ \ell \notin \mathfrak{L}_{G/\gamma} $.
• Apply blowups to Feynman motives to resolve non-transverse intersections, transforming $ X' \to X $ so that $ A \cup B $ becomes normal crossing and enabling the hierarchy to be applied in degenerate cases.
• Use duality and relative intersection numbers to verify the consistency of the hierarchy with de Rham cohomology and monodromy representations.

Experimental results

Research questions

• RQ1 How can the Picard-Lefschetz theorem be generalized to relative homology groups $ H_d(Y_t \setminus A_t, B_t \setminus A_t) $ in families of complex manifolds?
• RQ2 What are the constraints on iterated monodromy variations in parameter integrals, and how do they form a hierarchy?
• RQ3 Why do certain Feynman integrals (e.g., massless triangle) fail to satisfy classical hierarchy constraints, and how can this be resolved geometrically?
• RQ4 How does the blowup of Feynman motives reveal new singularity structures that restore hierarchy constraints?
• RQ5 Can the classical S-matrix hierarchy be derived and refined from a modern geometric and homological framework?

Key findings

• The paper proves a relative Picard-Lefschetz theorem that describes the variation of relative homology in families with transverse divisors, establishing the foundation for hierarchy constraints.
• Iterated variations vanish under specific conditions: $ \operatorname{Var}_{\ell} \circ \operatorname{Var}_{\ell_{A_{12}}} = 0 $ for all $ \ell \in \mathfrak{L}_G \setminus \{\ell_\delta\} $, except at the special singularity $ \ell_\delta $, where $ \operatorname{Var}_{\ell_\delta} \circ \operatorname{Var}_{\ell_{A_{12}}} = -\operatorname{Var}_{\ell_{A_{12}}} $.
• The hierarchy constraints are refined by introducing a new blowup $ X \to X' $ of the Feynman motive, which resolves non-transverse intersections and enables the application of the simple type condition.
• The singularity $ \ell_\delta $, associated with the merging of two lines in $ A_1 \cap A_2 $, is shown to be of 'second type' but not simple, and its monodromy violates naive hierarchy predictions, highlighting the need for refined geometric resolution.
• The vanishing cycle for $ A_{12} $ is shown to be $ \nu = \delta_{A_2} \delta_{A_{12}} \eta $, with $ \eta = [\mathbb{S}^1] \in H_1(A_2 \cap A_{12} \setminus A_1) \cong \mathbb{Z} $, confirming the iterated coboundary structure.
• The framework successfully re-derives the analytic structure of Aomoto polylogarithms and massive one-loop Feynman integrals, and extends to degenerate cases like the ice cream cone diagram, where standard Picard-Lefschetz theory fails.