Set Of Benchmarks Research Articles

How Good are Current Pocket-Based 3D Generative Models?: The Benchmark Set and Evaluation of Protein Pocket-Based 3D Molecular Generative Models.

The development of a three-dimensional (3D) molecular generative model based on protein pockets has recently attracted a lot of attention. This type of model aims to achieve the simultaneous generation of molecular graphs and 3D binding conformation under the constraint of protein binding. Various pocket-based generative models have been proposed; however, currently, there is a lack of systematic and objective evaluation metrics for these models. To address this issue, a comprehensive benchmark data set, named POKMOL-3D, is proposed to evaluate protein pocket-based 3D molecular generative models. It includes 32 protein targets together with their known active compounds as a test set to evaluate the versatility of generation models to mimic the real-world scenario. Additionally, a series of two-dimensional (2D) and 3D evaluation metrics with some newly created ones was integrated to assess the quality of generated molecular structures and their binding conformations. It is expected that this work can enhance our comprehension of the effectiveness and weakness of current 3D generative models and stimulate the discussion on challenges and useful guidance for developing the next wave of molecular generative models.

Kinase-Bench: Comprehensive Benchmarking Tools and Guidance for Achieving Selectivity in Kinase Drug Discovery.

Developing selective kinase inhibitors remains a formidable challenge in drug discovery because of the highly conserved structural information on adenosine triphosphate (ATP) binding sites across the kinase family. Tailoring docking protocols to identify promising kinase inhibitor candidates for optimization has long been a substantial obstacle to drug discovery. Therefore, we introduced "Kinase-Bench," a pioneering benchmark suite designed for an advanced virtual screen, to improve the selectivity and efficacy of kinase inhibitors. Our comprehensive data set includes 6875 selective ligands and 422,799 decoys for 75 kinases, using extensive bioactivity and structural data from the ChEMBL database and decoys generated by the Directory of Useful Decoys-Enhanced version. Our benchmarking sets and retrospective case studies were designed to provide useful guidance in discovering selective kinase inhibitors. We employed a Glide High-Throughput Virtual Screen and Standard Precision complemented by three scoring functions and customized protein-ligand interaction filters that target specific kinase residue interactions. These innovations were successfully implemented in our virtual screen efforts targeting JAK1 inhibitors, achieving selectivity against its family member, TYK2. Consequently, we identified novel potential hits: Compound 2 (JAK1 IC50: 980.5 nM, TYK2 IC50: 4.5 μM) and the approved pan-AKT inhibitor Capivasertib (JAK1 IC50: 275.9 nM, TYK2 IC50: 10.9 μM). Using the Kinase-Bench protocol, both compounds demonstrated substantial JAK1 selectivity, making them strong candidates for further investigation. Our pharmaceutical results underscore the utility of tailored virtual screen protocols in identifying selective kinase inhibitors with substantial implications for rational drug design. Kinase-Bench offers a robust toolset for selective kinase drug discovery with the potential to effectively guide future therapeutic strategies effectively.

Effects of reducing redundant parameters in parameter optimization for symbolic regression using genetic programming

Gradient-based local optimization has been shown to improve results of genetic programming (GP) for symbolic regression (SR) – a machine learning method for symbolic equation learning. Correspondingly, several state-of-the-art GP implementations use iterative nonlinear least squares (NLS) algorithms for local optimization of parameters. An issue that has however mostly been ignored in SR and GP literature is overparameterization of SR expressions, and as a consequence, bad conditioning of NLS optimization problem. The aim of this study is to analyze the effects of overparameterization on the NLS results and convergence speed, whereby we use Operon as an example GP/SR implementation. First, we demonstrate that numeric rank approximation can be used to detect overparameterization using a set of six selected benchmark problems. In the second part, we analyze whether the NLS results or convergence speed can be improved by simplifying expressions to remove redundant parameters with equality saturation. This analysis is done with the much larger Feynman symbolic regression benchmark set after collecting all expressions visited by GP, as the simplification procedure is not fast enough to use it within GP fitness evaluation. We observe that Operon frequently visits overparameterized solutions but the number of redundant parameters is small on average. We analyzed the Pareto-optimal expressions of the first and last generation of GP, and found that for 70% to 80% of the simplified expressions, the success rate of reaching the optimum was better or equal than for the overparameterized form. The effect was smaller for the number of NLS iterations until convergence, where we found fewer or equal iterations for 51% to 63% of the expressions after simplification.

