Accurate modeling of galaxy distributions is paramount for cosmological analysis using galaxy redshift surveys. However, this endeavor is often hindered by the computational complexity of resolving the dark matter halos that host these galaxies. To address this challenge, we propose the development of effective assembly bias models down to small scales, i.e., going beyond the local density dependence capturing non-local cosmic evolution. We introduce a hierarchical cosmic web classification that indirectly captures up to third-order long- and short-range non-local bias terms. This classification system also enables us to maintain positive definite parametric bias expansions. Specifically, we subdivide the traditional cosmic web classification, which is based on the eigenvalues of the tidal field tensor, with an additional classification based on the Hessian matrix of the negative density contrast. We obtain the large-scale dark matter field on a mesh with ~3.9 h ^-1 Mpc cell side resolution through Augmented Lagrangian Perturbation Theory. To assess the effectiveness of our model, we conduct tests using a reference halo catalogue extracted from the UNIT project simulation, which was run within a cubical volume of 1 h ^-1 Gpc side. The resulting mock halo catalogs, generated through our approach, exhibit a high level of accuracy in terms of the one-, two- and three-point statistics. They reproduce the reference power-spectrum within better than 2 percent accuracy up to wavenumbers k ~ 0.8 h Mpc^-1 and provide accurate bispectra within the scales that are crucial for cosmological analysis. This effective bias approach provides a forward model appropriate for field-level cosmological inference and holds significant potential for facilitating cosmological analysis of galaxy redshift surveys, particularly in the context of projects such as DESI, EUCLID, and LSST.

中文翻译：

---

分层宇宙网和装配偏差


星系分布的准确建模对于使用星系红移巡天进行宇宙学分析至关重要。然而，这一努力常常受到解决这些星系暗物质晕的计算复杂性的阻碍。为了应对这一挑战，我们建议开发小尺度的有效组装偏差模型，即超越局域密度依赖性，捕获非局域宇宙演化。我们引入了一种分层宇宙网络分类，可以间接捕获高达三阶的长程和短程非局部偏差项。该分类系统还使我们能够维持正定参数偏差扩展。具体来说，我们将基于潮汐场张量特征值的传统宇宙网分类细分为基于负密度对比的 Hessian 矩阵的附加分类。我们在网格上获得了约 3.9 的大规模暗物质场H通过增强拉格朗日微扰理论实现^-1 Mpc 单元侧分辨率。为了评估我们模型的有效性，我们使用从 UNIT 项目模拟中提取的参考光环目录进行测试，该模拟在 1 的立方体积内运行H ^-1 Gpc 侧。通过我们的方法生成的模拟光环目录在一点、两点和三点统计方面表现出很高的准确性。它们以高于 2% 的准确度再现高达波数的参考功率谱k 〜0.8 H Mpc ^-1并提供对宇宙学分析至关重要的尺度内的准确双谱。 这种有效的偏差方法提供了一个适合现场级宇宙学推断的正向模型，并且在促进星系红移巡天的宇宙学分析方面具有巨大的潜力，特别是在 DESI、EUCLID 和 LSST 等项目的背景下。