TRACE: Learning to Compute on Circuit Graphs

Learning to compute, the ability to model the functional behavior of a circuit graph, is a fundamental challenge for graph representation learning. Yet, the dominant paradigm is architecturally mismatched for this task. This flawed assumption, central to mainstream message passing neural networks (MPNNs) and their conventional Transformer-based counterparts, prevents models from capturing the position-aware, hierarchical nature of computation. To resolve this, we introduce TRACE, a new paradigm built on an architecturally sound backbone and a principled learning objective. First, TRACE employs a Hierarchical Transformer that mirrors the step-by-step flow of computation, providing a faithful architectural backbone that replaces the flawed permutation-invariant aggregation. Second, we introduce function shift learning, a novel objective that decouples the learning problem. Instead of predicting the complex global function directly, our model is trained to predict only the function shift, the discrepancy between the true global function and a simple local approximation that assumes input independence. We validate this paradigm on various circuits modalities, including Register Transfer Level graphs, And-Inverter Graphs and post-mapping netlists. Across a comprehensive suite of benchmarks, TRACE substantially outperforms all prior architectures. These results demonstrate that our architecturally-aligned backbone and decoupled learning objective form a more robust paradigm for the fundamental challenge of learning the functional behavior of a circuit graph.

• Publication year

  2025

• Venue

  Unknown venue

• Publication date

  2025-09-26

• Fields of study

  Computer Science

• Identifiers
  arXiv 2509.21886
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• FGNN2: A Powerful Pretraining Framework for Learning the Logic Functionality of Circuits

  2025cited by this paper
• DeepCell: Multiview Representation Learning for Post-Mapping Netlists

  2025influential reference
• ForgeEDA: A Comprehensive Multimodal Dataset for Advancing EDA

  2025influential reference
• CircuitFusion: Multimodal Circuit Representation Learning for Agile Chip Design

  2025influential reference
• NetTAG: A Multimodal RTL-and-Layout-Aligned Netlist Foundation Model via Text-Attributed Graph

  2025cited by this paper
• A Self-Supervised, Pre-Trained, and Cross-Stage-Aligned Circuit Encoder Provides a Foundation for Various Design Tasks

  2025influential reference
• DeepCircuitX: A Comprehensive Repository-Level Dataset for RTL Code Understanding, Generation, and PPA Analysis

  2025influential reference
• Circuit Representation Learning with Masked Gate Modeling and Verilog-AIG Alignment

  2025cited by this paper
• DeepSeq2: Enhanced Sequential Circuit Learning with Disentangled Representations

  2024influential reference
• DeepGate3: Towards Scalable Circuit Representation Learning

  2024cited by this paper
• Gamora: Graph Learning based Symbolic Reasoning for Large-Scale Boolean Networks

  2023cited by this paper
• DeepGate2: Functionality-Aware Circuit Representation Learning

  2023influential reference
• Simplifying and Empowering Transformers for Large-Graph Representations

  2023cited by this paper
• On EDA-Driven Learning for SAT Solving

  2022cited by this paper
• Recipe for a General, Powerful, Scalable Graph Transformer

  2022cited by this paper
• A graph placement methodology for fast chip design

  2021cited by this paper
• DeepGate: Learning Neural Representations of Logic Gates

  2021cited by this paper
• GRANNITE: Graph Neural Network Inference for Transferable Power Estimation

  2020cited by this paper
• On the Bottleneck of Graph Neural Networks and its Practical Implications

  2020cited by this paper
• Representation Learning with Contrastive Predictive Coding

  2018cited by this paper
• Ithemal: Accurate, Portable and Fast Basic Block Throughput Estimation using Deep Neural Networks

  2018cited by this paper
• Btor2 , BtorMC and Boolector 3.0

  2018cited by this paper
• Deeper Insights into Graph Convolutional Networks for Semi-Supervised Learning

  2018cited by this paper
• Neural Message Passing for Quantum Chemistry

  2017cited by this paper
• RT-Level ITC'99 Benchmarks and First ATPG Results

  2000cited by this paper

• No citing papers are available for this paper.