Mingxuan Zhu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Mingxuan Zhu has a total of 23 co-authors. Here are the top co-authors by number of shared publications:
- Dan Hao
Key Lab of HCST (PKU), MOE; School of Computer Science, Peking University, China
5 shared publications - Zeyu Sun
National Key Laboratory of Space Integrated Information System, Institute of Software, Chinese Academy of Sciences, China
3 shared publications - Guoqing Wang
Key Lab of HCST (PKU), MOE; School of Computer Science, Peking University, China
2 shared publications - Jinhao Dong
Renmin University of China, Beijing, China
2 shared publications - Qingyuan Liang
Peking University, Beijing, China
2 shared publications - Yujie Liu
Key Laboratory of High Confidence Software Technologies (Peking University), Ministry of Education; School of Computer Science, Peking University, China
2 shared publications
Latest Publications
- Open Access
- 16 November 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Software Engineering and Methodology
Isolating Compiler Bugs through Compilation Steps Analysis
Compilers are essential to software systems, and their bugs can propagate to dependent software. Ensuring compiler correctness is critical. However, isolating compiler bugs remains challenging due to the internal complexity of compiler execution. Existing techniques primarily mutate compilation inputs to generate passing and failing tests, but often lack causal analysis of internal steps, limiting their effectiveness. To address this limitation, we propose CompSCAN, a novel compiler bug isolation technique that applies analysis over the sequence of compilation steps. CompSCAN follows a three-stage process: (1) extracting the array of compilation steps that leads to the original failure, (2) identifying bug-causing steps and collecting corresponding compiler code elements, and (3) calculating suspicious scores for each code element and outputting a suspicious ranking list as the bug isolation result. We evaluate CompSCAN on 185 real-world LLVM and GCC bugs. Results show that CompSCAN outperforms state-of-the-art techniques in both effectiveness and efficiency. CompSCAN successfully isolates 50, 85, 100, and 123 bugs within the Top-1/3/5/10 ranks, respectively. Compared with ETEM and ODFL, two state-of-the-art compiler bug isolation techniques, CompSCAN achieves relative improvements of 44.51% / 50.18% / 36.24% / 24.49% over ETEM, and 31.58% / 49.12% / 44.93% / 21.78% over ODFL on those metrics. Moreover, CompSCAN runs faster on average per bug than both baselines.
- 16 November 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Software Engineering and Methodology
Open AccessIsolating Compiler Bugs through Compilation Steps Analysis
Compilers are essential to software systems, and their bugs can propagate to dependent software. Ensuring compiler correctness is critical. However, isolating compiler bugs remains challenging due to the internal complexity of compiler execution. Existing techniques primarily mutate compilation inputs to generate passing and failing tests, but often lack causal analysis of internal steps, limiting their effectiveness. To address this limitation, we propose CompSCAN, a novel compiler bug isolation technique that applies analysis over the sequence of compilation steps. CompSCAN follows a three-stage process: (1) extracting the array of compilation steps that leads to the original failure, (2) identifying bug-causing steps and collecting corresponding compiler code elements, and (3) calculating suspicious scores for each code element and outputting a suspicious ranking list as the bug isolation result. We evaluate CompSCAN on 185 real-world LLVM and GCC bugs. Results show that CompSCAN outperforms state-of-the-art techniques in both effectiveness and efficiency. CompSCAN successfully isolates 50, 85, 100, and 123 bugs within the Top-1/3/5/10 ranks, respectively. Compared with ETEM and ODFL, two state-of-the-art compiler bug isolation techniques, CompSCAN achieves relative improvements of 44.51% / 50.18% / 36.24% / 24.49% over ETEM, and 31.58% / 49.12% / 44.93% / 21.78% over ODFL on those metrics. Moreover, CompSCAN runs faster on average per bug than both baselines.
