Identifiability and singular locus of secant varieties to spinor varieties

Basic Information

• Paper ID: 2302.05295
• Title: Identifiability and singular locus of secant varieties to spinor varieties
• Author: Vincenzo Galgano (University of Trento)
• Classification: math.AG (Algebraic Geometry)
• Publication Date: February 2023 (arXiv v2: December 10, 2023)
• Paper Link: https://arxiv.org/abs/2302.05295

Abstract

This paper analyzes the $Spin(V)$-structure of the secant variety of lines $\sigma_2(\mathbb{S})$ to the spinor variety $\mathbb{S}$ minimally embedded in its spin representation. The study determines the poset of $Spin(V)$-orbits and their dimensions, which are used to resolve identifiability and tangential identifiability problems in $\sigma_2(\mathbb{S})$, as well as to determine the second Terracini locus of $\mathbb{S}$. The paper ultimately proves that the singular locus $Sing(\sigma_2(\mathbb{S}))$ contains two $Spin(V)$-orbits of minimal dimension and is contained in the tangential variety $\tau(\mathbb{S})$.

Research Background and Motivation

1. Core Problem: Investigation of geometric properties of secant varieties, particularly their identifiability and singularity, which is a classical problem in algebraic geometry.
2. Significance:
   • The identifiability problem has important implications in both theory and applications, relating to the uniqueness of tensor decompositions
   • Study of the singular locus is crucial for understanding the geometric structure of secant varieties
   • As an important example of rational homogeneous varieties, the properties of secant varieties to spinor varieties have representative significance
3. Existing Limitations:
   • For general secant varieties, determining the singular locus $Sing(\sigma_r(X))$ is a central problem, but only a few cases have been completely resolved
   • The secant variety equations and singular locus of spinor varieties have not been fully determined
   • Understanding is limited except for the Legendrian variety corresponding to $Spin_{12}$
4. Research Motivation: To systematically study the orbit structure of the second secant variety by utilizing the Clifford algebra structure of spinor varieties and the natural properties of the $Spin_{2N}$ group action.

Core Contributions

1. Orbit Classification: Complete determination of the poset structure and dimensions of $Spin_{2N}$-orbits in $\sigma_2(S_N^+)$
2. Identifiability Theory: Complete resolution of identifiability and tangential identifiability problems via Clifford duality
3. Terracini Locus: Determination of the second Terracini locus of spinor varieties
4. Singular Locus Bounds: Provision of upper and lower bounds for the singular locus of $\sigma_2(S_N^+)$, with a conjecture for complete characterization
5. Theoretical Tools: Introduction of Clifford duality as a special case of non-abelian duality

Methodology in Detail

Problem Definition

Study of the second secant variety $\sigma_2(S_N^+)$ to the spinor variety $S_N^+ \subset \mathbb{P}(\bigwedge^{ev} E)$, where:

• Input: Spinor variety on an even-dimensional vector space $V = E \oplus E^{\vee}$
• Output: Orbit classification, identifiability determination, singular locus characterization
• Constraints: Focus on even $N$, utilizing the spin group of D-type Dynkin diagrams

Core Architecture

1. Clifford Algebra Framework

Establishment of correspondence between spinor varieties and maximal isotropic subspaces using Clifford algebra $Cl_q(V)$: $S_N^+ \cup S_N^- \leftrightarrow OGr^+(N,V) \cup OGr^-(N,V)$ Implementation via the map $\psi_a: V \to \bigwedge E$, $v \mapsto v \cdot a$.

2. Hamming Distance Classification

Definition of Hamming distance on spinor varieties: $d([a],[b]) = \frac{\text{codim}_{H_a}(H_a \cap H_b)}{2}$ Application of this distance to decompose $S_N^+ \times S_N^+$ into orbits.

3. Orbit Construction

• Secant Orbits: $\Sigma_{l,N} = \{[a+b] \in \sigma_2^{\circ}(S_N^+) | d([a],[b]) = l\}$
• Tangential Orbits: $\Theta_{l,N} = \{[q] \in \tau(S_N^+) | q \text{ corresponds to antisymmetric matrices of rank } 2l\}$

Technical Innovations

1. Clifford Duality

Introduction of a new duality theory: $\Phi: \bigwedge^{ev} E \otimes \bigwedge^{od} E^{\vee} \to E \oplus E^{\vee}$$e \otimes f \mapsto C_e(f)|_E + C_f(e)|_{E^{\vee}}$

This is a concrete realization of non-abelian duality in the context of spinor varieties, generalizing the classical antisymmetric catalecticant.

2. Inductive Arguments

Establishment of relationships between secant orbits of spinor varieties of different dimensions via the injectivity of the wedge product map: $(e_{[N-2l]} \wedge \bullet): \Sigma_{l,2l} \to \Sigma_{l,N}$ Implementation of inductive proofs.

3. Fibration Structure

Construction of fibration $\xi: \Sigma_{l,N} \to OGr(N-2l,V)$ whose fibers are isomorphic to dense orbits of smaller spinor varieties, used for dimension calculations.

