Search

Next-generation surveys will provide photometric and spectroscopic data of millions to billions of galaxies with unprecedented precision. This offers a unique chance to improve our understanding of the galaxy evolution and the unresolved nature of dark matter (DM). At galaxy scales, the density distribution of DM is strongly affected by the astrophysical feedback processes, which are difficult to fully account for in classical techniques to derive mass models. In this work, we explore the capability of supervised learning algorithms to predict the DM content of galaxies from luminous observational-like parameters, using the public catalog of the TNG100 simulation. In particular, we use Photometric, Structural and Kinematic parameters to predict the total DM mass, DM half-mass radius, DM mass inside one and two stellar half-mass radii. We adopt the coefficient of determination, $R^2$, as a reference metric to evaluate the accuracy of these predictions. We find that the Photometric features alone are able to predict the total DM mass with fair accuracy, while Structural and Photometric features together are more effective to determine the DM inside the stellar half mass radius, and the DM within twice the stellar half mass radius. However, using all observational quantities together (Photometry, Structural and Kinematics) incredibly improves the overall accuracy for all DM quantities. This first test shows that Machine Learning tools are promising approaches to derive predictions of the DM in real galaxies. The next steps will be to improve observational realism of the training sets, by closely select samples which accurately reproduce the typical observed luminous scaling relations. The trained pipelines will be suitable for real galaxy data collected from the next-generation surveys like Rubin/LSST, Euclid, CSST, 4MOST, DESI, to derive, e.g., the properties of their central DM fractions., Comment: 19 pages, 11 figures, 3 tables. Submitted to MNRAS. Comments are welcome

This study investigates the interaction between suppliers and fraudulent customer firms from the perspective of reputation damage and reputation recovery. Specifically, reputation damage from the regulatory penalty for corporate fraud induces the trust crisis and suppliers respond to fraudulent firms by reducing the trade credit supply. To repair a damaged reputation and rebuild the trust, fraudulent firms raise the ratio of prepayment to purchase volume when purchasing from small suppliers and increase the proportion of purchase from large suppliers in the next year. Channel analysis shows that the declining trust is one of potential mechanisms to reduce the trade credit. Furthermore, the negative effect is more pronounced for fraudulent firms with non-related party suppliers, higher supplier concentration, less analyst coverage, and for fraudulent firms located in regions with less-developed financial environment. Additionally, supplier sanctions have a spillover effect on non-fraudulent customer firms in the same industry. The conclusions are robust to a series of checks, including PSM–DiD, firm and year fixed effect, the alternative measure for trade credit financing, the industry-year fixed effect and the consideration of the monetary policy and financial crisis.

Sea freight is one of the most important ways for the transportation and distribution of coal and other bulk cargo. This paper proposes a method for optimizing the scheduling efficiency of the bulk cargo loading process based on deep reinforcement learning. The process includes a large number of states and possible choices that need to be taken into account, which are currently performed by skillful scheduling engineers on site. In terms of modeling, we extracted important information based on actual working data of the terminal to form the state space of the model. The yard information and the demand information of the ship are also considered. The scheduling output of each convey path from the yard to the cabin is the action of the agent. To avoid conflicts of occupying one machine at same time, certain restrictions are placed on whether the action can be executed. Based on Double DQN, an improved deep reinforcement learning method is proposed with a fully connected network structure and selected action sets according to the value of the network and the occupancy status of environment. To make the network converge more quickly, an improved new epsilon-greedy exploration strategy is also proposed, which uses different exploration rates for completely random selection and feasible random selection of actions. After training, an improved scheduling result is obtained when the tasks arrive randomly and the yard state is random. An important contribution of this paper is to integrate the useful features of the working time of the bulk cargo terminal into a state set, divide the scheduling process into discrete actions, and then reduce the scheduling problem into simple inputs and outputs. Another major contribution of this article is the design of a reinforcement learning algorithm for the bulk cargo terminal scheduling problem, and the training efficiency of the proposed algorithm is improved, which provides a practical example for solving bulk cargo terminal scheduling problems using reinforcement learning.

Aiming at the problems of low efficiency, poor spatial location and incomplete attribute information in current transmission line modeling methods, a refined modeling method for transmission lines based on full area feature weighted template matching is proposed. First, collect the point cloud of the scary line, and extract the point cloud data of the tower and power line; Secondly, statistical outlier filter is used to denoise point cloud data; Finally, the GIM model of the transmission line is constructed by using the whole region feature weighted template matching method, which realizes the integrity and lightweight of the geometric, appearance and semantic information of the three-dimensional model of the transmission line, improves the efficiency of the transmission line modeling, and is verified by taking a 500kV transmission line as an example.

This study examines whether penalties issued to Chinese listed companies by securities regulators for violations of corporate law affect the cost of debt, and the moderating role of corporate social responsibility (CSR) fulfillment on this relationship. Our sample consists of firms listed on Shanghai and Shenzhen stock exchanges from 2011 to 2017 and the data are collected from the announcements of China Securities Regulatory Commission. The findings are as follows: (1) punishment announcements by regulatory authorities increase the cost of debt; and (2) the effect of punishment announcements on the cost of debt is partially offset by prior CSR performance. These findings are shown to be robust. The reputation insurance effect of CSR is more pronounced in state-owned enterprises and in an institutional environment with low marketization, a weak legal environment, and low information transparency. The findings support the reputation insurance hypothesis of CSR and employ the cost of debt as a governance mechanism.