- 25 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Software Engineering
- Vol. 51 (9) , 2583-2600
Directional Diffusion-Style Code Editing Pre-Training
Code pre-trained models have shown promising effectiveness in various software engineering tasks. Among these tasks, many tasks are related to software evolution and/or code editing. However, existing code pre-trained models often overlook the real-world code editing data and the evolutionary nature of the editing process. In this paper, to simulate the step-by-step code editing process of human developers, we propose DivoT5, a pre-trained model based on directional diffusion at the data level. In DivoT5, we adopt two categories of pre-training tasks. The first category is mask and denoising tasks augmented with a diffusion direction representing code evolution. That is, we first apply a noising process to the code snippets before evolution, and then ask the pre-training process to restore the snippets with noise into the code snippets after evolution. The second category is tasks aiming to reinforce the evolutionary direction. That is, we first generate various intermediate versions for each pair of snippets before and after evolution, and then ask the pre-training process to transform the intermediate versions into the snippet after evolution for each pair. We evaluate DivoT5 for two code-editing scenarios (including a number of tasks) and one non-editing scenario using four downstream tasks. For each downstream task, we fine-tune the pre-trained DivoT5 on multiple corresponding datasets and evaluate its effectiveness across diverse scenarios Our experimental results show that ivoT5 achieves state-of-the-art (SOTA) performance on most tasks in comparison to models of the same scale (220M), large-scale (770M, 6.7B) models in fine-tuning, and billion-scale (6.7B, 8B, ChatGPT) instruct models in few-shot settings. For one code-editing task (i.e., CodeReview in NL-based CodeRefinement task), DivoT5 pre-trained on top of CodeT5-small (60M) can even outperform CodeT5-base (220M) and other pre-trained models with 220M parameters except for DivoT5 pre-trained on top of CodeT5-base (220M).
- 25 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Software Engineering
- Vol. 51 (9) , 2583-2600
Directional Diffusion-Style Code Editing Pre-Training
Code pre-trained models have shown promising effectiveness in various software engineering tasks. Among these tasks, many tasks are related to software evolution and/or code editing. However, existing code pre-trained models often overlook the real-world code editing data and the evolutionary nature of the editing process. In this paper, to simulate the step-by-step code editing process of human developers, we propose DivoT5, a pre-trained model based on directional diffusion at the data level. In DivoT5, we adopt two categories of pre-training tasks. The first category is mask and denoising tasks augmented with a diffusion direction representing code evolution. That is, we first apply a noising process to the code snippets before evolution, and then ask the pre-training process to restore the snippets with noise into the code snippets after evolution. The second category is tasks aiming to reinforce the evolutionary direction. That is, we first generate various intermediate versions for each pair of snippets before and after evolution, and then ask the pre-training process to transform the intermediate versions into the snippet after evolution for each pair. We evaluate DivoT5 for two code-editing scenarios (including a number of tasks) and one non-editing scenario using four downstream tasks. For each downstream task, we fine-tune the pre-trained DivoT5 on multiple corresponding datasets and evaluate its effectiveness across diverse scenarios Our experimental results show that ivoT5 achieves state-of-the-art (SOTA) performance on most tasks in comparison to models of the same scale (220M), large-scale (770M, 6.7B) models in fine-tuning, and billion-scale (6.7B, 8B, ChatGPT) instruct models in few-shot settings. For one code-editing task (i.e., CodeReview in NL-based CodeRefinement task), DivoT5 pre-trained on top of CodeT5-small (60M) can even outperform CodeT5-base (220M) and other pre-trained models with 220M parameters except for DivoT5 pre-trained on top of CodeT5-base (220M).
- Open Access
- 19 June 2025
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Software Engineering
- Vol. 2 (FSE) , 847-867
PDCAT: Preference-Driven Compiler Auto-tuning
Compilers are crucial software tools that usually convert programs in high-level languages into machine code. A compiler provides hundreds of optimizations to improve the performance of the compiled code, which are controlled by enabled or disabled optimization flags. However, the vast number of combinations of these flags makes it extremely challenging to select the desired settings for compiler optimization flags (i.e., an optimization sequence) for a given target program. In the literature, many auto-tuning techniques have been proposed to select a desired optimization sequence via different strategies across the entire optimization space. However, due to the huge optimization space, these techniques commonly suffer from the widely recognized efficiency problem. To address this problem, in this paper, we propose a preference-driven selection approach PDCAT, which reduces the search space of optimization sequences through three components. In particular, PDCAT first identifies combined optimizations based on compiler documentation to exclude optimization sequences violating the combined constraints, and then categorizes the optimizations into a common optimization set (whose optimization flags are fixed) and an exploration set containing the remaining optimizations. Finally, within the search process, PDCAT assigns distinct enable probabilities to the explored optimization flags and finally selects a desired optimization sequence. The former two components reduce the search space by removing invalid optimization sequences and fixing some optimization flags, whereas the latter performs a biased search in the search space. To evaluate the performance of the proposed approach PDCAT, we conducted an extensive experimental study on the latest version of the GCC compiler with two widely used benchmarks, cBench and PolyBench. The results show that PDCAT significantly outperforms the four compared techniques, including the state-of-art technique SRTuner. Moreover, each component of PDCAT not only contributes to its performance, but also improves the acceleration performance of the compared techniques.
