Extending the Nahm transformation for parabolic Higgs bundles on the projective line beyond the generic regime: critical weights, flat factors, and non-semisimple leading term at infinity

The Nahm transformation constructed by Szabó (2016) maps parabolic harmonic bundles on the complex projective line — with logarithmic poles at finitely many finite points and one second-order pole at infinity — to transformed parabolic Higgs bundles, inducing a holomorphic map of the corresponding moduli spaces, but it is established only under the genericity hypotheses of Assumption 2.3 of that paper and under semisimplicity of the leading-order term at infinity. This problem asks for the completion of the framework: define the transform when the nilpotent part of a graded residue acts nontrivially on its 0-eigenspace (so that 0 becomes a critical weight for the local model Dirac operator) or when 0 is an eigenvalue of the residue at infinity (so that flat factors may occur); treat the case of a non-semisimple second-order leading term at infinity; find an intrinsic, uniform algebraic description of the transformed R-parabolic sheaf replacing the manual addition/deletion operations on the parabolic divisor; and establish the behaviour of the induced moduli map beyond the Dolbeault complex structure.

Background

A Higgs bundle on a complex curve is a holomorphic vector bundle $E$ together with a holomorphic section $\theta$ of $\operatorname{End}(E)\otimes K$ (the Higgs field), $K$ being the canonical bundle, satisfying $\theta\wedge\theta=0$ . Given a Hermitian metric $h$ on the underlying smooth bundle, the triple $(V,\bar\partial_E,\theta,h)$ is a harmonic bundle — the central notion of non-abelian Hodge theory introduced by C. Simpson — when the connection $D=D^{\mathrm{Ch}}_h+\theta+\theta^{*}$ is flat, where $D^{\mathrm{Ch}}_h$ is the Chern connection of $(\bar\partial_E,h)$ and $\theta^{*}$ is the $h$ -adjoint of $\theta$ ; equivalently, the pair solves Hitchin's self-duality equations. Harmonic bundles connect the de Rham world of flat connections with the Dolbeault world of Higgs bundles. On a punctured compact curve one further records the behaviour near the punctures by a parabolic structure: at each puncture $z_i$ the fiber carries a decreasing filtration $F^\bullet_i$ with graded pieces $\mathrm{Gr}^j_i=F^j_i/F^{j+1}_i$ , and each graded piece $\mathrm{Gr}^j_i$ is assigned a real number $\alpha^j_i$ , with $1>\alpha^{l_i-1}_i>\dots>\alpha^0_i\geq 0$ (so that $\alpha^0_i$ is the smallest weight), called a parabolic weight. In the situation at hand the Higgs field has logarithmic (first-order) poles at finitely many points $z_1,\dots,z_n$ of the affine line and one second-order pole at $z_0=\infty$ ; near infinity one may write $\theta=\tfrac{A}{2}\,dz+B\,dz/z+\dots$ in the global affine coordinate $z$ (the differential $dz$ itself carries a second-order pole at $z=\infty$ when read on $\mathbb P^1$ , so $A$ is the leading-order term there), and this coefficient $A$ is assumed semisimple, that is, diagonalizable over $\mathbb C$ . The residue $\mathrm{res}_{z_i}(\theta)$ at a logarithmic pole need not be semisimple: its Jordan decomposition contains a nilpotent part, and it is exactly this non-semisimple case — corresponding, on the flat side, to nontrivial unipotent monodromy — that the source paper addresses.

The Nahm transform, in this circle of ideas, is an analogue of the Fourier transform and of the ADHM–Nahm construction of instantons and monopoles. For a harmonic bundle as above on $\mathbb P^1$ one twists the flat connection by the rank-one family $-\zeta\,dz$ depending on a spectral parameter $\zeta$ , and takes the $L^2$ -cohomology of the resulting twisted elliptic (Dirac) complex; under suitable hypotheses these $L^2$ cohomology spaces have locally constant dimension in $\zeta$ , so they form the fibers of a holomorphic bundle $\widehat E$ over a dual affine line, and the construction produces a transformed Higgs bundle $(\widehat E,\widehat\theta)$ with singularities of the same general type, carrying a transformed parabolic structure. Szabó established this transform for integrable connections with semisimple residues in a 2007 memoir, showed in 2012 that it matches the minimal Laplace transform of $\mathcal D$ -modules (modules over the sheaf of algebraic differential operators), developed with Aker in 2014 an algebraic version in terms of spectral data on Hirzebruch surfaces (rational ruled surfaces), and proved a Plancherel theorem for it in 2015. The source paper of the present problem (2016) extends the transform to residues that need not be semisimple and proves, as its main result (Theorem 6.2), that the transformed Higgs field $\widehat\theta$ is meromorphic with a second-order pole at infinity with semisimple leading term and first-order poles at the image punctures, that it is compatible with the transformed parabolic filtration, and that the transform acts holomorphically on moduli spaces for their Dolbeault complex structures — the complex structure under which the points of the moduli space are the Higgs bundles themselves; non-abelian Hodge theory endows the same moduli spaces with further complex structures, notably the de Rham one, under which points are flat connections.

