[論文レビュー] Deep learning topological inference-guided $T_{cc}^{+}$ pole parameter extraction

Julius B. Pagayon, Klarence Tomas R. Cervantes+1 more|arXiv (Cornell University)|Mar 18, 2026

Quantum many-body systems被引用数 0

ひとこと要約

paperは合成で解析的に制御された振幅を用いて深層ニューラルネットを訓練し、四つのシートを持つS行列のポール拓扑を分類する。次に uniformized S-matrix と K-matrix フォーマリズムを用いて Tcc+ ポールパラメータを抽出し、第二のシート上に浅い D0D*+ 結合状態として Tcc+ を結論づける。

ABSTRACT

We perform a data-driven study of the doubly charmed tetraquark candidate $T_{cc}^+$. An ensemble of deep neural network classifiers, trained on synthetic amplitudes with controlled analytic structures, identifies a dominant pole topology characterized by an isolated pole on the $[bt]$ Riemann sheet which is robust against left-hand cut effects. A subsequent pole parameter extraction was performed via the uniformized $\mathcal{S}$-matrix and a complementary $\mathcal{K}$-matrix parameterization, which respectively provides a model-independent baseline and dynamical insight on the pole position and trajectory of the resonant state. Using this two-pronged approach, we submit that the $T_{cc}^{+}$ is a shallow $D^0D^{*+}$ bound state in the second Riemann sheet of the complex plane.

研究の動機と目的

• Motivate and resolve the uncertain nature of the Tcc+ near-threshold state through a topological pole analysis.
• Develop a data-driven framework that uses deep neural networks to infer pole topology from near-threshold scattering line shapes.
• Employ a uniformized S-matrix and a complementary K-matrix parameterization to extract pole positions and trajectories.
• Quantify the robustness of the dominant pole topology against left-hand cut effects in a coupled-channel setting.

提案手法

• Construct a coupled-channel S-matrix formulation with four Riemann sheets and uniformization to a single ω-plane.
• Define a Jost-like function Q(ω) whose zeros place poles on selected sheets; include a regulator pole to satisfy asymptotic unitarity.
• Model left-hand cut effects phenomenologically via an arctan phase modification of diagonal S-matrix elements with a scale Λ and strength ζk.
• Generate 35 pole topology classes as labels via triplets [N_bt,N_bb,N_tb] and create two ensembles with/without left-hand cuts for training.
• Build a 114-dimensional input vector from discretized energy points and differential event rates to feed fully connected DNN classifiers.
• Train an ensemble of 216 DNN models (with manual grid search and Optuna optimization) using Adam, dropout, and weight decay, for pole-classification.

実験結果

リサーチクエスチョン

• RQ1 What is the dominant pole topology of the Tcc+ state across the coupled D0D*+ and D0D0π+ channels?
• RQ2 Can DNN-based classification reliably identify pole topologies from near-threshold line shapes under left-hand cut effects?
• RQ3 How do uniformized S-matrix and K-matrix parameterizations compare in extracting pole positions and trajectories?
• RQ4 Is Tcc+ best described as a shallow D0D*+ bound state on the second Riemann sheet in the analyzed framework?

主な発見

• An ensemble of DNN classifiers identifies a dominant pole topology on the [bt] sheet robust against left-hand cut effects.
• A two-pronged pole parameter extraction (uniformized S-matrix and coupled-channel K-matrix) provides model-independent baseline and dynamical insight into pole trajectories.
• Using the outlined framework, the Tcc+ is concluded to be a shallow D0D*+ bound state in the second Riemann sheet of the complex plane.
• The approach reduces a 35-topology space to a probable class efficiently, enabling robust inference from LHCb-like line shapes.
• Inclusion of left-hand cut effects via the arctan phase is shown to influence line shapes and pole interpretation, while preserving unitarity.