The Path Not Taken: Duality in Reasoning about Program Execution
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866908987714174976 |
|---|---|
| author | Hasanov, Eshgin Sibat, Md Mahadi Hassan Karmaker, Santu Yadavally, Aashish |
| author_facet | Hasanov, Eshgin Sibat, Md Mahadi Hassan Karmaker, Santu Yadavally, Aashish |
| contents | Large language models (LLMs) have shown remarkable capabilities across diverse coding tasks. However, their adoption requires a true understanding of program execution rather than relying on surface-level patterns. Existing benchmarks primarily focus on predicting program properties tied to specific inputs (e.g., code coverage, program outputs). As a result, they provide a narrow view of dynamic code reasoning and are prone to data contamination. We argue that understanding program execution requires evaluating its inherent duality through two complementary reasoning tasks: (i) predicting a program's observed behavior for a given input, and (ii) inferring how the input must be mutated toward a specific behavioral objective. Both tasks jointly probe a model's causal understanding of execution flow. We instantiate this duality in DexBench, a benchmark comprising 445 paired instances, and evaluate 13 LLMs. Our results demonstrate that dual-path reasoning provides a robust and discriminative proxy for dynamic code understanding. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_20917 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | The Path Not Taken: Duality in Reasoning about Program Execution Hasanov, Eshgin Sibat, Md Mahadi Hassan Karmaker, Santu Yadavally, Aashish Machine Learning Artificial Intelligence Computation and Language Programming Languages Software Engineering I.2.7; D.2 Large language models (LLMs) have shown remarkable capabilities across diverse coding tasks. However, their adoption requires a true understanding of program execution rather than relying on surface-level patterns. Existing benchmarks primarily focus on predicting program properties tied to specific inputs (e.g., code coverage, program outputs). As a result, they provide a narrow view of dynamic code reasoning and are prone to data contamination. We argue that understanding program execution requires evaluating its inherent duality through two complementary reasoning tasks: (i) predicting a program's observed behavior for a given input, and (ii) inferring how the input must be mutated toward a specific behavioral objective. Both tasks jointly probe a model's causal understanding of execution flow. We instantiate this duality in DexBench, a benchmark comprising 445 paired instances, and evaluate 13 LLMs. Our results demonstrate that dual-path reasoning provides a robust and discriminative proxy for dynamic code understanding. |
| title | The Path Not Taken: Duality in Reasoning about Program Execution |
| topic | Machine Learning Artificial Intelligence Computation and Language Programming Languages Software Engineering I.2.7; D.2 |
| url | https://arxiv.org/abs/2604.20917 |