Q-BENCH: A Benchmark for Multi-modal Foundation Models on Low-level Vision from Single Images to Pairs.

The rapid development of Multi-modality Large Language Models (MLLMs) has navigated a paradigm shift in computer vision, moving towards versatile foundational models. However, evaluating MLLMs in low-level visual perception and understanding remains a yet-to-explore domain. To this end, we design benchmark settings to emulate human language responses related to low-level vision: the low-level visual perception (A1) via visual question answering related to low-level attributes (e.g. clarity, lighting); and the low-level visual description (A2), on evaluating MLLMs for low-level text descriptions. Furthermore, given that pairwise comparison can better avoid ambiguity of responses and has been adopted by many human experiments, we further extend the low-level perception-related questionanswering and description evaluations of MLLMs from single images to image pairs. Specifically, for perception (A1), we carry out the LLVisionQA+ dataset, comprising 2,990 single images and 1,999 image pairs each accompanied by an open-ended question about its low-level features; for description (A2), we propose the LLDescribe+ dataset, evaluating MLLMs for low-level descriptions on 499 single images and 450 pairs. Additionally, we evaluate MLLMs on assessment (A3) ability, i.e. predicting score, by employing a softmax-based approach to enable all MLLMs to generate quantifiable quality ratings, tested against human opinions in 7 image quality assessment (IQA) datasets. With 24 MLLMs under evaluation, we demonstrate that several MLLMs have decent low-level visual competencies on single images, but only GPT-4V exhibits higher accuracy on pairwise comparisons than single image evaluations (like humans). We hope that our benchmark will motivate further research into uncovering and enhancing these nascent capabilities of MLLMs. Datasets will be available at https://github.com/Q-Future/Q-Bench.

Graph protection under multiple simultaneous attacks: A heuristic approach

This work focuses on developing a meta-heuristic approach to protect network nodes from simultaneous attacks, specifically addressing the k-strong Roman domination problem. The objective is to assign integer weights to the nodes, representing the number of stationed armies, to meet protection constraints while minimizing the total number of armies. A network is protected if it can repel any simultaneous attack on k nodes. A node is protected if it can defend itself or if a neighboring node provides an army while retaining at least one army for self-defense. This problem formulation can be used in practical scenarios, e.g. developing counter-terrorism strategies or in coping with supply chain disruptions. The problem is difficult as even verifying the feasibility of a single solution generally requires an exponential time. Two exact approaches are proposed in the literature but applicable to small random graphs. For larger graphs, we propose an effective variable neighborhood search, where the feasibility of a solution is verified by introducing the concept of relaxed feasibility. Experiments are conducted with random networks from the literature and two introduced ad-hoc wireless and real-world networks. Extensive experimental evaluations show the robustness of the proposed approach compared to the existing approaches from the literature by significantly outperforming them in all three benchmark sets. Furthermore, we demonstrate the practical application of the proposed variable neighborhood search approach, where its solution is used to position fire stations within the city so that simultaneous fires can be extinguished efficiently while reducing the number of required fire trucks.

