Sort by
A robust numerical method based on a deep-learning operator for solving the 2D acoustic wave equation

The numerical solution of a wave equation plays a crucial role in computational geophysics problems, which forms the foundation of inverse problems and directly impacts the high-precision imaging results of earth models. However, common numerical methods often lead to significant computational and storage requirements. Due to the heavy reliance on forward-modeling methods in inversion techniques, particularly full-waveform inversion, enhancing the computational efficiency and reducing the storage demands of traditional numerical methods have become key issues in computational geophysics. We develop the deep Lax-Wendroff correction (DeepLWC) method, a deep-learning-based numerical format for solving 2D hyperbolic wave equations. DeepLWC combines the advantages of traditional numerical schemes with a deep neural network. We provide a detailed comparison of this method with the representative traditional Lax-Wendroff correction method. Our numerical results indicate that the DeepLWC significantly improves the calculation speed (by more than 10 times) and reduces the space needed for storage by more than 10,000 times compared with traditional numerical methods. In contrast to the more popular physics-informed neural network method, DeepLWC maximizes the advantages of traditional mathematical methods in solving partial differential equations and uses a new sampling approach, leading to improved accuracy and faster computations. It is particularly worth pointing out that DeepLWC introduces a novel research paradigm for numerical equation solving, which can be combined with various traditional numerical methods, enabling acceleration and reduction in the storage requirements of conventional approaches.

Relevant
Discrimination of New and Aged Seeds Based on On-Line Near-Infrared Spectroscopy Technology Combined with Machine Learning

The harvest year of maize seeds has a significant impact on seed vitality and maize yield. Therefore, it is vital to identify new seeds. In this study, an on-line near-infrared (NIR) spectra collection device (899–1715 nm) was designed and employed for distinguishing maize seeds harvested in different years. Compared with least squares support vector machine (LS-SVM), k-nearest neighbor (KNN), and extreme learning machine (ELM), the partial least squares discriminant analysis (PLS-DA) model has the optimal recognition performance for maize seed harvest years. Six different preprocessing methods, including Savitzky–Golay smoothing (SGS), standard normal variate transformation (SNV), multiplicative scatter correction (MSC), Savitzky–Golay 1 derivative (SG-D1), Savitzky–Golay 2 derivative (SG-D2), and normalization (Norm), were used to improve the quality of the spectra. The Monte Carlo cross-validation uninformative variable elimination (MC-UVE), competitive adaptive reweighted sampling (CARS), bootstrapping soft shrinkage (BOSS), successive projections algorithm (SPA), and their combinations were used to obtain effective wavelengths and decrease spectral dimensionality. The MC-UVE-BOSS-PLS-DA model achieved the classification with an accuracy of 88.75% using 93 features based on Norm preprocessed spectral data. This study showed that the self-designed NIR collection system could be used to identify the harvested years of maize seed.

Open Access
Relevant
MeshLink: A surface structured mesh generation framework to facilitate automated data linkage

With the rapid progress in engineering simulation, there is an increasing industrial need for accurate mesh generation methods. Furthermore, the development of data-driven methods require an innovative mesh generation framework that can integrate deep learning models to facilitate automatic data linkage. This paper develops a surface structured mesh generation framework named MeshLink. This framework extends the support for mesh data and associated algorithms through the Mesh-based Feature Information Framework (MFIF). First, in order to map the model to the parametric domain, we discretize the input model using triangular meshes. To handle complex geometric shapes, we develop a structured mesh generation technique based on conformal mapping. Then, we generate the surface structured mesh of the model based on inverse mapping algorithm. The MeshLink framework overcomes the limitations of traditional mesh generation workflows by integrating deep learning models. We adopted a structured mesh evaluation model based on graph neural network, which enhance the efficiency of the framework. Finally, based on the mesh quality evaluation results, we use the corresponding mesh optimization algorithm to generate high-quality surface structure meshes. The MeshLink framework not only provides a tool that supports high-quality surface structured mesh generation, but also facilitates the storage, linking and retrieval of mesh data sources.

Relevant
N-Doping CQDs as an Efficient Fluorescence Probe Based on Dynamic Quenching for Determination of Copper Ions and Alcohol Sensing in Baijiu.

