Stopping Criterion Research Articles

Automated University Lecture Schedule Generator based on Evolutionary Algorithm

university is a complicated work so in the implementation it have violation of the constraints and it also takes a lot of time since it is created manually. In this paper evolutionary algorithm (EA) is used to create an effective and feasible schedules based on the real data input that is obtained from each department. The objective functions in EA contribute in gaining the fitness function to solve the constraints problem in the schedule by applying weighting for each hard constraints. The objective function is gained from the total of infringement in each soft constraints addition by score weighting. The genetic operator used in EA is stochastic variation Operator. As far as the reproduction operator is concerned, the tournament selection was used with size 3. Crossover operator is conducted after selection process with crossover probability equal to 0.05 and mutation rate is 0.1. The size of population was set to 9 and stopping criteria algorithm was left run for fitness value = 1. The simulation result shows that EA can create lecture schedules efficiently and feasibly. Moreover, it is also faster with the execution time of the proposed EA is less than 30 and easier than creating manually.

DxFormer: a decoupled automatic diagnostic system based on decoder-encoder transformer with dense symptom representations.

Symptom-based automatic diagnostic system queries the patient's potential symptoms through continuous interaction with the patient and makes predictions about possible diseases. A few studies use reinforcement learning (RL) to learn the optimal policy from the joint action space of symptoms and diseases. However, existing RL (or Non-RL) methods focus on disease diagnosis while ignoring the importance of symptom inquiry. Although these systems have achieved considerable diagnostic accuracy, they are still far below its performance upper bound due to few turns of interaction with patients and insufficient performance of symptom inquiry. To address this problem, we propose a new automatic diagnostic framework called DxFormer, which decouples symptom inquiry and disease diagnosis, so that these two modules can be independently optimized. The transition from symptom inquiry to disease diagnosis is parametrically determined by the stopping criteria. In DxFormer, we treat each symptom as a token, and formalize the symptom inquiry and disease diagnosis to a language generation model and a sequence classification model, respectively. We use the inverted version of Transformer, i.e. the decoder-encoder structure, to learn the representation of symptoms by jointly optimizing the reinforce reward and cross-entropy loss. We conduct experiments on three real-world medical dialogue datasets, and the experimental results verify the feasibility of increasing diagnostic accuracy by improving symptom recall. Our model overcomes the shortcomings of previous RL-based methods. By decoupling symptom query from the process of diagnosis, DxFormer greatly improves the symptom recall and achieves the state-of-the-art diagnostic accuracy. Both code and data are available at https://github.com/lemuria-wchen/DxFormer. Supplementary data are available at Bioinformatics online.

Preliminary Safety and Efficacy of Bgb-11417, a Novel Bcl-2 Inhibitor, in Combination with Azacitidine in Patients with Acute Myeloid Leukemia (AML)

