本论文研究的致密恒星主要包括白矮星和热亚矮星，它们是恒星演化的最后阶段。传统理论认为它们的磁活动微弱，难以产生耀发现象。然而，最新的观测数据挑战了这一认知。为了系统地探索致密恒星耀发活动的普遍性和多样性，我们基于TESS第1–5周期共观测的38012条2分钟曝光时间的光变曲线数据，对7505颗白矮星和4113颗热亚矮星开展了全面的耀发搜寻与性质研究。
首先，我们开发了一套专门针对致密恒星光变曲线的耀发检测和验证流程。其中，我们创新性地提出了一种去除光变曲线中短周期信号方法，有效克服了致密恒星普遍存在的短时标变化对耀发检测的干扰；并通过交叉比对激变变星星表和小行星数据库，排除了这两类天体可能引入的虚假信号；最后利用机器学习实现了高效的耀发候选事件验证。这一系列措施确保了我们的检测结果具有较高的可靠性和完整性。
我们从193颗致密恒星中识别出了1016个耀发事件，其中182个来自58颗热亚矮星，834个来自135颗白矮星，由此构建了首个致密恒星耀发样本。为了进一步评估潜在的来自视场中周边天体或者伴星的污染，我们利用自主开发的tpfi软件包生成了TESS目标像素文件证认图，对每一个天体的周边天区进行目视检查。我们还使用GaiaDR3数据和光谱能量分布拟合来寻找潜在的主序伴星。
结果表明，我们的致密恒星耀发事件样本普遍受到了不同程度的污染，还没有确凿的证据支持任何一例耀发完全源自单个致密恒星。对于白矮星，观测到的耀发很可能主要源自其晚型主序伴星，因为后者通常光度与白矮星相当，并且耀发发生频繁。相比之下，热亚矮星样本则更值得深入探究。基于热亚矮星的光度通常显著高于其主序伴星的事实，我们推测，在热亚矮星双星系统中观测到的某些耀发事件可能确实源自热亚矮星本身，或者源自其与伴星的相互作用。
为了检验这一设想，我们从耀发热亚矮星中进一步筛选出13颗无视场污染的目标，构建了一个高纯度的精选热亚矮星样本。这一精选样本的累积耀发频率分布显示出了与晚型主序星明显不同的特征：其累积耀发频率分布的幂律指数（=1.70±0.19）略低于晚型星的典型值（∼2），这表明高能量耀发在其中占据了更高的比例。考虑到热亚矮星在内部结构上与早型主序星的相似性，即二者均具有辐射主导的包层，以及后者也表现出类似的耀发频率分布特征，我们推测热亚矮星耀发的物理机制可能与晚型星有本质区别，且更类似于早型星。
总体来说，本研究在致密恒星耀发研究领域迈出了关键的一步。尽管受限于当前数据，我们仍无法确定地证认单个致密恒星上的耀发现象，但通过系统的样本构建和分析，我们已经初步发现了一批值得深入研究的特殊个例，为将来进一步的观测、理论工作指明了方向。未来，随着更多的大型巡天项目的推进，我们有望获得前所未有的海量、高质量数据，从而真正厘清致密恒星耀发乃至整个致密恒星磁活动的来龙去脉。

The compact stars studied in this thesis are primarily white dwarfs and hot subdwarfs, which represent the final stages of stellar evolution. Traditionally, they were thought to exhibit weak magnetic activity and were unlikely to produce flares. However, the latest observational data have challenged this perception. To systematically explore the prevalence and diversity of flaring activity in compact stars, we conducted a comprehensive search and characterization of flare events on 7505 white dwarfs and 4113 hot subdwarfs using all of the 38 012 2-minute cadence light curve data from Cycles 1–5 of the TESS mission.
First, we developed a dedicated flare detection and validation pipeline specifically tailored for light curves of compact stars. We innovatively proposed a method to remove short-period signals from the light curves, effectively overcoming the interference of short-timescale variations commonly found in compact stars on flare detection. We also cross-matched with catalogs of cataclysmic variables and asteroids to eliminate false signals that might be introduced by these two types of objects. Finally, we used machine learning to achieve efficient validation of flare candidates. These measures ensure that our detection results have high reliability and completeness.
We identified 1016 flare events from 193 compact stars, including 182 events from 58 hot subdwarfs and 834 events from 135 white dwarfs, thus constructing the first flare sample for compact stars. To further assess potential contamination from nearby objects in the field of view or companion stars, we generated TESS target pixel file validation images using our self-developed tpfi software package and visually inspected the surrounding regions of each object. We also used Gaia DR3 data and spectral energy distribution fitting to search for potential main-sequence companions.
The results indicate that our compact star flare event sample is generally affected by contamination to varying degrees, and there is no conclusive evidence supporting any flare originating entirely from a single compact star. For white dwarfs, the observed flares are most likely primarily from their late-type main-sequence companions, as the latter usually have comparable luminosity to white dwarfs and exhibit frequent flares. In contrast, the hot subdwarf sample is more worthy of in-depth investigation. Based on the fact that hot subdwarfs are usually significantly more luminous than their main-sequence companions, we speculate that certain flare events observed in hot subdwarf binary systems may indeed originate from the hot subdwarfs themselves or from their interactions with companions.
To test this hypothesis, we further selected 13 targets from the flaring hot subdwarfs that have no field contamination and constructed a high-purity refined hot subdwarf sample. The cumulative flare frequency distribution of this refined sample shows distinct characteristics compared to late-type main-sequence stars: the power-law index of its cumulative flare frequency distribution ( = 1.70±0.19) is slightly lower than the typical value for late-type stars ( ∼ 2), indicating a higher proportion of high-energy flares. Considering the similarity in internal structure between hot subdwarfs and earlytype main-sequence stars, both having radiation-dominated envelopes, and the fact that the latter also exhibit similar flare frequency distribution characteristics, we speculate that the physical mechanism of hot subdwarf flares may be fundamentally different from that of late-type stars and more similar to early-type stars.
Overall, this study takes a key step forward in the field of compact star flare research. Although limited by current data, we cannot definitively identify flare phenomena on individual compact stars, through systematic sample construction and analysis, we have preliminarily discovered a number of special cases worthy of in-depth study, pointing the direction for future observations and theoretical work. As more large-scale survey projects progress, we expect to obtain unprecedented massive and high-quality data, thus truly unraveling the origins and evolution of compact star flares and magnetic activity in compact stars as a whole.