arXiv subjects

KaiFan Ji

Publications and source records attributed to KaiFan Ji.

10 recordsLinked to original sources

arXiv2026

Reconstruction of ASO-S/HXI Solar Flare Hard X-ray Source Images with Physics-Constrained Deep Network

Solar flare hard X-ray imaging is a key diagnostic of flare energy release and electron acceleration. The ASO-S Hard X-ray Imager (HXI) employs 91 bi-grid sub-collimators, compressing the two-dimensional source distribution into a 91-dimensional counts vector---an inherently underdetermined inverse problem. The conventional CLEAN algorithm relies on a point-source prior and manual parameter tuning, while existing deep-learning methods (HXI-DLA) learn data-driven mappings without guaranteeing consistency with the forward physical equation. This paper introduces a physics-constrained deep learning framework whose core innovation is a counts mean--shape decoupling theory (DC--AC decomposition) derived from modulation imaging principles: the counts mean is proportional to total source energy and the normalized counts shape is determined by source position and scale, yielding two independently enforceable physical constraints. Based on this theory, HXI-PINN embeds the forward equation into both the network architecture---via ReLU non-negativity and counts-mean rescaling enforcing zero-error energy closure---and the optimization objective, where counts-domain constraints dominate the loss. Unlike data-driven approaches, HXI-PINN replaces heuristic regularization with executable hard constraints, ensuring every reconstruction satisfies the governing physics. Experiments on simulated Gaussian sources, soft X-ray morphologies, and a real HXI flare event confirm that the framework generalizes across source configurations, with advantages over CLEAN on ring-shaped sources and over HXI-DLA on complex morphologies. This work demonstrates that ``physical constraints + deep prior'' is an effective paradigm for underdetermined inversion---constraints anchor the solution in the feasible subspace satisfying the forward equation, while the deep prior selects the optimal solution within it.

astro-ph.SR↗

arXiv2025

Fundamental Parameters for Totally Eclipsing Contact Binaries Observed by TESS

Totally eclipsing contact binaries provide a unique opportunity to accurately determine mass ratios through photometric methods alone, eliminating the need for spectroscopic data. Studying low mass ratio (LMR) contact binaries is crucial for advancing our understanding of binary star evolution and the formation of rare optical transients known as red novae. We identified 143 totally eclipsing contact binaries from the Transiting Exoplanet Survey Satellite. These high-precision light curves reveal a distinct O'Connell effect, which we interpret by introducing a cool spot on the primary star. Training a neural network model that includes cool spot parameters can generate a high-precision light curve 2 orders of magnitude faster than Phoebe. Utilizing the neural network (NNnol3) model combined with the Markov Chain Monte Carlo algorithm, we rapidly derived the fundamental parameters of these systems. By leveraging the relationship between orbital period and semimajor axis using the Random Sample Consensus algorithm, we estimated their absolute parameters. Our analysis identified 96 targets with mass ratios below 0.25, all of which were not listed in any previous catalog, thus signifying the discovery of new LMR system candidates. Assuming all 143 binary systems are affected by a third light during parameter estimation, we train a neural network (NNl3) model considering the third light. Then we calculate the residuals between the mass ratio ql3 (considering the third light) and qnol3 (neglecting it). For these residuals, the 25th percentile (Q1) is 0.012, the median (Q2) is 0.026, and the 75th percentile (Q3) is 0.05.

astro-ph.SR↗

arXiv2025

Detection of Semidetached Eclipsing Binaries from TESS

Semidetached binaries, distinguished by their mass transfer phase, play a crucial role in elucidating the physics of mass transfer within interacting binary systems. To identify these systems in eclipsing binary light curves provided by large-scale time-domain surveys, we have developed a methodology by training two distinct models that establish a mapping relationship between the parameters (orbital parameters and physical parameters) of semidetached binaries and their corresponding light curves. The first model corresponds to scenarios where the more massive star fills its Roche lobe, while the second model addresses situations where the less massive star does so. In consideration of the O'Connell effect observed in the light curves, we integrated a cool spot parameter into our models, thereby enhancing their applicability to fit light curves that exhibit this phenomenon. Our two-model framework was then harmonized with the Markov Chain Monte Carlo algorithm, enabling precise and efficient light-curve fitting and parameter estimation. Leveraging 2 minute cadence data from the initial 67 sectors of the Transiting Exoplanet Survey Satellite, we successfully identified 327 systems where the less massive component fills its Roche lobe, alongside three systems where the more massive component fills its Roche lobe. Additionally, we offer comprehensive fundamental parameters for these binary systems, including orbital inclination, relative radius, mass ratio, and effective temperature.