Background/Introduction: The efficacy of Bcl-2 inhibitors for treating AML was recognized by the approval of venetoclax and azacitidine (aza) in patients (pts) with newly diagnosed AML ineligible for intensive chemotherapy. However, survival rates beyond 2 years are low. Compared to venetoclax, BGB-11417 is a much more potent Bcl-2 inhibitor (>10-fold in biochemical assays) with improved selectivity and the potential to achieve deeper target inhibition and responses in the clinical setting. Here, we present updated preliminary results of pts with AML enrolled in BGB-11417-103 (NCT04771130). Methods: BGB-11417-103 is an ongoing dose finding/expansion study evaluating BGB-11417 and aza in pts with AML (treatment-naïve [TN] unfit for intensive chemotherapy or relapsed/refractory [R/R]), myelodysplastic syndrome (MDS), or MDS/myeloproliferative neoplasm. For all indications, pts who received prior aza or Bcl-2 inhibitors were excluded. For pts with AML, the 10-day regimen comprised BGB-11417 at 40 mg (cohort 1), 80 mg (cohort 2), or 160 mg (cohort 3); 160 mg was also used in a 28-day regimen (cohort 4). All dose levels were given in combination with aza (75 mg/m2 x 7 days). In cycle 1, a 4-day ramp-up of BGB-11417 starting at 1/8 target dose was used. Patients with dose-limiting toxicities (DLTs) in cycle 1 were assessed against the number of patients dosed. The safety stopping criteria were based on the number of patients with events where posterior probability of event rate exceeding 0.25 was at least 80%. Responses were determined using 2017 European LeukemiaNet (ELN) criteria with assessment of hematologic improvement. Treatment-emergent adverse events (TEAEs), minimal residual disease (MRD) by flow cytometry, and pharmacokinetics (PK) were also assessed. Results: As of June 1, 2022, 51 pts with AML were treated (16 in cohort 1; 17 in cohort 2; 14 in cohort 3; 4 in cohort 4). Median age was 77 yrs (TN, n=28) and 64 yrs (R/R, n=23) (Table 1). At baseline, 23 pts (45.1%) had adverse-risk cytogenetics, 37 (72.5%) had grade ≥3 neutropenia, and 27 (52.9%) had grade ≥3 thrombocytopenia. At a median follow-up of 2.8 mo (range 0.1-12.3) and median duration of treatment of 1.9 mo (range 0.0-12.3), 2 of 41 evaluable pts experienced DLTs (grade 4 neutropenia and grade 4 thrombocytopenia at 80 mg) (Table 2), which did not meet the safety stopping protocol criteria. Laboratory tumor lysis syndrome (TLS; Howard criteria) was observed in 1 pt (160 mg x 10 days) with a known history of chronic kidney disease; TLS resolved within 4 days. Twenty-one (41%) pts discontinued study drugs: 7 (13.7%) due to AEs, 5 (9.8%) due to disease progression, 4 (7.8%) proceeded to transplant, 3 (5.9%) withdrew consent, 1 (2.0%) per investigator decision, and 1 (2.0%) started new anti-cancer treatment. The most common TEAEs were neutropenia (58.8%), thrombocytopenia (45.1%), anemia (43.1%), febrile neutropenia (33.3%), nausea (39.2%), and constipation (37.3%). Neutropenia was the most common grade ≥3 TEAE observed in 30 (58.8%) pts; most cases did not require dose modification. Grade ≥3 infections occurred in 23 (45.1%) pts, mostly in cycle 1. The general incidence of febrile neutropenia and infections decreased with subsequent cycles. Four (7.8%) pts died due to unrelated TEAEs: bronchopulmonary aspergillosis, pneumonia, pulmonary sepsis, and possible intracranial hemorrhage after a fall. TEAEs leading to dose reduction (5.9%) or dose interruption (11.8%) were infrequent. In TN AML, complete remission (CR)/CR with partial hematologic recovery (CRh) was achieved in 16 (59.3%) and CR in 13 (48.1%) evaluable pts (Table 1). Of the 13 pts in CR, 7 (53.8 %) reached CR by end of cycle 1. In cohort 2 with the longest follow-up, CR was achieved in 8 (72.7%) pts with TN AML. Notably, 7 of these pts are continuing study treatment (median of 5 cycles, range 1-12) without progression. In R/R AML, CR/CRh was seen in 50% of evaluable pts, with a CR rate of 25%. Nine (50.0%; 6 TN, 3 R/R) of 18 evaluable pts with CR/CRh achieved MRD negativity (<0.1% per ELN 2018). In pts with AML treated with 40, 80, and 160 mg, PK increased in a dose-dependent manner at steady state with no drug-drug interaction observed with aza. Conclusion: These results demonstrate that BGB-11417 in combination with aza was effective and generally well tolerated at the 4 dose levels tested in pts with AML, with no dose-dependent toxicities observed. Further evaluation of the 28-day regimen is underway. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

Anticoagulation and Antiplatelet Strategies in Non-Critically Ill Patients with Covid-19