To address an accurate detection of heavy metal ions in Baijiu production, a nitrogen-doping carbon quantum dots (N-CQDs) was prepared by hydrothermal method from citric acid and urea. The as-prepared N-CQDs had an average particle size of 2.74 nm, and a large number of functional groups (amino, carbonyl group, etc.) attached on its surface, which obtained a 9.6% of quantum yield (QY) with relatively high and stable fluorescence performance. As a fluorescent sensor, the fluorescence of N-CQDs at 380 nm excitation wavelength could be quenched quantitatively by adding Cu2+, due to the dynamic quenching of electron transfer caused by the binding of amine groups and Cu2+, which showed excellent sensitivity and selectivity to Cu2+ in the range of 0.5-5 μM with a detection limit (LOD) of 0.032 μM. In addition, the N-CQDs as well as could be applied to quantitative determine alcohol content in the range of 10-80 V/V% depending on the fluorescence enhancement. Upon the experiment, the fluorescent mechanism was studied by Molecular dynamics (MD) simulations, which demonstrated that solvent effect played an influential role on sensing alcohol content in Baijiu. Overall, the work provided a theoretically guide for the design of fluorescence sensors to monitor heavy metal ion in liquid drinks and sense alcohol content.

Relevant
Comparison of anti-allergic activities of different types of lotus seed resistant starch in OVA-induced mouse model

Currently, evidence from observational studies suggests dietary fiber intake may be associated with decreased risk of food allergy. As a type of dietary fiber, resistant starch was also widely reported to possess anti-allergic properties. However, there is a relative paucity of studies assessing the influence of resistant starch types on their anti-allergic activity and its possible underlying mechanisms. In the current study, the anti-allergic effects of RS3-type (retrograded starch), RS4-type (chemically modified starch, cross-bonded), and RS5-type (starch-palmitic acid complex) of lotus seed resistant starch were evaluated in the OVA (100 mg/kg)-induced food allergic mice model. The results showed that oral administration of RS3 or RS4 lotus seed resistant starch (0.3 g/100 g b.w.) for 25 days significantly improved adverse symptoms of food allergy such as weight loss, increases in allergy symptom score and diarrhea rate; with significant reduction of serum specific antibody IgE, TNF-α, IL-4 levels and improved Th1/Th2 balance being observed. The mechanism may involve the regulation of lotus seed resistant starch on intestinal flora and the metabolites short-chain fatty acids and bile acids. Taken together, the findings may enhance understanding towards ameliorative effects of resistant starch on food allergy, and offer valuable insights for the exploration of novel anti-allergic bioactive compounds.

Relevant
Metagenomic and flavoromic profiling reveals the correlation between the microorganisms and volatile flavor compounds in Monascus-fermented cheese

The Monascus-fermented cheese (MC) is a unique cheese product that undergoes multi-strain fermentation, imparting it with distinct flavor qualities. To clarify the role of microorganisms in the formation of flavor in MC, this study employed SPME (arrow)-GC–MS, GC-O integrated with PLS-DA to investigate variations in cheese flavors represented by volatile flavor compounds across 90-day ripening periods. Metagenomic datasets were utilized to identify taxonomic and functional changes in the microorganisms. The results showed a total of 26 characteristic flavor compounds in MC at different ripening periods (VIP>1, p < 0.05), including butanoic acid, hexanoic acid, butanoic acid ethyl ester, hexanoic acid butyl ester, 2-heptanone and 2-octanone. According to NR database annotation, the genera Monascus, Lactococcus, Aspergillus, Lactiplantibacillus, Staphylococcus, Flavobacterium, Bacillus, Clostridium, Meyerozyma, and Enterobacter were closely associated with flavor formation in MC. Ester compounds were linked to Monascus, Meyerozyma, Staphylococcus, Lactiplantibacillus, and Bacillus. Acid compounds were linked to Lactococcus, Lactobacillus, Staphylococcus, and Bacillus. The production of methyl ketones was closely related to the genera Monascus, Staphylococcus, Lactiplantibacillus, Lactococcus, Bacillus, and Flavobacterium. This study offers insights into the microorganisms of MC and its contribution to flavor development, thereby enriching our understanding of this fascinating dairy product.

Relevant