Grammatical Redundancy in Scales: Using the “ConGRe” Process to Create Better Measures

As theoretical models become more complex, there is more pressure to use less time-consuming methods generally, and shorter scales specifically. Although reliability is related to scale length, reliability cutoffs are easily met, even in very short scales, by writing or selecting items that are worded in nearly identical ways, that is, grammatical redundancy. However, grammatical redundancy increases reliability at the cost of domain sampling—a crucial early step in scale construction and one of the two pillars of content validity. Without it, a scale cannot capture the intended construct. The purpose of this paper is to provide scale developers (and shorteners) with a process for quantifying, identifying, and reducing grammatical redundancy without compromising conceptual redundancy, a process that we label ConGRe. Our process involves indices from the linguistics literature that can be used to guide decisions during item writing, that is, prior to data collection. We examine their relation to more traditional psychometric indicators and provide a set of benchmarks. Overall, we demonstrate that it is possible to reduce grammatical redundancy, thus avoiding scale deficiency, without sacrificing traditional psychometric properties.

A Survey of General-purpose Polyhedral Compilers

Since the 1990s, many implementations of polyhedral compilers have been written and distributed, either as source-to-source translating compilers or integrated into wider-purpose compilers. This article provides a survey on those various available implementations as of today, 2024. First, we list and describe most commonly available polyhedral schedulers and compiler implementations. Then, we compare the general-purpose polyhedral compilers using two main criteria—robustness and performance—on the PolyBench/C set of benchmarks.

Blackbird: structural variant detection using synthetic and low-coverage long-reads.

Recent benchmarks of structural variant (SV) detection tools revealed that the majority of human genome structural variations (SVs), especially the medium-range (50-10,000 bp) SVs cannot be resolved with short-read sequencing, but long-read SV callers achieve great results on the same datasets. While improvements have been made, high-coverage long-read sequencing is associated with higher costs and input DNA requirements. To decrease the cost one can lower the sequence coverage, but the current long-read SV callers perform poorly with coverage below 10X. Synthetic long-read (SLR) technologies hold great potential for structural variant (SV) detection, although utilizing their long-range information for events smaller than 50 kbp has been challenging. Results: In this work, we propose a hybrid novel integrated alignment- and local-assembly-based algorithm, Blackbird, that uses SLR together with low-coverage long reads to improve SV detection and assembly. Without the need for a computationally expensive whole genome assembly, Blackbird uses a sliding window approach and barcode information encoded in SLR to accurately assemble small segments and use long reads for an improved gap closing and contig assembly. We evaluated Blackbird on simulated and real human genome datasets. Using the HG002 GIAB benchmark set, we demonstrated that in hybrid mode, Blackbird demonstrated results comparable to state-of-the-art long-read tools, while using less long-read coverage. Blackbird requires only 5X coverage to achieve F1 scores (0.835 and 0.808 for deletions and insertions) similar to PBSV (0.856 and 0.812) and Sniffles2 (0.839 and 0.804) using 10X Pacbio Hi-Fi long-read coverage.

BN SO20 - Day-case laparoscopic cholecystectomy: Predictors of failure

Abstract Background Laparoscopic Cholecystectomy (LC) is the main treatment option for symptomatic gallstone disease. It is commonly performed as a day case operation and Getting It Right First Time (GIRFT) have suggested that this should be the default position for all trusts. NHSE Model Hospital analyses hospital metrics and sets national benchmarks. They suggest that the day case rate for LC should be greater than 71.4%. The aim of our study is to analyse all elective LC’s over a two year period and identify factors that increase the likelihood of day-case failure and the common reasons for this. Method All patients who had LC between 1st July 2020 and 30th June 2022 were identified. All emergency cholecystectomies were excluded from analysis. Patient and operative details were retrieved from their electronic record and a Charlson co-morbidity index was calculated. Statistical analysis was performed using SPSS with Chi-squared tests performed for categorical data and ANOVA for continuous data. Odds ratios with 95% confidence intervals were calculated. Statistical significance was taken as p<0.05. Results 788 patients had elective cholecystectomies with 74.0% being successful day-case. Factors increasing day-case failure included: Admission with gallbladder disease within past year (OR 3.15 (2.26-4.37), CBD diameter>8mm (OR 2.17(1.44-3.26), thick-walled gallbladder (OR 2.20(1.59-3.04), on-table cholangiogram (OR 4.16 (2.21-7.84), sub-total cholecystectomy (OR 30.6(9.19-102) and male sex (OR 1.55(1.11-2.18). Patients who failed day-case had higher mean age (59 vs 49) and higher co-morbidity index (2.16 vs 1.44). The most common reasons for day-case failure were complex procedure (41%), and symptom control (35%). Only 20% had planned admissions. 30-day morbidity and re-admission rate were higher for failed day-case patients. Conclusion Our hospital complies with the day case benchmark set out by Model Hospital. Furthermore, only 5.2% of patients undergoing elective cholecystectomy had a pre-operative decision for an inpatient stay demonstrating that day case is the default for the vast majority of patients. Despite this, improvement could still be made with post-operative symptom control to try and increase this number further.