Background Optimal thromboprophylaxis for hospitalized patients with Covid-19 is uncertain. Most published clinical trials comparing different anticoagulation protocols in hospitalized patients with Covid-19 have recruited patients in high-income countries (HIC) and results may not be generalizable to patients in low and middle-income countries (LMIC). Furthermore, the incidence of VTE varies by ethnic background with higher rates reported amongst Caucasian compared to diverse Asian populations. We report the results of an international (incorporating both HIC and LMIC regions) adaptive platform randomized trial in hospitalized non-critically ill patients with Covid-19 comparing the effectiveness of low-dose, intermediate-dose, therapeutic-dose anticoagulation, and low-dose plus aspirin on survival and need for organ support. Methods The Australasian COVID-19 Trial (ASCOT) is an investigator-initiated, open-label, pragmatic adaptive platform randomized clinical trial of therapeutics (anticoagulation, antiviral, and therapeutic antibody domains) for non-critically ill patients hospitalized with Covid-19. In our anticoagulation trial, we initially randomly assigned patients to low-dose low-molecular-weight heparin (LMWH) thromboprophylaxis, intermediate-dose or low-dose plus aspirin. In response to external evidence, the aspirin intervention was discontinued and a therapeutic-dose arm added. The primary endpoint was death or the requirement for new organ support by day-28, analyzed with a Bayesian logistic model. Enrolment was closed due to operational constraints. Results Between February 2021 and April 2022, we screened 2,203 patients in 32 hospitals (15 in Australia, 4 in New Zealand, 11 in India, and 2 in Nepal) and enrolled 1574 into the anticoagulation domain (619 assigned to low-dose, 620 to intermediate-dose, 285 to low dose plus aspirin, and 50 to therapeutic-dose). Eighteen participants withdrew consent and 30 were lost to follow-up leaving 1526 included in the primary analysis population (1273 from India, 138 from Australia and New Zealand, 115 from Nepal). The median age of participants was 49 years, 633 (41%) were female, 1291 (83%) were Indian, 480 (31%) were vaccinated against SARS-CoV-2, and 629 (40%) had one or more comorbidities. Adherence to the trial protocol was 98% in the low-dose, 98% in the intermediate-dose, 97% in the low-dose plus aspirin and 92% in the therapeutic-dose anticoagulation arms. All participants received enoxaparin except for one participant in the intermediate group who received tinzaparin. At 28 days after randomization, 35/596 (5.9%) participants assigned to low-dose, 25/601 (4.2%) intermediate-dose, 20/279 (7.2%) low-dose plus aspirin, and 7/50 (14%) therapeutic-dose anticoagulation had died or required organ support. Compared to low-dose thromboprophylaxis, the median adjusted odds ratio (aOR) for the primary outcome for intermediate-dose was 0.74 (95% credible interval [CrI] 0.43 to 1.27, posterior probability of effectiveness [aOR < 1] [Pr] 86%), for low-dose plus aspirin 0.88 (95% CrI 0.47 to 1.64, Pr 65%) and for therapeutic-dose 2.22 (95% CrI 0.77 to 6.20, Pr 7%). The results were consistent in a number of sensitivity analyses including when restricting the analysis populations to concurrent randomizations. The results were also broadly consistent across the predefined subgroups, with the posterior probability of effectiveness for intermediate-dose compared to low-dose ranging from 56% (corticosteroid exposure) to 94% (>7 days of symptom onset to hospitalization). Overall thrombotic and bleeding rates were 0.8% and 0.4%, respectively. Conclusions In hospitalized adults with Covid-19, compared to low-dose, there was an 86% posterior probability that intermediate-dose, and 7% posterior probability that therapeutic-dose, reduced the odds of death or requirement for organ support by day 28. No treatment strategy met pre-specified statistical stopping criteria prior to trial closure. Trial registration number: NCT04483960

Lithium-ion battery design optimization based on a dimensionless reduced-order electrochemical model

Model-based optimal cell design is an efficient approach to maximize the energy density of lithium-ion batteries. This maximization problem is solved in this paper for a lithium iron phosphate (LFP) cell. We consider half-cells as opposed to full-cells typically considered, which are intermediate steps during battery manufacturing for electrode characterization and they are gaining popularity by themselves as lithium-metal batteries. First, a dimensionless reduced-order electrochemical model is used instead of high-order models. Second, sensitivity equations are analyzed to determine the ranking of the design parameters according to their effect on the energy density, which is often lacking in other contributions. Three parameters, namely electrode thickness, LFP particle radius and electrode cross sectional area, are shown to have the most influential effects. Third, a novel adaptive particle swarm optimization with a specific stopping criterion is used for LIB design optimization. The proposed optimization framework is tested in simulation on a LFP half-cell battery. The results show that the design optimization yields 250 Wh kg−1 for an LFP electrode of 310μm thickness, 10 nm particle radius and 2⋅10−4 m2 cross-sectional area, which is an increase of energy density of 61 Wh kg−1 with respect to an initial design proposed in the literature.