astro-ph.SR↗

arXiv2025

A method of Extracting Flat Field from Real Time Solar Observation Data

Existing methods for obtaining flat field rely on observed data collected under specific observation conditions to determine the flat field. However, the telescope pointing and the column fixed pattern noise of the CMOS detector change during actual observations, causing residual signals in real time observation data after flat field correction, such as interference fringes and column fixed pattern noise. In actual observations, the slight wobble of the telescope caused by the wind leads to shifts in the observed data. In this paper, a method of extracting the flat field from the real time solar observation data is proposed. Firstly, the average flat field obtained by multi-frame averaging is used as the initial value. A set of real-time observation data is input into the KLL method to calculate the correction amount for the average flat field. Secondly, the average flat field is corrected using the calculated correction amount to obtain the real flat field for the current observation conditions. To overcome the residual solar structures caused by atmospheric turbulence in the correction amount, real-time observation data are grouped to calculate the correction amounts. These residual solar structures are suppressed by averaging multiple groups, improving the accuracy of the correction amount. The test results from space and ground-based simulated data demonstrate that our method can effectively calculate the correction amount for the average flat field. The NVST 10830 A/Ha data were also tested. High-resolution reconstruction confirms that the correction amount effectively corrects the average flat field to obtain the real flat field for the current observation conditions. Our method works for chromosphere and photosphere data.

astro-ph.IM↗

arXiv2025

A High-Accuracy Alignment Approach for Solar Images of Different Wavelengths

Image alignment plays a crucial role in solar physics research, primarily involving translation, rotation, and scaling. \G{The different wavelength images of the chromosphere and transition region have structural complexity and differences in similarity, which poses a challenge to their alignment.} Therefore, a novel alignment approach based on dense optical flow (OF) and the RANSAC algorithm is proposed in this paper. \G{It takes the OF vectors of similar regions between images to be used as feature points for matching. Then, it calculates scaling, rotation, and translation.} The study selects three wavelengths for two groups of alignment experiments: the 304 {\AA} of the Atmospheric Imaging Assembly (AIA), the 1216 {\AA} of the Solar Disk Imager (SDI), and the 465 {\AA} of the Solar Upper Transition Region Imager (SUTRI). Two methods are used to evaluate alignment accuracy: Monte Carlo simulation and Uncertainty Analysis Based on the Jacobian Matrix (UABJM). \G{The evaluation results indicate that this approach achieves sub-pixel accuracy in the alignment of AIA 304 {\AA} and SDI 1216 {\AA}, while demonstrating higher accuracy in the alignment of AIA 304 {\AA} and SUTRI 465 {\AA}, which have greater similarity.

astro-ph.IM↗

arXiv2024

Fundamental Parameters of a Binary System Consisting of a Red Dwarf and a Compact Star

TIC 157365951 has been classified as a $\delta$ Scuti type by the International Variable Star Index (VSX). Through the spectra from Large Sky Area Multi-Object Fiber Spectroscopic Telescope (LAMOST) and its light curve, we further discovered that it is a binary system. This binary system comprises a red dwarf star and a compact star. Through the spectral energy distribution (SED) fitting, we determined the mass of the red dwarf star as $M_1 = 0.31 \pm 0.01 M_{\odot}$ and its radius as $R_1 = 0.414 \pm 0.004 R_{\odot}$. By fitting the double-peaked H${\rm \alpha}$ emission, we derived the mass ratio of $q = 1.76 \pm 0.04 $, indicating a compact star mass of $M_2 = 0.54 \pm 0.01 M_{\odot}$. Using Phoebe to model the light curve and radial velocity curve for the detached binary system, we obtained a red dwarf star mass of $M_1 = 0.29 \pm 0.02 M_{\odot}$, a radius of $R_1 = 0.39 \pm 0.04 R_{\odot}$, and a Roche-lobe filling factor of $f = 0.995\pm0.129$, which is close to the $f=1$ expected for a semi-detached system. The Phoebe model gives a compact star mass $M_2 = 0.53 \pm 0.05 M_{\odot}$. Constraining the system to be semidetached gives $M_1 = 0.34 \pm 0.02 M_{\odot}$, $R_1 = 0.41 \pm 0.01 R_{\odot}$, and $M_2 = 0.62 \pm 0.03 M_{\odot}$. The consistency of the models is encouraging. The value of the Roche-lobe filling factor suggests that there might be ongoing mass transfer. The compact star mass is as massive as a typical white dwarf.

astro-ph.SR↗

arXiv2024

The Application of Machine Learning in Tidal Evolution Simulation of Star-Planet Systems

