本次实验全部评测数据均由数值仿真程序自主运算生成，各数据文件均为纯文本格式（.txt），单列数值存储，其中以Fitness_result_for_X.txt命名的文件记录算法在对应20个测试问题上30次独立运行所获得的目标函数最优值。搭建了覆盖500维、1000维、1500维、2000维梯度维度的多组高维优化测试样本，划分完全可分、部分可分、完全不可分、重叠耦合四类不同结构的优化样本子集；生成的数据分别用于算法横向性能对比、维度可扩展性测试、算法核心模块消融验证三大实验场景，每组样本完成30次独立仿真迭代后，输出均值、中位数、标准差、显著性检验p值、综合排名、收敛轨迹等多维度统计数据，完整量化算法的求解精度、收敛速度与寻优稳定性。
", "ds_source": "本数据集为完全仿真生成数据，不源自特定文献、实测或第三方下载
", "ds_process_way": "本文所有实验数据均依托自研仿真代码实时计算产出，以高维优化数学模型作为底层计算基础，通过自定义维度调节模块自动生成不同维度规模的测试样本，为所有样本统一设置3000×D次适应度评估计算上限，对每组样本循环执行30次独立进化仿真，程序全程自动采集迭代过程中粒子位置、速度、种群适应度、种群规模动态变化等原始运算数据，再通过内置数据统计脚本对原始仿真记录做批量处理，自动生成显著性检验指标、算法整体排名、收敛曲线坐标等可供分析的最终数据集，整套数据生成流程不导入外部采集或公开存储数据，全部数值由仿真程序实时运算得到。
", "ds_quality": "本数据集在生成和汇集过程中实施了系统性质量控制，确保了数据的完整性、准确性和代表性。
", "ds_acq_start_time": "2024-01-01 00:00:00", "ds_acq_end_time": null, "ds_acq_place": "南京", "ds_acq_lon_east": null, "ds_acq_lat_south": null, "ds_acq_lon_west": null, "ds_acq_lat_north": null, "ds_acq_alt_low": null, "ds_acq_alt_high": null, "ds_share_type": "open-access", "ds_total_size": 1087462, "ds_files_count": 0, "ds_format": "txt", "ds_space_res": "", "ds_time_res": "", "ds_coordinate": "无", "ds_projection": "", "ds_thumbnail": "e1d1d1e3-03eb-47a7-abf4-005b5c0c0d6a.png", "ds_thumb_from": 0, "ds_ref_way": "", "paper_ref_way": "", "ds_ref_instruction": "", "ds_from_station": "", "organization_id": "9ecaaa78-39e9-411e-9f24-274e12aa643f", "ds_serv_man": "杨强", "ds_serv_phone": "13570466708", "ds_serv_mail": "mmmyq@126.com", "doi_value": "", "subject_codes": [ "410" ], "quality_level": 0, "publish_time": "2026-07-09 10:59:07", "last_updated": "2026-07-09 10:59:07", "protected": false, "protected_to": "2028-06-30 00:00:00", "lang": "zh", "cstr": "", "i18n": { "en": { "title": "PTLSO Large-Scale Optimization Simulation Dataset", "ds_format": "txt", "ds_source": "This dataset is completely synthetically generated through simulation and does not originate from specific literature, field measurements, or third-party downloads.", "ds_quality": "During the generation and collection of this dataset, systematic quality control measures were implemented to ensure its completeness, accuracy, and representativeness.", "ds_ref_way": "", "ds_abstract": "All evaluation data in this experiment were automatically generated by numerical simulation programs. Each data file is in plain text format (.txt) and stores numerical values in a single column. Files named with the pattern Fitness_result_for_X.txt record the optimal objective function values obtained by the algorithm over 30 independent runs on the corresponding 20 test problems. Multiple high‑dimensional optimization test samples covering dimensionalities of 500, 1000, 1500, and 2000 were constructed, which are divided into four categories according to problem structures: fully separable, partially separable, fully non‑separable, and overlapping. The generated data are used for three major experimental scenarios: horizontal performance comparison among algorithms, scalability testing with respect to dimensionality, and ablation study of the core algorithm modules. For each set of samples, after 30 independent simulation iterations, multi‑dimensional statistical data are output, including mean, median, standard deviation, p‑values from significance tests, overall ranking, and convergence trajectories, which completely quantify the algorithm's solution accuracy, convergence speed, and optimization stability.", "ds_time_res": "", "ds_acq_place": "Nanjing", "ds_space_res": "", "ds_projection": "", "ds_process_way": "All experimental data in this paper are computed and generated in real time by self-developed simulation codes, with mathematical models of high-dimensional optimization serving as the fundamental computational basis. Test samples of different dimensional scales are automatically generated via a custom dimension adjustment module. A unified upper limit of 3000×D fitness evaluations is configured for all samples, and 30 independent evolutionary simulations are repeatedly executed for each group of samples. The program automatically collects raw computational data throughout iterations, including particle positions, velocities, swarm fitness values, and dynamic variations of population size. Built-in data statistics scripts then perform batch processing on raw simulation records to automatically generate analyzable final datasets containing significance test metrics, overall algorithm rankings, and coordinates of convergence curves. No externally collected or publicly archived datasets are imported in the entire data generation pipeline; all numerical values are calculated in real time by the simulation program.", "ds_ref_instruction": "" } }, "submit_center_id": "ncdc", "data_level": 0, "recommendation_value": 0, "license_type": "https://creativecommons.org/licenses/by/4.0/", "doi_reg_from": "reg_local", "cstr_reg_from": "reg_local", "doi_not_reg_reason": null, "cstr_not_reg_reason": null, "is_paper_in_submitting": false, "belong_to_nieer": false, "ds_topic_tags": [ "大规模优化", "概率更新", "锦标赛学习", "粒子群优化", "高维优化" ], "ds_subject_tags": [ "工程与技术科学基础学科" ], "ds_class_tags": [], "ds_locus_tags": [], "ds_time_tags": [], "ds_contributors": [ { "true_name": "杨强", "email": "mmmyq@126.com", "work_for": "南京信息工程大学", "country": "中国" } ], "ds_meta_authors": [ { "true_name": "杨强", "email": "mmmyq@126.com", "work_for": "南京信息工程大学", "country": "中国" }, { "true_name": "陆振宇", "email": "luzhenyu76@163.com", "work_for": "南京信息工程大学人工智能学院（未来技术学院）", "country": "中国" } ], "ds_managers": [ { "true_name": "杨强", "email": "mmmyq@126.com", "work_for": "南京信息工程大学", "country": "中国" } ], "category": "其他" }