EPITOPE STUDY RESULTS: PHASE 3, RANDOMIZED, DOUBLE-BLIND, PLACEBO-CONTROLLED STUDY OF EPICUTANEOUS IMMUNOTHERAPY IN PEANUT-ALLERGIC TODDLERS

<h3>Introduction</h3> No approved peanut allergy treatments exist for children <4 years old. This study aimed to assess the efficacy and safety of epicutaneous immunotherapy with a patch containing 250 µg peanut protein among peanut-allergic children ages 1-3 years. <h3>Methods</h3> In a Phase 3, multicenter, randomized, double-blind, placebo-controlled trial (EPITOPE), peanut-allergic children developing symptoms meeting stopping criteria during double-blind, placebo-controlled food challenge (DBPCFC) at an eliciting dose (ED) ≤300 mg peanut protein were randomized 2:1 to 12 months of daily treatment with the peanut patch (n=244) or placebo patch (n=118). The primary outcome was the percent difference in responders between active and placebo, determined by DBPCFC ED at baseline and Month 12. Safety was assessed by treatment-emergent adverse event (TEAE) rate. <h3>Results</h3> Of 362 participants randomized (median age: 2.5 years, 68.8% male), 84.8% completed treatment. The primary efficacy endpoint was met by 67% receiving active peanut patch vs. 33.5% receiving placebo (difference=33.4%; 95% CI: 22.4–44.5% [p<i><</i>0.001]). Additionally, 64.2% active vs. 29.6% placebo recipients (difference=34.7%; [p<i><</i>0.001]) achieved a peanut protein ED ≥1000 mg. The majority of TEAEs were mild or moderate application site reactions. Serious TEAEs occurred in 8.6% peanut patch vs. 2.5% placebo recipients, with 1.6% (n=4) peanut patch subjects experiencing treatment-related anaphylaxis, and 3.3% (n=8) peanut patch subjects discontinuing due to a TEAE. <h3>Conclusions</h3> Twelve months of epicutaneous immunotherapy with a patch containing 250 µg peanut protein was associated with a statistically significant response vs. placebo among peanut-allergic children aged 1-3 years. Rates of treatment-related anaphylaxis and discontinuations due to TEAEs were low.

Early arrival waveform inversion using data uncertainties and matching filters with application to near-surface seismic refraction data

We develop an early arrival waveform inversion (EAWI) technique for high-resolution near-surface velocity estimation by iteratively updating the P-wave velocity model to minimize the difference between the observed and calculated seismic refraction data. Traditional EAWI uses a least-squares penalty function and an acoustic forward-modeling engine. Conventional least-squares error is sensitive to data with low signal-to-noise ratio (S/N) and iterations of EAWI stop at a local-minimum data misfit or at the preassigned maximum number of iterations. These stopping criteria can result in overfitting the data. In addition, fitting the elastic field data with an acoustic modeling engine can introduce artifacts in velocity estimation, especially in land data with significant elastic effects. To overcome these challenges, we develop a robust EAWI (REAWI) method by (1) incorporating the data uncertainties into the penalty function and (2) mitigating the elastic effects using a matching filter workflow. The data uncertainties are estimated from waveform reciprocal errors. When full-waveform reciprocity is not available, trace interpolation is applied. The proposed method prevents closely fitting data with low S/N, avoids overall overfitting by stopping the iterations when a normalized chi-square ([Formula: see text]) waveform misfit of one is achieved, and is less affected by elastic effects. Numerical examples and application to near-surface refraction data at a groundwater contamination site suggest that the final REAWI models are more accurate than the corresponding EAWI models, at the same level of misfit. This is the first known application of a matching filter workflow to real land data. The final REAWI models satisfy an appropriate misfit between the real data and predicted elastic P-wave data, making this approach in this respect equivalent to elastic waveform inversion. We also develop a method to analyze model constraint by examining the energy of the wavefield Fréchet derivative thereby avoiding the influence of the data residuals in traditional Fréchet kernels.

