Special Session 1: Advancing Cyberspace Security with Neural Networks and AI
 Special Session 2: AI for Earth Sciences and Complex Systems

HP3C 2026 solicits proposals for special sessions within the technical scope of the conference. Special sessions supplement the regular program of the conference and provide state-of-the-art results of research on novel, challenging, and emerging topics.
Approved special session will be submitted for inclusion to various abstracting and indexing databases. Papers submitted to special sessions will have to be evaluated and peer-reviewed along the very same criteria of the regular sessions.
Invited sessions consist of at least 5 thematically related invited papers. Invited session proposals consist of a brief statement of purpose and extended abstracts of the included invited papers.
Please provide all the information requested.

1. Special Session包括至少5篇主题相关的论文，提案包括目的的简要说明和邀请论文的详细摘要。   
2. Special Session提案应由组织者提交，他们将作为special session主席或联合主席，促进Special Session的召开和处理Special Session文章审查过程。   
3.  提交给Special Session全文论文将按照与常会相同的标准进行评估和同行评审，并列入会议论文集。 也可以仅参与报告，文章不出版。

● HP3C special session proposal submission | Proposal 提交链接：http://hgg922om6hjbwthb.mikecrm.com/RbJFXBh

Proposals should include the following information:

• Title;
• Name(s) of organizer(s);
• Email of main contact person;
• Brief bio(s) of organizer(s);
• Brief description;
• Related topics;
• Potential participants;

● - 可申请专题列表 - HP3C 2026 solicits special sessions. You may formulate your own session topics or select one from the following options.

1. Compiler and Runtime Co-Design for Next-Generation HPC Systems

As HPC systems evolve towards greater heterogeneity and scale, traditional compilation techniques face challenges due to disconnection from runtime systems. This special session focuses on deep co-design and co-optimization between compile-time and runtime systems, aiming to enhance the performance and portability of complex HPC applications in dynamic environments. We will explore how to achieve full-stack performance optimization from static code generation to dynamic resource scheduling through compiler awareness of runtime states, runtime feedback-driven compilation, and cross-layer abstraction design.

2. Processing-in/Near-Memory Computing Architectures and Compilation Techniques

Processing-in-Memory and Near-Memory Computing are key technologies to break the "memory wall" and improve energy efficiency. This session aims to discuss hardware architectures, circuit designs, system integration supporting these novel computing paradigms, and the crucial programming models and compilation techniques. We focus on full-stack innovation from devices and architectures to software toolchains, to unleash the performance potential of in/near-memory computing for data-intensive HPC and AI applications.

3. Software Stacks and Algorithms for Quantum-Classical Hybrid Computing Systems

Processing-in-Memory and Near-Memory Computing are key technologies to break the "memory wall" and improve energy efficiency. This session aims to discuss hardware architectures, circuit designs, system integration supporting these novel computing paradigms, and the crucial programming models and compilation techniques. We focus on full-stack innovation from devices and architectures to software toolchains, to unleash the performance potential of in/near-memory computing for data-intensive HPC and AI applications.

4. High-Performance Sparse and Tensor Computation & Compilation

Sparsity and high-dimensional tensors are core data characteristics in scientific computing and artificial intelligence. This session is dedicated to addressing performance and programming challenges in sparse/irregular and tensor computations. We focus on specialized architectures for sparse/tensor data, efficient storage formats, automated sparsity discovery and optimization, and high-performance tensor algebra compilation techniques. The goal is to build efficient hardware-software co-designed solutions for challenges ranging from physical simulation to large model training.

5. Security and Trusted Execution for High-Performance Computing Systems

As HPC systems become more open and complex, their security faces severe challenges. This session focuses on full-stack security and trust issues in HPC systems, covering hardware security extensions (e.g., SGX, CXL security), secure compilation, to system-level protection and privacy-preserving computing. We explore how to provide confidentiality, integrity guarantees, and trusted execution environments for HPC workloads without significant performance penalty, especially for multi-tenant cloud HPC and sensitive data processing scenarios.

6. AI for Scientific Computing: Algorithms, Software, and Systems

Artificial Intelligence, particularly Scientific Machine Learning, is revolutionizing scientific discovery and engineering simulation. This session focuses on innovative algorithms, specialized software stacks, and supporting systems arising from the deep integration of AI and scientific computing. We explore physics-informed machine learning, neural networks for solving PDEs, AI-accelerated scientific simulation, and the underlying technologies such as automatic differentiation, high-performance training frameworks, and AI-HPC co-design, aiming to advance the paradigm shift in scientific research.

7. Programming Languages and Compilation for Emerging Computing Paradigms

Artificial Intelligence, particularly Scientific Machine Learning, is revolutionizing scientific discovery and engineering simulation. This session focuses on innovative algorithms, specialized software stacks, and supporting systems arising from the deep integration of AI and scientific computing. We explore physics-informed machine learning, neural networks for solving PDEs, AI-accelerated scientific simulation, and the underlying technologies such as automatic differentiation, high-performance training frameworks, and AI-HPC co-design, aiming to advance the paradigm shift in scientific research.