Direct Optimization Research Articles

Orbital-Optimized Versus Time-Dependent Density Functional Calculations of Intramolecular Charge Transfer Excited States.

The performance of time-independent, orbital-optimized calculations of excited states is assessed with respect to charge transfer excitations in organic molecules in comparison to the linear-response time-dependent density functional theory (TD-DFT) approach. A direct optimization method to converge on saddle points of the electronic energy surface is used to carry out calculations with the local density approximation (LDA) and the generalized gradient approximation (GGA) functionals PBE and BLYP for a set of 27 excitations in 15 molecules. The time-independent approach is fully variational and provides a relaxed excited state electron density from which the extent of charge transfer is quantified. The TD-DFT calculations are generally found to provide larger charge transfer distances compared to the orbital-optimized calculations, even when including orbital relaxation effects with the Z-vector method. While the error on the excitation energy relative to theoretical best estimates is found to increase with the extent of charge transfer up to ca. -2 eV for TD-DFT, no correlation is observed for the orbital-optimized approach. The orbital-optimized calculations with the LDA and the GGA functionals provide a mean absolute error of ∼0.7 eV, outperforming TD-DFT with both local and global hybrid functionals for excitations with a long-range charge transfer character. Orbital-optimized calculations with the global hybrid functional B3LYP and the range-separated hybrid functional CAM-B3LYP on a selection of states with short- and long-range charge transfer indicate that inclusion of exact exchange has a small effect on the charge transfer distance, while it significantly improves the excitation energy, with the best-performing functional CAM-B3LYP providing an absolute error typically around 0.15 eV.

Photovoltaic and Solar-Thermal Use Case Application Comparison with Witness Simulation and DCF Analysis

In this work, we investigate a system of solar and thermal heat collectors in combination with local and electric net storage. The system is implemented in a Witness Horizon simulation model, allowing for the investigation and scenario analysis of different use cases. In a practical industrial application example use-case, using simulation and Discounted Cash Flow (DCF) analysis, the economic feasibility of the best system variant could be calculated. The developed model includes the local consumption and production profile and it was evaluated by a DCF model after the dynamic energy modelisation with the generic industrial production system simulation software WITNESS Horizon. This model can be in future further extended to generic and more complex production systems, simultaneously integrating the DCF cost evaluation analysis and following direct economic optimisation as well as integrating it with the production portfolio theory approach by integrating diversified virtual production market aspects and market goal-oriented optimisation.

Wind-Induced Aerodynamic Effects on Set-Back Tall Buildings Using LES

Set-back modification represents a dependable method to reducing the wind effects on high-rise buildings, but the flow mechanism and aerodynamic responses of set-back tall buildings have not been explored systematically. Using the square building as the benchmark model, six set-back building models with varying steps and sizes were chosen for numerical simulation in this study. The results of mean wind pressure and local wind force distribution, power spectral densities of base moment, flow field, and structural response of the proposed building models are compared and analysed using the Large Eddy Simulation (LES) technique. The results demonstrate that the aerodynamic performance of set-back tall buildings dramatically improved, with the actual optimization in the cross-wind direction being more significant. Generally, a single set-back measure can enhance the aerodynamic performance of high-rise buildings more effectively than a double set-back measure. The outcome of this study provides an aerodynamic design guide for set-back tall buildings.

Прогнозування РНК, siRNA з подальшим визначенням імунологічної відповіді запропонованої вакцини (Vac534) проти COVID-19

Despite the decline in Coronavirus infections, care must be taken to avoid new mutations that allow the virus to escape vaccination and treatment. Viral RNA is responsible for virus replication and assembly in the host cell. Aim. The goal of this work was to predict the RNA secondary structure of the vaccination we developed before (Vac534). In addition, the degree of RNA overlaps while translating into ribosomes, refolding, and evaluating the protein immune response are all considered. Methods. Different immunobioinformatics tools and servers were utilized for analysis. RNA folding was executed through the application of RNAstructure 6.4 and RNAfold software. The unique sequence context led to the emergence of potential RNA structures at the site where ribosome binding occurs. A new method, C-IMMSIM, which relies on immune cell epitope prediction, was employed to gain fresh insights into comprehending the immune system. Results. А high probability of ≥ 99% is shown between nucleotides with the stability of loops and motifs of the RNA 2D structure. The predicted siRNA sequences, which were located in three places, were used to calculate total energy, self-folding, and a free end with a high accessibility score. RNA translation into the ribosome is required to determine the optimum direction of translation. The short docking fragment of 24 nucleotides of RNA (Vac534) generated six robust binds with MPRO at high energy. Conclusions. The immunological evaluation of the vaccination is critical for stimulating immune cells and detecting interleukins and cytokine production. This study is viewed as a step toward developing Coronavirus treatments. The number of loops and motifs is increasing, with a probability of above 99% for most sequences. The predicted siRNA sequences, which were located in three places, were used to calculate total energy, self-folding, and a free end with a high accessibility score.