A posteriori error estimations and convergence criteria in fast Fourier transform‐based computational homogenization

AbstractA stopping criterion for fast Fourier transform (FFT)‐based iterative schemes in computational homogenization is proposed and investigated numerically. This criterion is based on the separate evaluation and comparison of the discretization and iteration errors on the computed fields. Some estimators for these errors are proposed and their performances are assessed on a set of 2D problems in the frameworks of both the classical FFT‐based methods and these that use a modified version of the featured Green's operator. In particular, two novel strategies for estimating the discretization error are investigated: either using an image processing approach or transposing to the FFT‐based setting the constitutive relation error that is well‐established in the context of the finite element method. It is then shown that the resulting stopping criterion leads to a better control of the global error on the computed effective property compared to the classical criterion based on the residual of the iterative scheme alone.

Computationally efficient and sub-optimal trajectory planning framework based on trajectory-quality growth rate analysis.

A planetary exploration rover has been used for scientific missions or as a precursor for a future manned mission. The rover's autonomous system is managed by a space-qualified, radiation-hardened onboard computer; hence, the processing performance for such a computer is strictly limited, owing to the limitation to power supply. Generally, a computationally efficient algorithm in the autonomous system is favorable. This study, therefore, presents a computationally efficient and sub-optimal trajectory planning framework for the rover. The framework exploits an incremental search algorithm, which can generate more optimal solutions as the number of iterations increases. Such an incremental search is subjected to the trade-off between trajectory optimality and computational burden. Therefore, we introduce the trajectory-quality growth rate (TQGR) to statistically analyze the relationship between trajectory optimality and computational cost. This analysis is conducted in several types of terrain, and the planning stop criterion is estimated. Furthermore, the relation between terrain features and the stop criterion is modeled offline by a machine learning technique. Then, using the criterion predicted by the model, the proposed framework appropriately interrupts the incremental search in online motion planning, resulting in a sub-optimal trajectory with less computational burden. Trajectory planning simulation in various real terrain data validates that the proposed framework can, on average, reduce the computational cost by 47.6% while maintaining 63.8% of trajectory optimality. Furthermore, the simulation result shows the proposed framework still performs well even though the planning stop criterion is not adequately predicted.

Robust design optimisation under lack-of-knowledge uncertainty

Design optimization is common practice in engineering where the goal is to find the optimal combination of design parameters under prescribed constraints. However, some parameters may be impossible to define in a deterministic sense and may only be known with significant uncertainty. This limitation has led to an alternative definition of design optimality called robustness, where attention is payed to the variation around the optimal performance. Straightforward methods to solve robust optimization problems are usually limited in two ways: (1) the computation burden of the so-called ‘double-loop’ optimization problem hinders application to realistic models, and (2) the formalisms are typically limited to probabilistic descriptions of the uncertainty. This paper presents a formulation of the robust optimization problem under interval uncertainty and proposes a new approach taking advantage of the so-called adaptive Gaussian processes to solve it efficiently. The proposed surrogate approach mitigates the computational burden of the resolution, and a dedicated learning function is proposed to ensure iterative minimization of the surrogate modelling error and convergence towards the robust optimum. The algorithm uses a stopping criterion related to the level of confidence associated with the optimality of the solution. The approach is illustrated on six analytical and engineering benchmark problems.

Optimization of Tuned Mass Damper for Submerged Floating Tunnel with Frequency-Domain Dynamics Simulation

In this study, the Tuned Mass Damper (TMD) optimization is carried out to reduce the resonant motion of Submerged Floating Tunnel (SFT) under wave excitations. The SFT dynamics is evaluated in frequency domain; a new approach to cost-effectively optimizing TMD parameters for a moored system is suggested. Discrete-Module-Beam (DMB) method is used to model the Tunnel; mooring lines are included as equivalent stiffness matrix through static-offset tests by the fully coupled model. Since the frequency-domain dynamics simulation model is employed, a significant reduction in optimization time can be achieved. TMD is installed at the tunnel’s mid-length to mitigate the lateral motion of the Tunnel and coupled with the Tunnel with translational and rotational springs and dampers. The optimization process for TMD parameters is performed through the Genetic Algorithm (GA). The GA generates the TMD mass and spring and damping coefficients. The dynamics simulation is performed under wave conditions and this process is repeated until the stopping criteria is satisfied. Results demonstrate that TMD with optimized parameters significantly reduces the lateral motion, especially near the system’s lowest lateral natural frequency. This frequency-domain optimization also works as intended with significantly decreased optimization time.

