[Paper Review] The canonical strip phenomenon for complete intersections in homogeneous spaces

Laurent Manivel|ArXiv.org|Apr 16, 2009

Mathematics and Applications6 references3 citations

TL;DR

This paper establishes that the canonical strip hypothesis—specifically the tight canonical strip (TCS) and canonical line (CL) hypotheses—holds for complete intersections in rational homogeneous spaces of Picard number one. Using representation theory and root system combinatorics, it proves that Hilbert polynomials of Fano and general type complete intersections have roots confined to the narrow strip or line Re(z) = -1/2, extending Golyshev's conjectures to this geometric setting.

ABSTRACT

We show that a refined version of Golyshev's canonical strip hypothesis does hold for the Hilbert polynomials of complete intersections in rational homogeneous spaces.

Motivation & Objective

• To verify refined versions of Golyshev's canonical strip conjectures for Hilbert polynomials of complete intersections in rational homogeneous spaces.
• To extend the canonical line (CL) and tight canonical strip (TCS) hypotheses beyond Fano threefolds and minimal threefolds to complete intersections in homogeneous spaces.
• To establish that the roots of the Hilbert polynomial $ H_{-K_X}(z) $ lie in the canonical strip or on the canonical line, depending on the canonical bundle type.
• To generalize results from hypersurfaces to complete intersections and branched coverings via induction and cohomological techniques.

Proposed method

• Apply the Weyl dimension formula to express the Hilbert polynomial $ H_L(z) $ of a rational homogeneous space $ X = G/P $ with Picard number one.
• Decompose $ H_L(z) $ into products of polynomials $ H_L^\ell(z) $ indexed by the pairing $ (\omega_0, \alpha) = \ell $, using root system data.
• Use the symmetry and unimodality of the coefficients $ h_{\ell,j} $ (number of roots with given $ (\omega_0, \alpha) $ and $ (\rho, \alpha) $) to analyze root distribution.
• Prove that the Hilbert polynomial $ H_{-K_Y}(z) $ for a Fano complete intersection $ Y $ has roots in the tight canonical strip $ -1 + \frac{1}{\iota_Y} < \mathrm{Re}(z) < -\frac{1}{\iota_Y} $ via induction on the number of hypersurfaces.
• Establish the canonical line hypothesis (CL) for general type complete intersections by showing $ H_{-K_Y}(z) $ is a polynomial in $ z^2 $ with positive coefficients, implying real non-positive roots for the associated $ P_X(z) $.
• Extend results to double coverings of such spaces by analyzing the Hilbert polynomial via $ H_{-K_Y}(z) = H_{-K_X}(z) + H_{-K_X}(z - \frac{d}{\iota_Y}) $, preserving root confinement.

Experimental results

Research questions

• RQ1 Do the roots of the Hilbert polynomial $ H_{-K_X}(z) $ for Fano complete intersections in rational homogeneous spaces of Picard number one lie within the tight canonical strip $ -1 + \frac{1}{\iota_X} < \mathrm{Re}(z) < -\frac{1}{\iota_X} $?
• RQ2 Does the canonical line hypothesis (CL) hold for general type complete intersections in such spaces, i.e., do all roots of $ H_{-K_X}(z) $ lie on $ \mathrm{Re}(z) = -\frac{1}{2} $?
• RQ3 Can the canonical strip behavior be extended to branched double coverings of rational homogeneous spaces?
• RQ4 How does the Hilbert polynomial of a complete intersection in an abelian variety satisfy the CL hypothesis?
• RQ5 What is the role of the Weyl dimension formula and root system combinatorics in controlling the location of the roots of $ H_{-K_X}(z) $?

Key findings

• The tight canonical strip hypothesis (TCS) holds for all Fano complete intersections in rational homogeneous spaces of Picard number one, with roots confined to $ -1 + \frac{1}{\iota_Y} < \mathrm{Re}(z) < -\frac{1}{\iota_Y} $.
• The canonical line hypothesis (CL) holds for all general type complete intersections in such spaces, as their Hilbert polynomial $ H_{-K_Y}(z) $ is a polynomial in $ z^2 $ with positive coefficients, implying roots on $ \mathrm{Re}(z) = -\frac{1}{2} $.
• For Calabi-Yau complete intersections in such spaces, the Hilbert polynomial $ H_{L_Y}(z) $ has purely imaginary roots.
• The result extends to double coverings of rational homogeneous spaces: if $ Y $ is a double cover branched over a hypersurface, then $ H_{-K_Y}(z) $ has roots either on $ \mathrm{Re}(z) = -\frac{1}{2} $ or in the tight canonical strip, so (TCS) holds.
• For complete intersections in abelian varieties, the canonical line hypothesis (CL) holds because the Hilbert polynomial factors into products of $ P_\ell(z) = z^\ell - (z-1)^\ell $, which when shifted by $ \frac{1}{2} $ yield polynomials with non-negative coefficients.
• The proof relies on the unimodal and symmetric nature of the coefficients $ h_{\ell,j} $ in the root decomposition of $ H_L(z) $, which ensures controlled root distribution via representation-theoretic and combinatorial arguments.