DecTree: a physics-based geochemical surrogate for surface complexation of uranium on clay

Abstract. Geochemistry is usually the computational bottleneck in coupled reactive transport simulations, which hampers the complexity of the systems and of the processes they can investigate. In recent years, promising speedups have been obtained by substituting the numerical solution of geochemical models with approximated surrogates borrowed from artificial intelligence and machine learning (AI/ML). In the framework of the DONUT/EURAD project a set of benchmarks were defined to assess the performance and the accuracy of different surrogate approaches in settings relevant to the safety assessment of nuclear waste repositories, such as the surface complexation and exchange of U(VI) on clay. In this context, this work introduces am original surrogate modelling approach based on recursive partitioning of parameter space, which exploits prior domain knowledge for the training. The surrogate, which can be represented as a decision tree, hence the DecTree name, performs dimensionality reduction by identifying functional relationships between outputs and input variables using a straightforward non-monotonic extension of the Spearman's rank correlation coefficient. New predictions are then interpolated from the partitioned training data. Applied to a low-dimensional geochemical model, DecTree shows virtually no training time and excellent accuracy, ensuring a throughput of around 500 000 predictions per second on a single CPU core.

A Novel Solution Approach Based on Dominance Evaluation Measure for Project Scheduling in Multi-Project Environments

The widely recognized measure for resources called resource strength (RS) does not fully capture the resources complexity of a project. Therefore, it cannot be used as a standalone measure to distinguish the complexity of various instances of project scheduling problems. Consequently, additional resource measures such as total amount of overflow (TAO) have been introduced, which should be used in conjunction with the RS. Extensive experimental studies have shown that as the value of TAO increases in a project, scheduling schemes with higher dimensional scheduling schemes such as bi-directional and tri-directional result in schedules with shorter makespans. In this study, an effective approach is proposed for integrating projects in multi-project environments, called the integrated project approach (IPA), taking into account the influence of TAO and building upon the relation between the TAO and the scheduling generation schemes. To assess the performance of IPA, we develop a new random multi-project generator based on the well-known benchmark sets, which utilizes TAO as a control tool to generate instances. The findings indicate that prioritizing the projects and frequency of the projects integration, facilitated by the proposed IPA, have a positive impact on the quality of multi-project schedules.

Automating TODO-missed Methods Detection and Patching

TODO comments are widely used by developers to remind themselves or others about incomplete tasks. In other words, TODO comments are usually associated with temporary or suboptimal solutions. In practice, all the equivalent suboptimal implementations should be updated (e.g., adding TODOs) simultaneously. However, due to various reasons (e.g., time constraints or carelessness), developers may forget or even are unaware of adding TODO comments to all necessary places, which results in the TODO-missed methods . These “hidden” suboptimal implementations in TODO-missed methods may hurt the software quality and maintainability in the long-term. Therefore, in this paper, we propose the novel task of TODO-missed methods detection and patching, and develop a novel model, namely TDPatcher ( T O D O-comment Patcher ), to automatically patch TODO comments to the TODO-missed methods in software projects. Our model has two main stages: offline learning and online inference. During the offline learning stage, TDPatcher employs the GraphCodeBERT and contrastive learning for encoding the TODO comment (natural language) and its suboptimal implementation (code fragment) into vector representations. For the online inference stage, we can identify the TODO-missed methods and further determine their patching position by leveraging the offline trained model. We built our dataset by collecting TODO-introduced methods from the top-10,000 Python GitHub repositories and evaluated TDPatcher on them. Extensive experimental results show the promising performance of our model over a set of benchmarks. We further conduct an in-the-wild evaluation which successfully detects 26 TODO-missed methods from 50 GitHub repositories.