Nutrient Deficiency Classification in Rice Plants Using DenseNet121

The leaves of plants often display signs of nutritional shortages. Therefore, to identify the nutrient shortages in plants, the color as well as the shape of the leaves can be wielded. Image classification is a quick and efficient methodology for this diagnosis task. In image classification, even though Deep Convolutional Neural Networks (DCNNs) are successful, little emphasis has been paid to their use in detecting plant nutrient deficits. Thus, to classify rice plant nutrient deficits, a DenseNet121 model is proposed in this paper. This proposed technique includes inserting additional new layers, early stopping criteria, model checkpoints, and five-fold cross-validation to enhance the model’s accuracy. After that, the model’s efficacy has been assessed utilizing specific performance metrics like accuracy, F1 score, precision, and recall. The performance of the suggested model is also analogized with the newer deep learning algorithms. From experimental results, the modified DenseNet121 attained 99.98% of accuracy, 99.99% of Precision, 99.98% of Recall, and 99.97% of F1-score. Lastly, to classify nutrient deficiencies in rice plants automatically on the web and mobile devices, an application was created for the farmers.

Runoff Estimation Using Advanced Soft Computing Techniques: A Case Study of Mangla Watershed Pakistan

A precise rainfall-runoff prediction is crucial for hydrology and the management of water resources. Rainfall-runoff prediction is a nonlinear method influenced by simulation model inputs. Previously employed methods have some limitations in predicting rainfall-runoff, such as low learning speed, overfitting issues, stopping criteria, and back-propagation issues. Therefore, this study uses distinctive soft computing approaches to overcome these issues for modeling rainfall-runoff for the Mangla watershed in Pakistan. Rainfall-runoff data for 29 years from 1978–2007 is used in the study to estimate runoff. The soft computing approaches used in the study are Tree Boost (TB), decision tree forests (DTFs), and single decision trees (SDTs). Using various combinations of past rainfall datasets, these soft computing techniques are validated and tested for the security of efficient results. The evaluation criteria for the models are some statistical measures consisting of root means square error (RMSE), mean absolute error (MAE), coefficient of determination (R2), and Nash–Sutcliffe efficiency (NSE). The outcomes of these computing techniques were evaluated with the multilayer perceptron (MLP). DTF was found to be a more accurate soft computing approach with the average evaluation parameters R2, NSE, RMSE, and MAE being 0.9, 0.8, 1000, and 7000 cumecs. Regarding R2 and RMSE, there are about 57% and 17% of improvement in the results of DTF compared to other techniques. Flow duration curves (FDCs) were employed and revealed that DTF performed better than other techniques. This assessment revealed that DTF has potential; researchers may consider it an alternative approach for rainfall-runoff estimations in the Mangla watershed.

Hitting the target: stopping active learning at the cost-based optimum

Active learning allows machine learning models to be trained using fewer labels while retaining similar performance to traditional supervised learning. An active learner selects the most informative data points, requests their labels, and retrains itself. While this approach is promising, it raises the question of how to determine when the model is ‘good enough’ without the additional labels required for traditional evaluation. Previously, different stopping criteria have been proposed aiming to identify the optimal stopping point. Yet, optimality can only be expressed as a domain-dependent trade-off between accuracy and the number of labels, and no criterion is superior in all applications. As a further complication, a comparison of criteria for a particular real-world application would require practitioners to collect additional labelled data they are aiming to avoid by using active learning in the first place. This work enables practitioners to employ active learning by providing actionable recommendations for which stopping criteria are best for a given real-world scenario. We contribute the first large-scale comparison of stopping criteria for pool-based active learning, using a cost measure to quantify the accuracy/label trade-off, public implementations of all stopping criteria we evaluate, and an open-source framework for evaluating stopping criteria. Our research enables practitioners to substantially reduce labelling costs by utilizing the stopping criterion which best suits their domain.