- 19 June 2025
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Software Engineering
- Vol. 2 (FSE) , 847-867
Open AccessPDCAT: Preference-Driven Compiler Auto-tuning
Compilers are crucial software tools that usually convert programs in high-level languages into machine code. A compiler provides hundreds of optimizations to improve the performance of the compiled code, which are controlled by enabled or disabled optimization flags. However, the vast number of combinations of these flags makes it extremely challenging to select the desired settings for compiler optimization flags (i.e., an optimization sequence) for a given target program. In the literature, many auto-tuning techniques have been proposed to select a desired optimization sequence via different strategies across the entire optimization space. However, due to the huge optimization space, these techniques commonly suffer from the widely recognized efficiency problem. To address this problem, in this paper, we propose a preference-driven selection approach PDCAT, which reduces the search space of optimization sequences through three components. In particular, PDCAT first identifies combined optimizations based on compiler documentation to exclude optimization sequences violating the combined constraints, and then categorizes the optimizations into a common optimization set (whose optimization flags are fixed) and an exploration set containing the remaining optimizations. Finally, within the search process, PDCAT assigns distinct enable probabilities to the explored optimization flags and finally selects a desired optimization sequence. The former two components reduce the search space by removing invalid optimization sequences and fixing some optimization flags, whereas the latter performs a biased search in the search space. To evaluate the performance of the proposed approach PDCAT, we conducted an extensive experimental study on the latest version of the GCC compiler with two widely used benchmarks, cBench and PolyBench. The results show that PDCAT significantly outperforms the four compared techniques, including the state-of-art technique SRTuner. Moreover, each component of PDCAT not only contributes to its performance, but also improves the acceleration performance of the compared techniques.
- Open Access
- 18 January 2025
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Software Engineering and Methodology
- Vol. 34 (2) , 1-28
Is It Hard to Generate Holistic Commit Message?
Commit messages are important for developers to understand the content and the reason for code changes. However, poor and even empty commit messages widely exist. To improve the quality of commit messages and development efficiency, many commit message generation methods have been proposed. Nevertheless, previous methods mainly focus on a brief generation problem, where both the input code change and the output commit messages are restricted to short. This may initiate a debate on the performance of these methods in practice. In this article, we attempt to remove the restrictions and move the needle forward to a holistic commit message generation problem. In particular, we conduct experiments to evaluate the performance of existing commit message generation methods in holistic commit message generation. In the experiments, we choose seven state-of-the-art commit generation methods and focus on two important scenarios in commit message generation (i.e., the within-project scenario and the cross-project scenario). To conduct our experiments, we publish a holistic commit message dataset HORDA with test data manually labeled. In our evaluations, we find that in generating holistic commit messages, the IR-based method has a better performance than non-pre-trained generation-based methods in the within-project scenario, contradicting previous research findings. Further, while the pre-trained generation-based methods are better than non-pre-trained generation-based methods, they are still constrained by the limitations of generation models.
- 18 January 2025
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Software Engineering and Methodology
- Vol. 34 (2) , 1-28
Open AccessIs It Hard to Generate Holistic Commit Message?
Commit messages are important for developers to understand the content and the reason for code changes. However, poor and even empty commit messages widely exist. To improve the quality of commit messages and development efficiency, many commit message generation methods have been proposed. Nevertheless, previous methods mainly focus on a brief generation problem, where both the input code change and the output commit messages are restricted to short. This may initiate a debate on the performance of these methods in practice. In this article, we attempt to remove the restrictions and move the needle forward to a holistic commit message generation problem. In particular, we conduct experiments to evaluate the performance of existing commit message generation methods in holistic commit message generation. In the experiments, we choose seven state-of-the-art commit generation methods and focus on two important scenarios in commit message generation (i.e., the within-project scenario and the cross-project scenario). To conduct our experiments, we publish a holistic commit message dataset HORDA with test data manually labeled. In our evaluations, we find that in generating holistic commit messages, the IR-based method has a better performance than non-pre-trained generation-based methods in the within-project scenario, contradicting previous research findings. Further, while the pre-trained generation-based methods are better than non-pre-trained generation-based methods, they are still constrained by the limitations of generation models.