The construction is carried out under an explicit genericity hypothesis (Assumption 2.3 of the source): (1) $0$ is not an eigenvalue of the induced residue $\mathrm{res}_{z_0}(\theta)^j$ at infinity; (2) at a finite puncture with parabolic weight $\alpha^j_i=0$ , the nilpotent part $N^j_i$ of the induced residue acts trivially on the $0$ -eigenspace of $\mathrm{res}_{z_i}(\theta)^j$ ; (3) at a finite puncture with $\alpha^j_i>0$ , $0$ is not an eigenvalue of the induced residue. The proofs show exactly where these hypotheses are used. First, the Fredholm property of the twisted Dirac operators (finite-dimensional kernel and cokernel, with closed range) relies on the weight $0$ not being a critical weight — a value at which the translation-invariant local model Dirac operator on a graded piece loses its spectral gap at $0$ ; Assumption 2.3(2) is precisely what excludes this. Second, condition (1) excludes flat factors, Higgs subbundles of the form $(\mathcal O_{\mathbb P^1}(m),\zeta\,dz)$ , the analogue of instantons with flat factors, which would contribute $\zeta$ -independent $L^2$ classes. Third, the local analysis near the irregular point at infinity is taken over from the semisimple case and uses the diagonalizability of $A$ . Finally, the transformed parabolic structure is assembled as an $\mathbf R$ -parabolic sheaf — a family of subsheaves of $\widehat E$ , indexed by a real parameter and decreasing as the parameter grows, encoding parabolic weights along the spectral curve, i.e. the curve in the product of the two projective lines cut out by the characteristic equation of the Higgs field — and this assembly is not canonical: for branches of the spectral curve that do not meet the divisor at infinity, twists must be added manually (Remark 5.7 of the source), an operation inverse to a deletion operation appearing elsewhere in the construction. The moduli-map properties are established only for the Dolbeault complex structure. These exclusions and gaps define the problem stated below.

Problem Statement

Complete the Nahm-transformation framework for parabolic harmonic bundles on the projective line — with logarithmic poles at finitely many finite points and one second-order pole at infinity — by removing the genericity hypotheses under which it is currently established and by clarifying its structural and moduli-theoretic content. Concretely, extending Szabó (2016):

(1) Define and study the transform when Assumption 2.3 of the source fails through its clause (2) (case (a)) or its clause (1) (case (b)); failure of its clause (3) (a $0$ -eigenvalue of the induced residue at a positive-weight finite puncture) is outside the scope of this item: (a) at some finite puncture $z_i$ , the nilpotent part $N^j_i$ of the induced residue acts nontrivially on the $0$ -eigenspace of $\mathrm{res}_{z_i}(\theta)^j$ , so that the parabolic weight $0$ becomes a critical weight for the local model Dirac operator; or (b) $0$ is an eigenvalue of $\mathrm{res}_{z_0}(\theta)^j$ at infinity, so that flat factors of the form $(\mathcal O_{\mathbb P^1}(m),\zeta\,dz)$ may occur. Construct the transformed object in these regimes, retaining the defining properties of the transform established in the generic regime (Theorem 6.2 of the source: $\widehat\theta$ is meromorphic, with a second-order pole at infinity with semisimple leading term and first-order poles at the image punctures; its residues are compatible with the transformed parabolic filtration; and the induced map on the moduli spaces is holomorphic for their Dolbeault complex structures).

(2) Define and study the transform when the leading-order term $A$ of the second-order pole at infinity is not semisimple, providing both the analytic construction ( $L^2$ theory of the twisted elliptic complex near an irregular, non-semisimple singularity) and the algebraic description of the transformed object.

(3) Decide whether the transformed $\mathbf R$ -parabolic sheaf admits an intrinsic, uniform algebraic description — one not requiring the manual addition or deletion of twists on the parabolic divisor for spectral-curve branches that fail to meet the divisor at infinity (Remark 5.7 of the source) — and, if such a description exists, determine how the choices it involves affect the moduli-theoretic properties of the transform, namely the holomorphicity of the induced map between moduli spaces for the Dolbeault complex structures (Theorem 6.2(3) of the source) and the compatibility of the transform with the transformed parabolic filtration.