Design and flexural behavior of steel fiber-reinforced concrete beams with regular oriented fibers and GFRP bars

In this study, steel fiber-reinforced concrete (SFRC) beam structures with regular oriented steel fibers are designed and prepared by using an assembled solenoid device. Fiber orientation is not arranged in a single direction but is similar to the direction of tensile stress of the beam in bending load conditions. With application of inductive test method and image processing technology, fiber orientation coefficient of SFRC was evaluated and increased from 0.5 to 0.7–0.8 after fiber alignment. To describe and analyze the flexural behavior of SFRC beams with regular oriented fibers and glass fiber-reinforced polymer (GFRP) rebars, 14 concrete beams with different parameters were fabricated and subjected to a four-point bending test. Specifically, the parameters included fiber content, fiber orientation, GFRP reinforcement ratio and different layer sections. The results showed that optimizing fiber orientations to the tension zone of the section led to a higher flexural strength and can well restrain the development of crack widths with comparison to randomly distributed fibers. The peak load and energy absorption increased by 7 % and 12.4 % after fiber alignment with a fiber volume fraction of 1.5 %. The combination of randomly and regularly distributed steel fibers in the form of two-layer composite members can be an efficient solution to improve the flexural performance for purpose of the optimum distribution and direction of steel fibers, and reduce the engineering cost.

Enhancing acoustic levitation capacity through array geometry optimization

The contactless micromanipulation capability and microgravity environment provided by the acoustic levitation technology have excellent application prospects in many fields, such as droplet dynamics, solidification state physics, life sciences, etc. However, the applications are severely constrained by the limitation of the transducer power. In this paper, we explore an array geometry optimization framework to improve the levitation capability. First, the acoustic pressure attenuation and directional characteristics of commercial transducers are analyzed to uncover the mechanism of the array geometry affecting the acoustic radiation force (ARF). Second, based on the Gor'kov potential theory, the array geometry form of N-elements is derived, and the arrangement in the spherical ring (ASR) method is proposed to obtain the element coordinates. Then, spherical radius and array height are adjusted to obtain the optimal array geometry. On this basis, the directional angle of each transducer is further optimized. Finally, simulation and physical verification are carried out. The simulation results indicate that the ARF increases and then decreases with the spherical radius and array height, which are the existing maximum values. Directional angle optimization (DAO) can improve the ARF, particularly when the array height exceeds the sphere diameter. These findings are consistent with the analytical results. Notably, we built a prototype physical system, and the physical experiments showed that the optimized array improved the axial ARF by 201.2% and the radial ARF by 115.9% over the conventional array. The works presented in this paper offer a valuable design approach for acoustic levitation arrays, thereby contributing to the advancement and application of ultrasonic array levitation technology.