- Open Access
- 27 October 2024
- Published by Association for Computing Machinery (ACM)
- p. 819-830
Compiler Bug Isolation via Enhanced Test Program Mutation
Compilers are one of the most fundamental software systems. A large number of software systems rely on compilers for execution. Compiler bugs can significantly hinder software developers from diagnosing issues within their software. Therefore, it is essential to ensure the correctness of compilers and to isolate and fix compiler bugs. Isolating bugs within compilers is challenging due to compilers' complexity and large codebase. The prior studies on compiler bug isolation struggle to generate sufficient test cases for bug isolation and are not effective enough. To improve compiler bug isolation, in this paper, we propose a new compiler bug isolation technique ETEM which generates diverse test programs using feature mutation operators. Specifically, ETEM first searches for simplified optimization configurations that turn off fine-grained options as much as possible to narrow the coverage. Then, ETEM uses a joint reinforcement learning-based search method for mutation operators and mutate positions to get more high-quality test programs. Finally, ETEM uses rank-based aggregation to calculate each file's suspicious value. This paper also evaluates the effectiveness of the proposed technique and the contribution of each component on a dataset consisting of 120 real bugs in GCC and LLVM. The experimental results indicate that ETEM outperforms existing techniques in terms of efficiently isolating compiler bugs. Furthermore, each component of ETEM contributes to its effectiveness. Specifically, ETEM effectively isolates about 23.3%/47.5%/61.4%/76.6% bugs within Top-1/5/10/20. The overall improvements of ETEM over the state-of-the-art techniques ODFL in terms of Top-1/5/10/20 are 64.7%/35.7%/20.3%/15.0%.
- 27 October 2024
- Published by Association for Computing Machinery (ACM)
- p. 819-830
Open AccessCompiler Bug Isolation via Enhanced Test Program Mutation
Compilers are one of the most fundamental software systems. A large number of software systems rely on compilers for execution. Compiler bugs can significantly hinder software developers from diagnosing issues within their software. Therefore, it is essential to ensure the correctness of compilers and to isolate and fix compiler bugs. Isolating bugs within compilers is challenging due to compilers' complexity and large codebase. The prior studies on compiler bug isolation struggle to generate sufficient test cases for bug isolation and are not effective enough. To improve compiler bug isolation, in this paper, we propose a new compiler bug isolation technique ETEM which generates diverse test programs using feature mutation operators. Specifically, ETEM first searches for simplified optimization configurations that turn off fine-grained options as much as possible to narrow the coverage. Then, ETEM uses a joint reinforcement learning-based search method for mutation operators and mutate positions to get more high-quality test programs. Finally, ETEM uses rank-based aggregation to calculate each file's suspicious value. This paper also evaluates the effectiveness of the proposed technique and the contribution of each component on a dataset consisting of 120 real bugs in GCC and LLVM. The experimental results indicate that ETEM outperforms existing techniques in terms of efficiently isolating compiler bugs. Furthermore, each component of ETEM contributes to its effectiveness. Specifically, ETEM effectively isolates about 23.3%/47.5%/61.4%/76.6% bugs within Top-1/5/10/20. The overall improvements of ETEM over the state-of-the-art techniques ODFL in terms of Top-1/5/10/20 are 64.7%/35.7%/20.3%/15.0%.
- Open Access
- 18 April 2024
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Software Engineering and Methodology
- Vol. 33 (4) , 1-38
Compiler Autotuning through Multiple-phase Learning
Widely used compilers like GCC and LLVM usually have hundreds of optimizations controlled by optimization flags, which are enabled or disabled during compilation to improve the runtime performance (e.g., small execution time) of the compiler program. Due to the large number of optimization flags and their combination, it is difficult for compiler users to manually tune compiler optimization flags. In the literature, a number of autotuning techniques have been proposed, which tune optimization flags for a compiled program by comparing its actual runtime performance with different optimization flag combinations. Due to the huge search space and heavy actual runtime cost, these techniques suffer from the widely recognized efficiency problem. To reduce the heavy runtime cost, in this article we propose a lightweight learning approach that uses a small number of actual runtime performance data to predict the runtime performance of a compiled program with various optimization flag combinations. Furthermore, to reduce the search space, we design a novel particle swarm algorithm that tunes compiler optimization flags with the prediction model. To evaluate the performance of the proposed approach, CompTuner, we conduct an extensive experimental study on two popular C compilers, GCC and LLVM, with two widely used benchmarks, cBench and PolyBench. The experimental results show that CompTuner significantly outperforms the six compared techniques, including the state-of-the-art technique BOCA.
