Bogdan Alexandru Stoica

Weichen Li, Jiamin Song, Bogdan Alexandru Stoica et al.

Code transformation is a foundational capability in the software development process, where its effectiveness relies on constructing a high-quality code representation to characterize the input code semantics and guide the transformation. Existing approaches treat code transformation as an end-to-end learning task, leaving the construction of the representation needed for semantic reasoning implicit in model weights or relying on rigid compiler-level abstractions. We present SemRep, a framework that improves code transformation through generative code representation learning. Our key insight is to employ the semantics-preserving transformations as the intermediate representation, which serves as both a generative mid-training task and the guidance for subsequent instruction-specific code transformations. Across general code editing and optimization tasks (e.g., GPU kernel optimization), SemRep outperforms the extensively finetuned baselines with strictly the same training budget by 6.9% in correctness, 1.1x in performance, 13.9% in generalization, and 6.7% in robustness. With the improved exploration of diverse code transformations, SemRep is particularly amenable to evolutionary search. Combined with an evolutionary coding agent, SemRep finds optimizations that 685B larger-weight baselines fail to discover while achieving the same performance with 25% less inference compute.

Bogdan Alexandru Stoica, Swarup K. Sahoo, James R. Larus et al.

Dynamic program slicing can significantly reduce the code developers need to inspect by narrowing it down to only a subset of relevant program statements. However, despite an extensive body of research showing its usefulness, dynamic slicing is still short from production-level use due to the high cost of runtime instrumentation. As an alternative, we propose statistical program slicing, a novel hybrid dynamic-static slicing technique that explores the trade-off between accuracy and runtime cost. Our approach relies on modern hardware support for control flow monitoring and a novel, cooperative heap memory tracing mechanism combined with static program analysis for data flow tracking. We evaluate statistical slicing for debugging on 21 failures from 6 widely deployed applications and show it recovers 94% of the program statements on a dynamic slice with only 5% overhead.