Adaptive Timestepping, Linearization, and A Posteriori Error Control for Multiphase Flow of Immiscible Fluids in Porous Media with Wells

Summary This work focuses on the development of adaptive timesteps, stopping criteria, and error control strategies for reservoir simulations with fully implicit (FIM) solvers. Using a rigorous error control framework, an adaptive time selector combined with nonlinear stopping criteria is used to control nonlinear iterations as well as to balance accuracy and robustness for challenging nonlinear simulations. In reservoir simulation, efficiently solving a system of nonlinear equations arising from the FIM method can be computationally burdensome for complex recovery processes. Theoretically, an FIM reservoir simulator has no stability limit on the timestep size. In practice, standard Newton’s method often fails to converge for large timestep sizes and must therefore cut the timestep multiple times to achieve convergence, resulting in a large number of unnecessary iterations. Another cause of nonlinear convergence issues is the presence of wells, which are often presented as singular point/line sources strongly coupled to the reservoir model, posing additional restrictions on the timestep choice. Here, we use a posteriori error estimators to avoid unnecessary nonlinear iterations and timestep cuts when solving immiscible multiphase flow. First, we estimate error components (e.g., spatial, temporal, and nonlinear) and then apply these to balancing criteria, providing us with dynamic and adaptive strategies to control timestep and nonlinear iterations. The error estimators are fully and locally computable, inexpensive to use, and target the various error components, including well singularities. The method provides an adaptive criterion for stopping the nonlinear iteration process whenever the linearization error does not significantly affect the overall error. Simultaneously, timesteps are adapted to maintain a constant size of the temporal discretization error with respect to the total error. Altogether, this avoids using unnecessary linearization iterations, wasteful timestep cuts, and too small timesteps. To demonstrate the effectiveness of these adaptive features, we present results for a suite of cases, covering both standard benchmarks and conceptual problems incorporating highly heterogeneous media with multiple wells. The proposed timestep selector cooperates with the new stopping criteria to improve nonlinear solver performance and increases robustness for cases with high nonlinearity. Perhaps most important, the adaptive features ensure balanced temporal and spatial errors while maintaining sufficiently small nonlinear errors, which ensures solution accuracy by accurately reproducing saturation fronts, production plateau, and breakthrough times.

An algorithm-independent measure of progress for linear constraint propagation

Propagation of linear constraints has become a crucial sub-routine in modern Mixed-Integer Programming (MIP) solvers. In practice, iterative algorithms with tolerance-based stopping criteria are used to avoid problems with slow or infinite convergence. However, these heuristic stopping criteria can pose difficulties for fairly comparing the efficiency of different implementations of iterative propagation algorithms in a real-world setting. Most significantly, the presence of unbounded variable domains in the problem formulation makes it difficult to quantify the relative size of reductions performed on them. In this work, we develop a method to measure—independently of the algorithmic design—the progress that a given iterative propagation procedure has made at a given point in time during its execution. Our measure makes it possible to study and better compare the behavior of bounds propagation algorithms for linear constraints. We apply the new measure to answer two questions of practical relevance: (i) We investigate to what extent heuristic stopping criteria can lead to premature termination on real-world MIP instances. (ii) We compare a GPU-parallel propagation algorithm against a sequential state-of-the-art implementation and show that the parallel version is even more competitive in a real-world setting than originally reported.

Data-driven clustering of infectious disease incidence into age groups.

Understanding the patterns of infectious diseases spread in the population is an important element of mitigation and vaccination programs. A major and common characteristic of most infectious diseases is age-related heterogeneity in the transmission, which potentially can affect the dynamics of an epidemic as manifested by the pattern of disease incidence in different age groups. Currently there are no statistical criteria of how to partition the disease incidence data into clusters. We develop the first data-driven methodology for deciding on the best partition of incidence data into age-groups, in a well defined statistical sense. The method employs a top-down hierarchical partitioning algorithm, with a stopping criteria based on multiple hypotheses significance testing controlling the family wise error rate. The type one error and statistical power of the method are tested using simulations. The method is then applied to Covid-19 incidence data in Israel, in order to extract the significant age-group clusters in each wave of the epidemic.