Enhanced Lightweight YOLOX for Small Object Wildfire Detection in UAV Imagery

Target detection technology based on unmanned aerial vehicle (UAV)-derived aerial imagery has been widely applied in the field of forest fire patrol and rescue. However, due to the specificity of UAV platforms, there are still significant issues to be resolved such as severe omission, low detection accuracy, and poor early warning effectiveness. In light of these issues, this paper proposes an improved YOLOX network for the rapid detection of forest fires in images captured by UAVs. Firstly, to enhance the network’s feature-extraction capability in complex fire environments, a multi-level-feature-extraction structure, CSP-ML, is designed to improve the algorithm’s detection accuracy for small-target fire areas. Additionally, a CBAM attention mechanism is embedded in the neck network to reduce interference caused by background noise and irrelevant information. Secondly, an adaptive-feature-extraction module is introduced in the YOLOX network’s feature fusion part to prevent the loss of important feature information during the fusion process, thus enhancing the network’s feature-learning capability. Lastly, the CIoU loss function is used to replace the original loss function, to address issues such as excessive optimization of negative samples and poor gradient-descent direction, thereby strengthening the network’s effective recognition of positive samples. Experimental results show that the improved YOLOX network has better detection performance, with mAP@50 and mAP@50_95 increasing by 6.4% and 2.17%, respectively, compared to the traditional YOLOX network. In multi-target flame and small-target flame scenarios, the improved YOLO model achieved a mAP of 96.3%, outperforming deep learning algorithms such as FasterRCNN, SSD, and YOLOv5 by 33.5%, 7.7%, and 7%, respectively. It has a lower omission rate and higher detection accuracy, and it is capable of handling small-target detection tasks in complex fire environments. This can provide support for UAV patrol and rescue applications from a high-altitude perspective.

EmoStyle: Emotion-Aware Semantic Image Manipulation with Audio Guidance

With the flourishing development of generative models, image manipulation is receiving increasing attention. Rather than text modality, several elegant designs have delved into leveraging audio to manipulate images. However, existing methodologies mainly focus on image generation conditional on semantic alignment, ignoring the vivid affective information depicted in the audio. We propose an Emotion-aware StyleGAN Manipulator (EmoStyle), a framework where affective information from audio can be explicitly extracted and further utilized during image manipulation. Specifically, we first leverage the multi-modality model ImageBind for initial cross-modal retrieval between images and music, and select the music-related image for further manipulation. Simultaneously, by extracting sentiment polarity from the lyrics of the audio, we generate an emotionally rich auxiliary music branch to accentuate the affective information. We then leverage pre-trained encoders to encode audio and the audio-related image into the same embedding space. With the aligned embeddings, we manipulate the image via a direct latent optimization method. We conduct objective and subjective evaluations on the generated images, and our results show that our framework is capable of generating images with specified human emotions conveyed in the audio.

Smooth Information Criterion for Regularized Estimation of Item Response Models

Item response theory (IRT) models are frequently used to analyze multivariate categorical data from questionnaires or cognitive test data. In order to reduce the model complexity in item response models, regularized estimation is now widely applied, adding a nondifferentiable penalty function like the LASSO or the SCAD penalty to the log-likelihood function in the optimization function. In most applications, regularized estimation repeatedly estimates the IRT model on a grid of regularization parameters λ. The final model is selected for the parameter that minimizes the Akaike or Bayesian information criterion (AIC or BIC). In recent work, it has been proposed to directly minimize a smooth approximation of the AIC or the BIC for regularized estimation. This approach circumvents the repeated estimation of the IRT model. To this end, the computation time is substantially reduced. The adequacy of the new approach is demonstrated by three simulation studies focusing on regularized estimation for IRT models with differential item functioning, multidimensional IRT models with cross-loadings, and the mixed Rasch/two-parameter logistic IRT model. It was found from the simulation studies that the computationally less demanding direct optimization based on the smooth variants of AIC and BIC had comparable or improved performance compared to the ordinarily employed repeated regularized estimation based on AIC or BIC.

Directional optimization of elevator scheduling algorithms in complex traffic patterns

Elevator systems in buildings face challenges due to unpredictable passenger flow, which can make scheduling elevators complicated to optimize their operation. Most of the existing algorithms are developed based on pattern recognition and may not be effective in scenarios where patterns are difficult to classify, especially when elevator operations involve uncertain human behaviors. To address this issue, this paper proposes a time-dependent optimization model that considers long-term and multi-floor peaks, with a focus on peak floors. The proposed model accounts for dynamic scheduling patterns of elevators and represents passenger flow direction as a relation matrix. The proposed direction optimization method contains several functions to guide iteration direction and improve the efficiency of an iteration process based on classical algorithms. This method also ensures the stability of Markov chains by adjusting an iteration process. The feasibility of the proposed method is supported by the relevant theory, and experimental results show that the direction-optimized algorithms outperform classical algorithms, resulting in the superb operating efficiency of elevators at a lower cost. This paper contributes to the development of efficient algorithms for scheduling elevators in complex traffic patterns, which can improve performance of elevator group systems in buildings. The proposed method is not only limited to elevators but can also be extended to other transportation systems with flow requirements.