With the release of a large amount of astronomical data, an increasing number of close-in hot Jupiters have been discovered. Calculating their evolutionary curves using star-planet interaction models presents a challenge. To expedite the generation of evolutionary curves for these close-in hot Jupiter systems, we utilized tidal interaction models established on MESA to create 15,745 samples of star-planet systems and 7,500 samples of stars. Additionally, we employed a neural network (Multi-Layer Perceptron - MLP) to predict the evolutionary curves of the systems, including stellar effective temperature, radius, stellar rotation period, and planetary orbital period. The median relative errors of the predicted evolutionary curves were found to be 0.15%, 0.43%, 2.61%, and 0.57%, respectively. Furthermore, the speed at which we generate evolutionary curves exceeds that of model-generated curves by more than four orders of magnitude. We also extracted features of planetary migration states and utilized lightGBM to classify the samples into 6 categories for prediction. We found that by combining three types that undergo long-term double synchronization into one label, the classifier effectively recognized these features. Apart from systems experiencing long-term double synchronization, the median relative errors of the predicted evolutionary curves were all below 4%. Our work provides an efficient method to save significant computational resources and time with minimal loss in accuracy. This research also lays the foundation for analyzing the evolutionary characteristics of systems under different migration states, aiding in the understanding of the underlying physical mechanisms of such systems. Finally, to a large extent, our approach could replace the calculations of theoretical models.

astro-ph.EP↗

arXiv2024

A Method of Rapidly Deriving Late-type Contact Binary Parameters and Its Application in the Catalina Sky Survey

With the continuous development of large optical surveys, a large number of light curves of late-type contact binary systems (CBs) have been released. Deriving parameters for CBs using the the WD program and the PHOEBE program poses a challenge. Therefore, this study developed a method for rapidly deriving light curves based on the Neural Networks (NN) model combined with the Hamiltonian Monte Carlo (HMC) algorithm (NNHMC). The neural network was employed to establish the mapping relationship between the parameters and the pregenerated light curves by the PHOEBE program, and the HMC algorithm was used to obtain the posterior distribution of the parameters. The NNHMC method was applied to a large contact binary sample from the Catalina Sky Survey, and a total of 19,104 late-type contact binary parameters were derived. Among them, 5172 have an inclination greater than 70 deg and a temperature difference less than 400 K. The obtained results were compared with the previous studies for 30 CBs, and there was an essentially consistent goodness-of-fit (R2) distribution between them. The NNHMC method possesses the capability to simultaneously derive parameters for a vast number of targets. Furthermore, it can provide an extremely efficient tool for rapid derivation of parameters in future sky surveys involving large samples of CBs.

astro-ph.IM↗

arXiv2024

Detection of Contact Binary Candidates Observed By TESS Using Autoencoder Neural Network

Contact binary may be the progenitor of a red nova that eventually produces a merger event and have a cut-off period around 0.2 days. Therefore, a large number of contact binaries is needed to search for the progenitor of red novae and to study the characteristics of short-period contact binaries. In this paper, we employ the Phoebe program to generate a large number of light curves based on the fundamental parameters of contact binaries. Using these light curves as samples, an autoencoder model is trained, which can reconstruct the light curves of contact binaries very well. When the error between the output light curve from the model and the input light curve is large, it may be due to other types of variable stars. The goodness of fit (R2) between the output light curve from the model and the input light curve is calculated. Based on the thresholds for global goodness of fit (R2), period, range magnitude, and local goodness of fit (R2), a total of 1322 target candidates were obtained.

astro-ph.SR↗

arXiv2020

A Non-Linear Magnetic Field Calibration Method for Filter-Based Magnetographs by Multilayer Perceptron

For filter-based magnetographs, the linear calibration method under the weak-field assumption is usually adopted; this leads to magnetic saturation effect in the regions with strong magnetic field. This article explores a new method to overcome the above disadvantage using a multilayer perceptron network, which we call MagMLP, based on a back-propagation algorithm with one input layer, five hidden layers, and one output layer. We use the data from the \textit{Spectropolarimeter} (SP) on board \textit{Hinode} to simulate single-wavelength observations for the model training, and take into account the influence of the Doppler velocity field and the filling factor. The training results show that the linear fitting coefficient (LFC) of the transverse field reaches above 0.91, and that of the longitudinal field is above 0.98. The generalization of the models is good because the corresponding LFCs are above 0.9 for the test subsets. Compared with the linear calibration method, the MagMLP is much more effective on dealing with the magnetic saturation effect. Analyzing an active region, the results of the linear calibration present an evident magnetic saturation effect in the umbra regions; the corresponding systematic error reaches values greater than 1000 G in most areas, or even exceeds 2000 G at some pixels. However, the results of MagMLP at these locations are very close to the inversion results, and the systematic errors are basically within 300 G. In addition, we find that there are many "bright spots" and "dark spots" on the inclination angle images from the inversion results of \textit{Hinode}/SP with values of 180 and 0 degrees, respectively, where the inversion is not reliable and does not produce a good result; the MagMLP handles these points well.

astro-ph.IM↗