A vessel bifurcation liver CT landmark pair dataset for evaluating deformable image registration algorithms.

Evaluating deformable image registration (DIR) algorithms is vital for enhancing algorithm performance and gaining clinical acceptance. However, there is a notable lack of dependable DIR benchmark datasets for assessing DIR performance except for lung images. To address this gap, we aim to introduce our comprehensive liver computed tomography (CT) DIR landmark dataset library. This dataset is designed for efficient and quantitative evaluation of various DIR methods for liver CTs, paving the way for more accurate and reliable image registration techniques. Forty CT liver image pairs were acquired from several publicly available image archives and authors' institutions under institutional review board (IRB) approval. The images were processed with a semi-automatic procedure to generate landmark pairs: (1) for each case, liver vessels were automatically segmented on one image; (2) landmarks were automatically detected at vessel bifurcations; (3) corresponding landmarks in the second image were placed using two deformable image registration methods to avoid algorithm-specific biases; (4) a comprehensive validation process based on quantitative evaluation and manual assessment was applied to reject outliers and ensure the landmarks' positional accuracy. This workflow resulted in an average of ∼56 landmark pairs per image pair, comprising a total of 2220 landmarks for 40 cases. The general landmarking accuracy of this procedure was evaluated using digital phantoms and manual landmark placement. The landmark pair target registration errors (TRE) on digital phantoms were 0.37±0.26 and 0.55±0.34mm respectively for the two selected DIR algorithms used in our workflow, with 97% of landmark pairs having TREs below 1.5mm. The distances from the calculated landmarks to the averaged manual placement were 1.27±0.79mm. All data, including image files and landmark information, are publicly available at Zenodo (https://zenodo.org/records/13738577). Instructions for using our data can be found on our GitHub page at https://github.com/deshanyang/Liver-DIR-QA. The landmark dataset generated in this work is the first collection of large-scale liver CT DIR landmarks prepared on real patient images. This dataset can provide researchers with a dense set of ground truth benchmarks for the quantitative evaluation of DIR algorithms within the liver.

Accurate prediction of protein–ligand interactions by combining physical energy functions and graph-neural networks

We introduce an advanced model for predicting protein–ligand interactions. Our approach combines the strengths of graph neural networks with physics-based scoring methods. Existing structure-based machine-learning models for protein–ligand binding prediction often fall short in practical virtual screening scenarios, hindered by the intricacies of binding poses, the chemical diversity of drug-like molecules, and the scarcity of crystallographic data for protein–ligand complexes. To overcome the limitations of existing machine learning-based prediction models, we propose a novel approach that fuses three independent neural network models. One classification model is designed to perform binary prediction of a given protein–ligand complex pose. The other two regression models are trained to predict the binding affinity and root-mean-square deviation of a ligand conformation from an input complex structure. We trained the model to account for both deviations in experimental and predicted binding affinities and pose prediction uncertainties. By effectively integrating the outputs of the triplet neural networks with a physics-based scoring function, our model showed a significantly improved performance in hit identification. The benchmark results with three independent decoy sets demonstrate that our model outperformed existing models in forward screening. Our model achieved top 1% enrichment factors of 32.7 and 23.1 with the CASF2016 and DUD-E benchmark sets, respectively. The benchmark results using the LIT-PCBA set further confirmed its higher average enrichment factors, emphasizing the model’s efficiency and generalizability. The model’s efficiency was further validated by identifying 23 active compounds from 63 candidates in experimental screening for autotaxin inhibitors, demonstrating its practical applicability in hit discovery.Scientific contributionOur work introduces a novel training strategy for a protein–ligand binding affinity prediction model by integrating the outputs of three independent sub-models and utilizing expertly crafted decoy sets. The model showcases exceptional performance across multiple benchmarks. The high enrichment factors in the LIT-PCBA benchmark demonstrate its potential to accelerate hit discovery.