High-dimensional aerodynamic shape optimization framework using geometric domain decomposition and data-driven support strategy for wing design

Traditional aerodynamic shape optimization methods usually suffer from difficulties in optimization and modeling for high-dimensional aerodynamic optimization problems. To overcome these difficulties, a new high-dimensional aerodynamic shape optimization framework is proposed from the perspectives of geometric domain decomposition for dimensionality reduction and a data-driven model support strategy for changing the role of surrogate model and assisting the optimizer. Geometric domain decomposition uses proper orthogonal decomposition to transform the high-dimensional Class-Shape Function Transformation parametric method into low-dimensional geometric modes to reduce design dimensions for parameterizing the geometric shape, which reduces optimization and modeling difficulties. Also, a data-driven support strategy is proposed to improve the effectiveness and efficiency of the optimizer by learning from historical aerodynamic data; this strategy modifies the role of surrogate model in the optimization framework to assist the optimizer; the surrogate model, serving as an auxiliary role, can reduce the dependence on model accuracy, which overcomes the difficulty of high-dimensional modeling to some extent. To verify the effectiveness and efficiency of the developed framework, transonic drag reduction of wings is conducted as an example, and the optimization results are compared with those from direct optimization and surrogate-based optimization frameworks. These results show that the developed framework converges faster and offers better designs compared to other frameworks for high-dimensional aerodynamic shape optimization.

Determination of optimum printing direction of the polycarbonate parts produced by additive manufacturing

In this study, the effect of printing direction on the tensile properties of polycarbonate (PC) parts produced using a 3D printer was investigated. A total of 9 samples were produced with printing directions of 0°, 45° and 90° with infill density of 20%. Tensile tests were carried out to determine the maximum tensile strength (UTS) of the samples. As a result of the research, it was determined that the samples produced in the same direction as the tensile direction (0° printing direction) had the highest strength values; on the other hand, samples produced in the 90° printing direction have the lowest strength values. While the ultimate tensile strength values for samples produced in 0° printing direction were measured respectively as 22.4 MPa, 21.6 MPa and 20.7 MPa; as 7.1 MPa, 9.1 MPa and 9.2 MPa in 45° printing direction; it was determined as 7.2 MPa, 5.4 MPa and 5.8 MPa in the 90° printing direction. In addition, UTS/mass ratios were calculated by measuring the mass of the samples after the tensile test. According to the results obtained, samples in the 0° printing direction have the highest UTS/mass ratio compared to other printing directions, making them the most durable option for applications requiring lightweight structures. Tensile samples produced in FDM were analyzed with the finite element method (FEM) and analysis results were found to be equivalent to physical tests. This study contributes to examination of the influence of 3D printing direction, especially exposed to tensile, on material properties, and provides important information for the production of lightweight and high strength PC components in industries such as medicine, aerospace and automotive.

Design of a Broadband Integrated Mode Converter Based on Non-Uniform Binary Long-Period Grating by Direct Multiparameter Optimization

Design of a Broadband Integrated Mode Converter Based on Non-Uniform Binary Long-Period Grating by Direct Multiparameter Optimization

Research on Large Language Model for Coal Mine Equipment Maintenance Based on Multi-Source Text

The efficient management and utilization of coal mine equipment maintenance knowledge is an indispensable foundation for advancing the establishment of intelligent mines. This knowledge has problems such as scattered, low sharing, and insufficient management, which restricts the development of coal mine intelligence. For the above-mentioned problems, a large language model for the maintenance of coal mine equipment based on multi-source text (XCoalChat) was proposed to better manage and utilize the existing massive knowledge of coal mine equipment maintenance. The dataset of coal mine equipment maintenance based on ReliableCEMK-Self-Instruction was constructed to obtain a wide and diverse amount of knowledge through sample generation. Aiming at the illusory problem of the large language model, a knowledge graph enhancement method based on the “Coal Mine Equipment Maintenance System—Full Life Cycle—Specification” was proposed to improve the knowledge density. A triple-LoRA fine-tuning mechanism and DPO direct preference optimization method were introduced into the top of the baseline model, which guarantees that XCoalChat can handle multiple Q&A and maintenance decision analysis tasks with limited computing power. Compared with ChatGLM, Bloom, and LLama, the comprehensive assessment of XCoalChat was performed by experiments including coal mine dialog consulting, coal mine professional consulting, and maintenance decision analysis. The results showed that XCoalChat achieved the best response accuracy in professional consulting and maintenance decision analysis; XCoalChat also took the least reasoning time on average. XCoalChat outperformed other mainstream large language models, which verify that XCoalChat is an effective large language model in the field of coal mine equipment maintenance.