Experimental Setup

Theoretical Verification

This is a purely theoretical study with results verified through rigorous mathematical proofs:

1. Special Case Verification: Detailed analysis of cases $N=4,5,6,8$ to verify the general theory
2. Dimension Calculations: Computation of orbit dimensions via the fiber dimension theorem
3. Comparison with Known Results: Comparison with known results for the Legendrian variety ($N=6$)

Computational Examples

• $S_4^+$ Case: Corresponds to 6-dimensional quadric $Q_6 \subset \mathbb{P}^7$, decomposition locus parameterized by $\mathbb{P}^6 \setminus Q_5$
• $S_6^+$ Case: 15-dimensional Legendrian variety, orbit dimensions are 15, 24, 30, 31
• $S_8^+$ Case: Unique case satisfying $\Theta_{3,8} = \Theta_{4,8} \cap \Sigma_{3,8}$

Experimental Results

Main Theorems

Orbit Poset Structure (Theorem 3.12)

For $N \geq 6$, the $Spin_{2N}$-orbit poset in $\sigma_2(S_N^+)$ is:

S_N^+
  ↓
Θ_{2,N} = Σ_{2,N}
  ↓        ↓
Θ_{3,N} → Σ_{3,N}
  ↓        ↓
  ⋮        ⋮
  ↓        ↓
Θ_{N/2,N} → Σ_{N/2,N}

Identifiability Results (Main Theorem)

• Points in $\Sigma_{2,N}$ are not identifiable
• Points in $\Sigma_{l,N}$ ($l \geq 3$) are identifiable
• Points in $\Theta_{l,N}$ ($l \geq 3$) are tangentially identifiable

Dimension Formulas (Propositions 6.2, 6.4)

$\dim \Sigma_{l,N} = \begin{cases} \frac{N(N-1)}{2} + 4N - 15 & \text{if } l = 2 \\ \frac{N(N-1)}{2} + l(2N-1) - 2l^2 + 1 & \text{if } l \geq 3 \end{cases}$

$\dim \Theta_{l,N} = \frac{N(N-1)}{2} + l(2N-1) - 2l^2 \quad (l \geq 3)$

Key Results

Terracini Locus (Theorem 7.1)

The second Terracini locus is precisely the closure of the distance-2 orbit: $\text{Terr}_2(S_N^+) = \overline{\Sigma_{2,N}} = S_N^+ \sqcup \Sigma_{2,N}$

Singular Locus Bounds (Corollary 8.3)

For $N \geq 7$: $\overline{\Sigma_{2,N}} \subset \text{Sing}(\sigma_2(S_N^+)) \subset \tau(S_N^+)$

Singular Locus Conjecture (Conjecture 8.4)

$\text{Sing}(\sigma_2(S_N^+)) = \overline{\Sigma_{2,N}} = S_N^+ \sqcup \Sigma_{2,N}$

Related Work

Historical Background

1. Classical Results: Zak's pioneering work established fundamental theory of secant varieties to rational homogeneous varieties
2. Known Cases:
   • Singular locus of second secant varieties to Segre varieties MOZ15
   • Second and third secant varieties to Veronese varieties Kan99, Han18
   • Partial results for Grassmann varieties MM15

Concurrent Work

Galgano-Staffolani GS23 independently obtained the same orbit poset and identifiability results for Grassmann varieties, suggesting general behavior for varieties of minimal codimension.

Theoretical Contributions

The Clifford duality in this paper generalizes the antisymmetric duality of Arrondo et al. Arr+21, providing specialized tools for spinor varieties.

Conclusions and Discussion

Main Conclusions

1. Complete resolution of the orbit classification problem for second secant varieties to spinor varieties
2. Complete characterization of identifiability via Clifford duality
3. Determination of the Terracini locus and precise bounds for the singular locus

Limitations

1. Dimension Restrictions: Main results require $N \geq 6$; lower-dimensional cases need special treatment
2. Technical Assumptions: Restriction to even $N$; odd cases require additional work
3. Unproven Conjecture: Precise characterization of the singular locus remains conjectural, despite strong theoretical support

Future Directions

1. Generalization to Varieties of Minimal Codimension: Extension of results to more general minimal codimension homogeneous varieties
2. Higher-Order Secant Varieties: Study of analogous problems for $\sigma_r(S_N^+)$ ($r \geq 3$)
3. Computational Aspects: Development of effective algorithms for determining identifiability of specific points

In-Depth Evaluation

Strengths

1. Theoretical Depth: Skillful combination of Clifford algebras, Lie group representation theory, and algebraic geometry
2. Methodological Innovation: Clifford duality provides new tools for addressing problems on spinor varieties
3. Complete Results: Provides complete picture of orbit classification, dimensions, and identifiability
4. Technical Rigor: Detailed proofs with clear logic and accurate calculations

Weaknesses

1. Limited Scope: Results are specifically tailored to spinor varieties with limited generalizability
2. Computational Complexity: High computational complexity in practical applications
3. Conjecture Dependence: Key characterization of singular locus remains dependent on unproven conjecture

Impact

1. Theoretical Contribution: Provides new techniques and perspectives for secant variety research
2. Application Potential: Potential applications in quantum information (fermionic Fock spaces)
3. Foundation for Future Research: Establishes foundation for systematic study of varieties of minimal codimension

Applicable Scenarios

1. Pure Mathematics Research: Algebraic geometry, representation theory, invariant theory
2. Applied Mathematics: Tensor decomposition, multilinear algebra
3. Theoretical Physics: Spinor field theory, quantum information processing

References

This paper cites 50+ related references, primarily including:

• Zak Zak93: Fundamental theory of secant varieties
• Landsberg-Manivel LM01, LM07: Research on Legendrian varieties
• Arrondo et al. Arr+21: Theory of antisymmetric catalecticants
• Procesi Pro07: Clifford algebras and spin groups theory

This paper achieves significant breakthroughs in the study of secant varieties to spinor varieties. Beyond resolving specific geometric problems, it importantly develops new theoretical tools, providing a solid foundation for further research in related fields.