Physicians' attitudes and perceptions of neuromuscular blocker infusions in ARDS

PurposeThe perceptions and practices of ICU physicians regarding initiating neuromuscular blocker infusions (NMBI) in acute respiratory distress syndrome (ARDS) may not be evidence-based amidst the surge of severe ARDS during the SARS-CoV-2 pandemic and new practice guidelines. We identified ICU physicians' perspectives and practices regarding NMBI use in adults with moderate-severe ARDS. Materials and methodsAfter extensive development and testing, an electronic survey was distributed to 342 ICU physicians from three geographically-diverse U.S. health systems(n = 12 hospitals). ResultsThe 173/342 (50.5%) respondents (75% medical) somewhat/strongly agreed a NMBI should be reserved until: after a trial of deep sedation (142, 82%) or proning (59, 34%) and be dose-titrated based on train-of-four monitoring (107, 62%). Of 14 potential NMBI risks, 2 were frequently reported to be of high/very high concern: prolonged muscle weakness with steroid use (135, 79%) and paralysis awareness due to inadequate sedation (114, 67%). Absence of dyssychrony (93, 56%) and use ≥48 h (87, 53%) were preferred NMBI stopping criteria. COVID-19 + ARDS patients were twice as likely to receive a NMBI (56 ± 37 vs. 28 ± 19%, p < 0.01). ConclusionsMost intensivists agreed NMBI in ARDS should be reserved until after a deep sedation trial. Stopping criteria remain poorly defined. Unique considerations exist regarding the role of paralysis in COVID-19+ ARDS.

Sequential optimization method based on the adaptive Kriging model for the possibility-based design optimization

The possibility-based design optimization (PBDO) model under the fuzzy uncertainty can provide the optimal design parameters by taking a trade-off between the performance and security. In order to efficiently solve the PBDO model with implicit failure possibility constraints in engineering, a Sequential Optimization Method based on the Adaptive Kriging model (AK-SOM) is proposed in this paper. The AK-SOM firstly constructs the Kriging model of the performance function corresponding to each failure possibility constraint. The optimized design parameters of the PBDO based on the current Kriging model (It is called the current K-PBDO) can be obtained without any additional model evaluations. Then, the effective possibility constraints in the current K-PBDO can be identified by the effective constraint criterion proposed in this paper, and only the Kriging models of the effective constraints are updated. The sequential optimization process continues until both the relative error stopping criterion and the U-function stopping criterion are satisfied simultaneously. The effectiveness and superiority of the AK-SOM are verified by five examples, and the obtained experimental results indicate that the proposed AK-SOM has no restrictions on the expression of the performance function, and it can effectively improve the efficiency of solving the PBDO model while ensuring the accuracy. Additionally, the proposed method is employed in the PBDO of a GH4169 aero-engine turbine disk, and the optimal scheme of PBDO shows that the failure possibility of the disk is reduced and the maximum stress in the dangerous part is also greatly reduced.

Recursive Convex Model for Optimal Power Flow Solution in Monopolar DC Networks

This paper presents a new optimal power flow (OPF) formulation for monopolar DC networks using a recursive convex representation. The hyperbolic relation between the voltages and power at each constant power terminal (generator or demand) is represented as a linear constraint for the demand nodes and generators. To reach the solution for the OPF problem a recursive evaluation of the model that determines the voltage variables at the iteration t+1 (vt+1) by using the information of the voltages at the iteration t (vt) is proposed. To finish the recursive solution process of the OPF problem via the convex relaxation, the difference between the voltage magnitudes in two consecutive iterations less than the predefined tolerance is considered as a stopping criterion. The numerical results in the 85-bus grid demonstrate that the proposed recursive convex model can solve the classical power flow problem in monopolar DC networks, and it also solves the OPF problem efficiently with a reduced convergence error when compared with semidefinite programming and combinatorial optimization methods. In addition, the proposed approach can deal with radial and meshed monopolar DC networks without modifications in its formulation. All the numerical implementations were in the MATLAB programming environment and the convex models were solved with the CVX and the Gurobi solver.