The First Catalog of Candidate White Dwarf–Main-sequence Binaries in Open Star Clusters: A New Window into Common Envelope Evolution

Close binary systems are the progenitors to both Type Ia supernovae and the compact object mergers that can be detected via gravitational waves. To achieve a binary with a small radial separation, it is believed that the system likely undergoes common envelope (CE) evolution. Despite its importance, CE evolution may be one of the largest uncertainties in binary evolution due to a combination of computational challenges and a lack of observed benchmarks where both the post-CE and pre-CE conditions are known. Identifying post-CE systems in star clusters can partially circumvent this second issue by providing an independent age constraint on the system. For the first time, we conduct a systematic search for white dwarf and main-sequence binary systems in 299 Milky Way open star clusters. Coupling Gaia DR3 photometry and kinematics with multiband photometry from Pan-STARRS1 and the Two Micron All Sky Survey, we apply a machine learning-based approach and find 52 high-probability candidates in 38 open clusters. For a subset of our systems, we present follow-up spectroscopy from the Gemini and Lick Observatories and archival light curves from the Transiting Exoplanet Survey Satellite, Kepler/K2, and the Zwicky Transient Facility. Examples of M dwarfs with hot companions are spectroscopically observed, along with regular system variability. While the kinematics of our candidates are consistent with their host clusters, some systems have spatial positions offset relative to their hosts, potentially indicative of natal kicks. Ultimately, this catalog is a first step to obtaining a set of observational benchmarks to better link post-CE systems to their pre-CE progenitors.

Optimal proposal particle filters for detecting anomalies and manoeuvres from two line element data

Detecting the manoeuvres of satellites is an important goal within the broader task of space situational awareness. Moreover, it is advantageous to be able to detect other anomalous behaviour from satellites arising from, for instance, collisions or malfunction. The Two Line Element (TLE) data published by NORAD is the only publicy-available comprehensive source of data for satellite orbits. This paper presents a filtering approach for detecting anomalies in satellite orbits from TLE data. Particle filters are deployed to track the state of the satellites’ orbits. New TLEs that are deemed sufficiently unlikely given one’s belief of the current orbital state are designated as anomalies. The change in the orbits over time is modelled using the SGP4 model. However, it is shown that this model can have large errors in certain elements in some instances. The structure of these errors is analysed and a suitable model uncertainty is derived. The optimal proposal filter is used to further reduce the impact of the errors in the SGP4 model. This strategy improves the proposal distribution from which the particles are drawn before re-weighting, so that the resulting weights are less degenerate. The proposed techniques are evaluated on a benchmark set of 15 satellites for which an externally-obtained ground truth is available, along with simulated orbital data with inserted manoeuvres. Although this benchmark is larger than the evaluation sets used in other published work, it cannot be representative of the full population of satellites and so more emphasis is placed on the simulated data. The proposed techniques are compared against a baseline, which is similar to many previously-proposed techniques for satellite manoeuvre detection. The particle filters are shown to be superior at detecting the subtle in-track and cross-track manoeuvres in the simulated dataset. However, on the benchmark dataset, a simple baseline that focusses on the mean motion can perform equivalently to the best filters. This simple technique is likely overfit to the benchmark.

DTAP: Accelerating Strongly-Typed Programs with Data Type-Aware Hardware Prefetching

Queries on linked data structures, such as trees and graphs, often suffer from frequent cache misses and significant performance loss due to dependent and random pointer-chasing memory accesses. In this paper, we propose a software-hardware co-designed solution for accelerating linked data structures implemented in strongly typed languages. The solution incorporates a compiler extension and a hardware prefetcher. The compiler extension extracts type information from the code, annotates each load instruction, and forwards the type information to the hardware prefetcher. The prefetcher leverages the type information to fetch the referred objects and identify the associated pointers in advance. By doing so, the program can find these objects in the cache when it follows the prefetched pointers, thus minimizing cache misses. In the evaluation, the proposed solution achieves an average speedup of 1.37 × over a set of memory-intensive benchmarks.