The objective of this study was to determine and compare the effect of five different pretreatments including sulfite treatment (ST), osmotic dehydration (OD), steam blanching (SB), steam blanching plus osmotic dehydration (SB+OD), and ultrasound treatment (UT), on the microwave vacuum drying (MVD) kinetics, the physicochemical properties, and sensory quality of the apple slices. The results showed that the pretreatments prior to MVD could reduce the drying time of apple slices by 25-45% as compared to the non-pretreated apple slices, and the drying time in the SB+OD apple slices sample was shortest. Whether pretreated or not, MVD process was controlled by diffusion and characterized by a two-stage falling-rate drying. As calculated according to the Fick's law of diffusion, the moisture diffusivities ( D eff ) ranged from 1.64×10 -8 to 3.46×10 - 8 m 2 /s. Different pretreatment methods had a significant influence on the physicochemical properties and sensory qualities of the dried products (pSB>UT>OD>ST>the control. ntrol.

In many electronic and information engineering related disciplines, various complex machine learning and reinforcement learning optimization algorithms appear nowadays so as to handle huge amount of data related. Traditional optimization methods, which have to take the whole data space into consideration, cannot always perform well, and cannot always be completed in a short period of time as the data amount to be handled is huge. The object of the so-called optimization problem is to find the optimal solutions or quasi-optimal solutions in a complex and huge data searching space, and finding the efficient optimization algorithm which can handle data with characteristics of complexity, nonlinearity, and modeling difficulties for practical engineering, can be meaningful for electronic and information industries. Huber loss function is a typical traditional optimization method, with the error expressed via smooth mean absolute error (MAE) which has the advantages of MAE and mean square error (MSE) and uses MAE to enhance the robustness of MSE for those abnormal data points. However, Huber loss function employs gradient descent (GD) method to find the minimum value, which can easily be trapped into the local optimal solution. In this paper, a hybrid optimization algorithm, Huber loss function based on cockroach swarm algorithm with t-distribution has been proposed, it has better global convergence, and via numerical experiment, its efficiency in calculation has also been proved.

F1exible pressure sensors are important in applications of flexible electronics. Although extensive efforts have been made to improve the performance of pressure sensors through materials fabrication or circuitry design, the effect of loading mode on the sensor response has rarely been discussed before. In this study, an appropriate constitutive relationship of the composite material that is used as the dielectric layer in capacitive pressure sensors was characterized via tensile tests, thereby ensuring the stresses in the film sensor under a pressure can be simulated accurately by the finite element analysis. The stresses under the pressure head, as a major factor to affect the capacitance variation, were further fitted by a polynomial equation in terms of the variation of the capacitance of the film sensor with a designed experiment for both measuring the capacitance and force. The equation supports an empirical model that reveals the capacitance of the film sensor can be estimated directly when the stress in the sensor is obtained. The model helps to develop a new pressure sensor applied in flexible devices.

This study assessed the concentrations of specific volatile organic compounds (VOCs) inside vehicle cabins under different practical vehicle driving conditions in China. The mean concentrations of the VOCs, including benzene, toluene, xylene, ethylbenzene, styrene, formaldehyde, acetaldehyde, acetone, and acrolein, were 16.73 μg/m 3 , 66.02 μg/m 3 , 14.20 μg/m 3 , 6.78 μg/m 3 , 28.09 μg/m 3 , 16.43 μg/m 3 , 12.47 μg/m 3 , and 20.65 μg/m 3 (the sum of acetone and acrolein), respectively. All the specified VOCs inside vehicle cabins were not exceeded the limits of the national standard. The in-cabin VOCs concentrations were investigated for 16 private vehicles under three ventilation conditions: (i) fan off and recirculation (RC) off, (ii) fan on and RC off, and (iii) fan on and RC on. The VOCs concentrations increased 50.46% (mean of the measured VOCs) when the ventilation condition changed from (ii) to (i), and increased 51.38% (mean of the measured VOCs) when ventilation condition changed from (ii) to (iii). Two vehicle models (vehicle model A and vehicle model B) were tested in the study to investigate the influence on in-cabin VOCs concentrations of two typical interior trims (leather, fabric). The VOCs concentrations inside B vehicles (leather interiors) were averagely 1.42 times larger than the concentrations in A vehicles (fabric interiors). For new vehicles, the concentrations of benzene, toluene, xylene, ethylbenzene, formaldehyde, acetaldehyde, acetone and acrolein were larger than the concentrations inside old vehicles by 12.89%, 103.54%, 123.14%, 104.20%, 6.26%, 6.31%, and 10.67%, respectively. The VOCs concentrations significantly increased as the raise of ambient temperature. Toluene, styrene, ethylbenzene, and xylene were the most sensitive VOCs to temperature, which increased 513.6%, 544.8%, 767.0%, and 597.7% as the temperature increased from 11 °C to 25 °C.