(4) Establish the behaviour of the induced map between moduli spaces beyond holomorphicity for the Dolbeault complex structures (Theorem 6.2(3) of the source), with respect to the de Rham and remaining complex structures of the non-abelian Hodge moduli spaces and in metric or Hitchin-correspondence terms, in the presence of non-semisimple residues and, once (1)–(2) are resolved, in the degenerate regimes. A complete answer to (4) requires these directions jointly; establishing any single one of them does not by itself settle (4).

These deliverables share one scientific objective: a Nahm transform defined for the natural degenerate configurations of parabolic harmonic bundles on $\mathbb P^1$ — rather than only under Assumption 2.3 and semisimplicity of the leading term at infinity — with the analytic construction, the algebraic description of the transformed parabolic object, and the moduli-space mapping properties each in place to the same standard as in the generic regime. A resolution requires the four items (1)–(4) jointly, each independently assessed; resolving any single item is partial progress and does not by itself resolve the problem. Item (4) is not settled by resolving (1)–(2): its non-semisimple-residue part is independent of them, while its degenerate-regime part presupposes them.

References

Szilárd Szabó, Nahm transformation for parabolic Higgs bundles on the projective line — case of non-semisimple residues, Journal of Geometry and Physics 122 (2017) 80–91, DOI 10.1016/j.geomphys.2016.10.018, https://arxiv.org/abs/1606.03612
Szilárd Szabó, Nahm transformation for parabolic integrable connections on the projective line — case of generic regular graded residues, arXiv:1702.03306 (2017, preprint), https://arxiv.org/abs/1702.03306
Kürşat Aker and Szilárd Szabó, Algebraic Nahm transform for parabolic Higgs bundles on P^1, Geometry & Topology 18 (2014) 2487–2545, DOI 10.2140/gt.2014.18.2487, https://arxiv.org/abs/math/0610301
Szilárd Szabó, Nahm transform for integrable connections on the Riemann sphere, Mémoires de la Société Mathématique de France, Nouvelle Série 110 (2007) 1–114, DOI 10.24033/msmf.422, https://arxiv.org/abs/math/0511471
Szilárd Szabó, Nahm transform and parabolic minimal Laplace transform, Journal of Geometry and Physics 62 (2012) 2241–2258, DOI 10.1016/j.geomphys.2012.07.001, https://arxiv.org/abs/0704.2744
Szilárd Szabó, The Plancherel theorem for Fourier–Laplace–Nahm transform for connections on the projective line, Communications in Mathematical Physics 338 (2015) 753–769, DOI 10.1007/s00220-015-2348-2, https://arxiv.org/abs/1407.2440
Sergey Cherkis, Andrés Larraín-Hubach, Mark Stern, Instantons on multi-Taub-NUT spaces II: bow construction, Journal of Differential Geometry 127 (2024), no. 2, DOI 10.4310/jdg/1717772419, https://arxiv.org/abs/2103.12754
R. Donagi, T. Pantev, C. Simpson, Direct images in non-abelian Hodge theory, arXiv:1612.06388 (2016, preprint), https://arxiv.org/abs/1612.06388
Derek Harland, Parabolic Higgs bundles and cyclic monopole chains, Journal of Geometry and Physics 178 (2022) 104552, DOI 10.1016/j.geomphys.2022.104552, https://arxiv.org/abs/2012.01083
Szilárd Szabó, Exponential decay of harmonic 1-forms for wild harmonic bundles on curves, Comptes Rendus. Mathématique 362 (2024) 1015–1021, DOI 10.5802/crmath.607, https://doi.org/10.5802/crmath.607
Carlos Simpson, Harmonic bundles on noncompact curves, Journal of the American Mathematical Society 3 (1990) 713–770, DOI 10.1090/s0894-0347-1990-1040197-8, https://doi.org/10.1090/s0894-0347-1990-1040197-8
Takuro Mochizuki, Wild harmonic bundles and wild pure twistor D-modules, Astérisque 340 (2011), DOI 10.24033/ast.898, https://arxiv.org/abs/0803.1344
Takuro Mochizuki, Szilárd Szabó, Asymptotic behaviour of large-scale solutions of Hitchin's equations in higher rank, arXiv:2303.04913 (2023), https://arxiv.org/abs/2303.04913