Solving UAV 3D Path Planning Based on the Improved Lemur Optimizer Algorithm.

This paper proposes an Improved Lemur Optimization algorithm (ILO), which combines the advantages of the Spider Monkey Optimization algorithm, Simulated Annealing algorithm, and Lemur Optimization algorithm. Through the use of an adaptive nonlinear decrement model, adaptive learning factors, and updated jump rates, the algorithm enhances its global exploration and local exploitation capabilities. A Gaussian function model is used to simulate the mountain environment, and a mathematical model for UAV flight is established based on constraints and objective functions. The fitness function is employed to determine the minimum cost for avoiding obstacles in a designated airspace, and cubic spline interpolation is used to smooth the flight path. The Improved Lemur Optimization algorithm was tested using the CEC2017 benchmark set, assessing its search capability, convergence speed, and accuracy. The simulation results show that ILO generates high-quality, smooth paths with fewer iterations, overcoming the issues of premature convergence and insufficient local search ability in traditional genetic algorithms. It adapts to complex terrain, providing an efficient and reliable solution.

Smoother Semiclassical Dispersion for Density Functional Theory via D3S: Understanding and Addressing Unphysical Minima in the D3 Dispersion Correction Model.

One of the most widely used and computationally efficient models that accounts for London dispersion interactions within density functional theory (DFT) is the D3 dispersion correction model. In this work, we demonstrate that this model can induce the appearance of unphysical minima on the potential energy surface (PES) when the coordination number of atoms changes. Optimizing to these artifactual structures can lead to significant errors in determining the interaction energy between two molecules and in estimating the thermodynamic properties of the system. In several specific examples, such as Kuratowski-type H2-NiKur and H2-PdKur clusters, these local minima exhibited extremely high PES curvature, resulting in incorrect estimations of harmonic frequencies and significant overestimations of zero-point energy and enthalpy values. Although such erroneous behavior of the D3 model is relatively rare, it can occur across a wide range of chemical species, including molecules like the [Li(C6H6)]+ complex and the dispiro(acridan)-substituted pyracene (DSAP) molecule. Our analysis reveals that the root of the problem lies in the definition of the AB atomic-pair dependent C6AB coefficients in the D3 model. To address this issue, we propose a reparameterization of the D3 model by introducing a modified C6AB functional form that now depends on the specific pair of considered atoms. This new model, termed D3-Smooth (or D3S for short), is designed to smooth out the PES associated with the dispersion correction. By doing so, we demonstrate that D3S eliminates unphysical local minima while maintaining the quite satisfactory accuracy of the parent D3 method in interaction energy benchmark sets. For example, the RMS difference between using the D3(BJ) correction to B3LYP and the D3S(BJ) correction across the large MGCDB84 data set of nearly 5000 data points is only 0.12 kJ/mol. Similar results are obtained for every other D3-corrected functional tested. Consistent with this result, no significant improvement could be obtained for the B3LYP-D3S(0) correction by reoptimizing the damping function.

Lexical Effect Handlers, Directly

Lexically scoping effect handlers is a language-design idea that equips algebraic effects with a modular semantics: it enables local-reasoning principles without giving up on the control-flow expressiveness that makes effect handlers powerful. However, we observe that existing implementations risk incurring costs akin to the run-time search for dynamically scoped handlers. This paper presents a compilation strategy for lexical effect handlers, adhering to the lexical scoping principle and targeting a language with low-level control over stack layout. Key aspects of this approach are formalized and proven correct. We embody the ideas in a language called Lexa: the Lexa compiler translates high-level effect handling to low-level stack switching. We evaluate the Lexa compiler on a set of benchmarks; the results suggest that it generates efficient code, reducing running-time complexity from quadratic to linear in some cases.