- 18 April 2024
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Software Engineering and Methodology
- Vol. 33 (4) , 1-38
Open AccessCompiler Autotuning through Multiple-phase Learning
Widely used compilers like GCC and LLVM usually have hundreds of optimizations controlled by optimization flags, which are enabled or disabled during compilation to improve the runtime performance (e.g., small execution time) of the compiler program. Due to the large number of optimization flags and their combination, it is difficult for compiler users to manually tune compiler optimization flags. In the literature, a number of autotuning techniques have been proposed, which tune optimization flags for a compiled program by comparing its actual runtime performance with different optimization flag combinations. Due to the huge search space and heavy actual runtime cost, these techniques suffer from the widely recognized efficiency problem. To reduce the heavy runtime cost, in this article we propose a lightweight learning approach that uses a small number of actual runtime performance data to predict the runtime performance of a compiled program with various optimization flag combinations. Furthermore, to reduce the search space, we design a novel particle swarm algorithm that tunes compiler optimization flags with the prediction model. To evaluate the performance of the proposed approach, CompTuner, we conduct an extensive experimental study on two popular C compilers, GCC and LLVM, with two widely used benchmarks, cBench and PolyBench. The experimental results show that CompTuner significantly outperforms the six compared techniques, including the state-of-the-art technique BOCA.
- Open Access
- 1 January 2022
- Published by Tech Science Press in Computers, Materials & Continua
- Vol. 71 (2) , 3635-3646
Evolution of Desertification Types on the North Shore of Qinghai Lake
Land desertification is a widely concerned ecological environment problem. Studying the evolution trend of desertification types is of great significance to prevent and control land desertification. In this study, we applied the decision tree classification method, to study the land area and temporal and spatial change law of different types of desertification in the North Bank of Qinghai Lake area from 1987 to 2014, based on the current land use situation and TM remote sensing image data of Haiyan County, Qinghai Province, The results show that the area of mild desertification land and moderate desertification land in the study area has decreased, while the area of severe desertification land and extreme desertification land has increased significantly in the past 30 years. The area of desertification land decreased by 4.02 km2, of which the area of mild and moderate desertification land decreased by 39.73 km2 and 36.8 km2 respectively, and the area of severe and extreme desertification land increased by 32.78 km2 and 39.73 km2 respectively. As for the mutual transformation relationship, the transformation from severe desertification land to extreme desertification land is the main, and the junction of severe desertification land and extreme desertification land is the sensitive area of transformation. In the north shore of Qinghai Lake, the sandy land tends to expand eastward. The research provides reference basis for local land desertification monitoring, and has a great guidance for local effective land desertification and soil and water conservation.
- 1 January 2022
- Published by Tech Science Press in Computers, Materials & Continua
- Vol. 71 (2) , 3635-3646
Open AccessEvolution of Desertification Types on the North Shore of Qinghai Lake
Land desertification is a widely concerned ecological environment problem. Studying the evolution trend of desertification types is of great significance to prevent and control land desertification. In this study, we applied the decision tree classification method, to study the land area and temporal and spatial change law of different types of desertification in the North Bank of Qinghai Lake area from 1987 to 2014, based on the current land use situation and TM remote sensing image data of Haiyan County, Qinghai Province, The results show that the area of mild desertification land and moderate desertification land in the study area has decreased, while the area of severe desertification land and extreme desertification land has increased significantly in the past 30 years. The area of desertification land decreased by 4.02 km2, of which the area of mild and moderate desertification land decreased by 39.73 km2 and 36.8 km2 respectively, and the area of severe and extreme desertification land increased by 32.78 km2 and 39.73 km2 respectively. As for the mutual transformation relationship, the transformation from severe desertification land to extreme desertification land is the main, and the junction of severe desertification land and extreme desertification land is the sensitive area of transformation. In the north shore of Qinghai Lake, the sandy land tends to expand eastward. The research provides reference basis for local land desertification monitoring, and has a great guidance for local effective land desertification and soil and water conservation.