活动转板式马铃薯分级装置优化设计分析

The study focused on enhancing a movable rotating plate type grading device for potatoes, prioritizing safety and weight reduction. By optimizing the body frame and movable rotating plate using ANSYS Workbench 2022, a static analysis was conducted to confirm compliance with design requirements for strength and deformation. The optimized frame experienced a remarkable 32.3% weight reduction. Additionally, the direct optimization module of ANSYS Workbench 2022 was employed to lighten the movable rotating plate, resulting in a 22.86% reduction in the total mass of the device. This research serves as an invaluable reference for the structural design and optimization analysis of potato grading devices.

Firm failure prediction using genetic programming generated features

Many studies on predicting firm failure have focused on finding new features that improve the accuracy of the models. In this paper, genetic programming (GP) is used for this purpose. The main problem in GP is to specify a function that evaluates the fitness of the feature. Direct optimization of a machine learning (ML) model that uses a generated feature in most cases leads to high computational costs since evolving a population of N programs over G generations while evaluating each model using K-fold cross-validation requires N*G*K model learning cycles. Thus, many researchers use scores that measure the relationship of the generated features to the class label. However, our empirical analysis shows that most such scores correlate poorly with ML model performance. The novelty of our work is that we introduce several ways of combining different scores into a single measure of expected model performance. Experimental results on data from Hungarian firms (7167 observations, class imbalance 9.37) using five ML models (Logistic Regression, Random Forest, Gradient Boosting, Histogram Boosting, and AdaBoost) prove that the proposed way of setting the fitness function increases the ROC AUC of the listed models by 6.6%, 5.2%, 6.8%, 5.5% and 5.2% respectively. Moreover, by applying the found formula to the data from Czech firms (3872 observations, class imbalance of 74.92), which were not used for the feature search, we obtained increases in ROC AUC by 13.1%, 11.8%, 14.9%, 9.9%, and 8.2%, respectively. This indicates that the proposed method allows to find universal features, which opens the way to build effective models in case of insufficient data (small number of observations, extreme imbalance, etc.).

Identification of Wind Load Exerted on the Jacket Wind Turbines from Optimally Placed Strain Gauges Using C-Optimal Design and Mathematical Model Reduction

Wind turbine towers experience complex dynamic loads during actual operation, and these loads are difficult to accurately predict in advance, which may lead to inaccurate structural fatigue and strength assessment during the structural design phase, thereby posing safety risks to the wind turbine tower. However, online monitoring of wind loads has become possible with the development of load identification technology. Therefore, an identification method for wind load exerted on wind turbine towers was developed in this study to estimate the wind loads using structural strain, which can be used for online monitoring of wind loads. The wind loads exerted on the wind turbine tower were simplified into six equivalent concentrated forces on the topside of the tower, and the initial mathematical model for wind load identification was established based on dynamic load identification theory in the frequency domain, in which many candidate sensor locations and directions were considered. Then, the initial mathematical model was expressed as a linear system of equations. A numerical example was used to verify the accuracy and stability of the initial mathematical model for the wind load identification, and the identification results indicate that the initial mathematical model combined with the Moore–Penrose inverse algorithm can provide stable and accurate reconstruction results. However, the initial mathematical model uses too many sensors, which is not conducive to engineering applications. Therefore, D-optimal and C-optimal design methods were used to reduce the dimension of the initial mathematical model and determine the location and direction of strain gauges. The C-optimal design method adopts a direct optimisation search strategy, while the D-optimal design method adopts an indirect optimisation search strategy. Then, four numerical examples of wind load identification show that dimensionality reduction of the mathematical model leads to high accuracy, in which the C-optimal design algorithm provides more robust identification results. Moreover, the fatigue damage calculated based on the load identification wind loads closely approximates that derived from finite element simulation wind load, with a relative error within 6%. Therefore, the load identification method developed in this study offers a pragmatic solution for the accurate acquisition of the actual wind load of a wind turbine tower.

