---
格式版本: 2
标题: "Probabilistic Inverse Modeling of Contaminant Transport via a Conditioned-on-Design Bayesian Physics Informed Neural Network"
原文链接: "https://arxiv.org/abs/2608.17591"
发布日期: "2026-08-18"
发布时间校准状态: "found"
发布时间需复核: "否"
发布时间来源: "rule:local:strict_original_body"
发布时间证据: "**\\[v1\\]** Tue, 18 Aug 2026 09:55:52 UTC (10,873 KB)"
发布时间校准原因: "规则确认唯一严格发布时间，来源 local:strict_original_body"
发布时间校准置信度: "high"
发布时间候选数量: 18
发布时间严格候选数量: 6
发布时间原页读取状态: "source template page reused from URL open"
发布时间未找到原因: ""
发布时间校准时间: "2026-08-20T15:59:53+08:00"
发布时间仲裁状态: "skipped"
发布时间仲裁尝试次数: 0
发布时间仲裁耗时毫秒: 0
发现时间: "2026-08-20T15:54:04+08:00"
入库时间: "2026-08-20T07:59:53.380Z"
来源平台: "arXiv 学术论文搜索"
搜索渠道: "source_template"
搜索词: "https://arxiv.org/search/?query=inference%20networking&searchtype=all"
匹配关键词:
  []
相关厂家:
  []
相关专家:
  []
内容类型: "网页"
抓取工具: "Free Fetch + Defuddle"
清洗工具: "Defuddle Markdown + Defuddle/Readability 正文提取"
原始附件:
  []
AI优质: "否"
AI打分: 5
AI分档: "非优质"
AI质检状态: "不通过"
AI打分理由: "内容为科学计算领域的污染物输运反演论文，与超节点、AI Rack、机柜级AI基础设施完全无关，无任何相关技术或商业信息。"
AI质检模型: "ali-deepseek-v4-flash"
AI质检时间: "2026-08-20T16:05:33+08:00"
AI主题相关性: 0
AI来源权威性: 5
AI新颖性: 0
AI技术细节: 0
AI商业部署信号: 0
AI完整性: 5
AI摘要: "该研究提出一种条件于设计的贝叶斯物理信息神经网络（CoDe-BPINN），结合域分解PINN与贝叶斯推断网络，用于多孔介质反应输运的逆问题。"
AI摘要模型: "ali-deepseek-v4-flash"
AI摘要时间: "2026-09-07T03:22:22.774Z"
采集批次: "2026年8月20日14点19分32秒"
采集批次ID: "20260820-141932-079"
去重键: "https://arxiv.org/abs/2608.17591"
---

## Physics > Computational Physics

## Title:Probabilistic Inverse Modeling of Contaminant Transport via a Conditioned-on-Design Bayesian Physics Informed Neural Network

[View PDF](https://arxiv.org/pdf/2608.17591) [HTML (experimental)](https://arxiv.org/html/2608.17591v1)

> Abstract:We address the inverse problem of reactive transport in heterogeneous porous media, where unknown model parameters must be inferred from sparse experimental observations. The problem is complicated by strong nonlinearities, spatial heterogeneity, and limited data availability. We propose a Conditioned-on-Design Bayesian Physics-Informed Neural Network (CoDe-BPINN), which combines a domain-decomposed PINN solver with a Bayesian inference network that learns the conditional distribution of model parameters given experimental design variables. The framework is trained by maximizing a physics-informed Evidence Lower Bound (ELBO), enabling simultaneous reconstruction of spatiotemporal concentration fields, probabilistic parameter estimation, and uncertainty quantification. We demonstrate the approach using laboratory experiments on contaminant transport through a multilayer porous column with an iodinated contrast medium. The model accurately reproduces breakthrough dynamics while revealing systematic parameter dependence on flow conditions. In particular, it identifies a nonlinear decrease in effective sorption capacity with increasing flow rate and porosity, consistent with kinetic limitations and reduced adsorbent mass. The Bayesian formulation also uncovers a strong negative correlation between sorption affinity and sorption capacity, quantifying the intrinsic non-identifiability of the inverse problem. CoDe-BPINN provides a robust framework for parameter inference and uncertainty quantification in data-scarce reactive transport problems.

| Comments: |  |
| --- | --- |
| Subjects: | Computational Physics (physics.comp-ph); Probability (math.PR) |
| MSC classes: | 76S05, 68T07 |
| ACM classes: | I.6.5; J.2; G.1.0; G.1.8; G.1.6; G.3 |
| Cite as: | [arXiv:2608.17591](https://arxiv.org/abs/2608.17591) \[physics.comp-ph\] |
|  | (or [arXiv:2608.17591v1](https://arxiv.org/abs/2608.17591v1) \[physics.comp-ph\] for this version) |
|  | [https://doi.org/10.48550/arXiv.2608.17591](https://doi.org/10.48550/arXiv.2608.17591) |

## Submission history

From: Milad Panahi \[[view email](https://arxiv.org/show-email/beb18bb4/2608.17591)\]  
**\[v1\]** Tue, 18 Aug 2026 09:55:52 UTC (10,873 KB)

[Which authors of this paper are endorsers?](https://arxiv.org/auth/show-endorsers/2608.17591) | Disable MathJax ([What is MathJax?](https://info.arxiv.org/help/mathjax.html))