Paper List
-
Evolutionarily Stable Stackelberg Equilibrium
通过要求追随者策略对突变入侵具有鲁棒性，弥合了斯塔克尔伯格领导力模型与演化稳定性之间的鸿沟。
-
Recovering Sparse Neural Connectivity from Partial Measurements: A Covariance-Based Approach with Granger-Causality Refinement
通过跨多个实验会话累积协方差统计，实现从部分记录到完整神经连接性的重建。
-
Atomic Trajectory Modeling with State Space Models for Biomolecular Dynamics
ATMOS通过提供一个基于SSM的高效框架，用于生物分子的原子级轨迹生成，弥合了计算昂贵的MD模拟与时间受限的深度生成模型之间的差距。
-
Slow evolution towards generalism in a model of variable dietary range
通过证明是种群统计噪声（而非确定性动力学）驱动了模式形成和泛化食性的演化，解决了间接竞争下物种形成的悖论。
-
Grounded Multimodal Retrieval-Augmented Drafting of Radiology Impressions Using Case-Based Similarity Search
通过将印象草稿基于检索到的历史病例，并采用明确引用和基于置信度的拒绝机制，解决放射学报告生成中的幻觉问题。
-
Unified Policy–Value Decomposition for Rapid Adaptation
通过双线性分解在策略和价值函数之间共享低维目标嵌入，实现对新颖任务的零样本适应。
-
Mathematical Modeling of Cancer–Bacterial Therapy: Analysis and Numerical Simulation via Physics-Informed Neural Networks
提供了一个严格的、无网格的PINN框架，用于模拟和分析细菌癌症疗法中复杂的、空间异质的相互作用。
-
Sample-Efficient Adaptation of Drug-Response Models to Patient Tumors under Strong Biological Domain Shift
通过从无标记分子谱中学习可迁移表征，利用最少的临床数据实现患者药物反应的有效预测。
Geometric framework for biological evolution
Unknown
30秒速读
IN SHORT: This paper addresses the fundamental challenge of developing a coordinate-independent, geometric description of evolutionary dynamics that bridges genotype and phenotype spaces, revealing evolution as a learning process.
核心创新
- Methodology Establishes a generally covariant framework for evolutionary dynamics that operates consistently across genotype and phenotype spaces, enabling coordinate-independent analysis.
- Theory Demonstrates through maximum entropy principle that the inverse metric tensor equals the covariance matrix, transforming the Lande equation into a covariant gradient ascent equation.
- Methodology Models evolution as a learning process where the specific optimization algorithm is determined by the functional relationship g(κ) between metric tensor and noise covariance.
主要结论
- The maximum entropy principle yields fundamental identification: g^{αr,βs} = c^{αr,βs} (inverse metric equals genotypic covariance matrix).
- The Lande equation transforms to covariant gradient ascent: dx̄^i/dt = G^{ij}(x̄) ∂ℱ(x̄)/∂x̄^j, where G^{ij} = C^{ij} (inverse phenotype metric equals phenotypic covariance).
- Evolution implements specific learning algorithms determined by functional relation g(κ) between metric and noise covariance, with three regimes identified: quantum (α=1), efficient learning (α=1/2), and equilibration (α=0).
摘要: We develop a generally covariant description of evolutionary dynamics that operates consistently in both genotype and phenotype spaces. We show that the maximum entropy principle yields a fundamental identification between the inverse metric tensor and the covariance matrix, revealing the Lande equation as a covariant gradient ascent equation. This demonstrates that evolution can be modeled as a learning process on the fitness landscape, with the specific learning algorithm determined by the functional relation between the metric tensor and the noise covariance arising from microscopic dynamics. While the metric (or the inverse genotypic covariance matrix) has been extensively characterized empirically, the noise covariance and its associated observable (the covariance of evolutionary changes) have never been directly measured. This poses the experimental challenge of determining the functional form relating metric to noise covariance.