TETRIS: Towards Exploring the Robustness of Interactive Segmentation

Interactive segmentation methods rely on user inputs to iteratively update the selection mask. A click specifying the object of interest is arguably the most simple and intuitive interaction type, and thereby the most common choice for interactive segmentation. However, user clicking patterns in the interactive segmentation context remain unexplored. Accordingly, interactive segmentation evaluation strategies rely more on intuition and common sense rather than empirical studies (e.g., assuming that users tend to click in the center of the area with the largest error). In this work, we conduct a real-user study to investigate real user clicking patterns. This study reveals that the intuitive assumption made in the common evaluation strategy may not hold. As a result, interactive segmentation models may show high scores in the standard benchmarks, but it does not imply that they would perform well in a real world scenario. To assess the applicability of interactive segmentation methods, we propose a novel evaluation strategy providing a more comprehensive analysis of a model's performance. To this end, we propose a methodology for finding extreme user inputs by a direct optimization in a white-box adversarial attack on the interactive segmentation model. Based on the performance with such adversarial user inputs, we assess the robustness of interactive segmentation models w.r.t click positions. Besides, we introduce a novel benchmark for measuring the robustness of interactive segmentation, and report the results of an extensive evaluation of dozens of models.

Detecting and Preventing Hallucinations in Large Vision Language Models

Instruction tuned Large Vision Language Models (LVLMs) have significantly advanced in generalizing across a diverse set of multi-modal tasks, especially for Visual Question Answering (VQA). However, generating detailed responses that are visually grounded is still a challenging task for these models. We find that even the current state-of-the-art LVLMs (InstructBLIP) still contain a staggering 30 percent of the hallucinatory text in the form of non-existent objects, unfaithful descriptions, and inaccurate relationships. To address this, we introduce M-HalDetect, a Multimodal Hallucination Detection Dataset that can be used to train and benchmark models for hallucination detection and prevention. M-HalDetect consists of 16k fine-grained annotations on VQA examples, making it the first comprehensive multi-modal hallucination detection dataset for detailed image descriptions. Unlike previous work that only consider object hallucination, we additionally annotate both entity descriptions and relationships that are unfaithful. To demonstrate the potential of this dataset for hallucination prevention, we optimize InstructBLIP through our novel Fine-grained Direct Preference Optimization (FDPO). We also train fine-grained multi-modal reward models from InstructBLIP and evaluate their effectiveness with best-of-n rejection sampling (RS). We perform human evaluation on both FDPO and rejection sampling, and find that they reduce hallucination rates in InstructBLIP by 41% and 55% respectively. We also find that our reward model generalizes to other multi-modal models, reducing hallucinations in LLaVA and mPLUG-OWL by 15% and 57% respectively, and has strong correlation with human evaluated accuracy scores. The dataset is available at https://github.com/hendryx-scale/mhal-detect.

ESRL: Efficient Sampling-Based Reinforcement Learning for Sequence Generation

Applying Reinforcement Learning (RL) to sequence generation models enables the direct optimization of long-term rewards (e.g., BLEU and human feedback), but typically requires large-scale sampling over a space of action sequences. This is a computational challenge as presented by the practice of sequence generation problems, such as machine translation, where we often deal with a large action space (e.g., a vocabulary) and a long action sequence (e.g., a translation). In this work, we introduce two-stage sampling and dynamic sampling approaches to improve the sampling efficiency during training sequence generation models via RL. We experiment with our approaches on the traditional sequence generation tasks, including machine translation and abstractive summarization. Furthermore, we evaluate our approaches in RL from human feedback (RLHF) through training a large language model using the reward model. Experimental results show that the efficient sampling-based RL, referred to as ESRL, can outperform all baselines in terms of both training efficiency and memory consumption. Notably, ESRL yields consistent performance gains over the strong REINFORCE, minimum risk training, and proximal policy optimization methods. The code is available at https://github.com/wangclnlp/DeepSpeed-Chat-Extension/examples/esrl.