Search

Approximate Computing is a paradigm used by researchers as alternative to the diminishing of the evolution of hardware performance in the ongoing race for computational throughput in HPC. Precision reduction and mixed precision are the most studied among the existing techniques. In addition, some NVIDIA GPUs have Tensor Core architecture to speed up some classes of algorithms, such as matrix multiplication. This study aims to apply Approximate Computing techniques, like mixed precision, in matrix multiplication and stencil algorithms using OpenACC directives and cuTensor library to analyze performance gains versus accuracy losses. Results showed that it was possible to obtain a speedup of 16.60× with an optimized matrix multiplication algorithm present in the matmul intrinsic function using 16-bit floating-point data with Tensor Core, compared to a naive version using 64-bit floating-point. For this same case, accuracy loss went from 10−26 up to 10−1, approximately. For the stencil algorithm, it was possible to obtain a gain of 1.60× by only reducing variables precision from 64-bit to 16-bit floating-point, with accuracy loss from 0 to 10−9, for 300 iterations.

Instruments for monitoring severe meteorological phenomena (such as lightning, flooding and landslides) can be used to assist in decision-making by state agencies, in an attempt to mitigate their possible harmful effects. These phenomena usually occur suddenly on a short-term duration, under a limited region, imposing difficulties in being predicted by regular weather forecast models, requiring specific prediction systems. Very short-term weather forecasting systems, on order of a few hours, known as nowcasting, can include numerical models of physical phenomena and machine learning algorithms. This work presents a system for forecasting the incidence of lightning, a common phenomenon in electrically active storms, through the application and evaluation of two machine learning models, an Artificial Neural Network and a Random Forest model, which were able to detect the occurrence of atmospheric electrical discharges from the automatic recognition of patterns obtained from the data generated by the numerical weather forecasts. The Random Forest model presented the best results when trained with the set that includes the ten best correlated variables, reaching 99.77% of accuracy for the case study performed.

The conservation of tropical forests is urgent and necessary due to the important role they play in the global ecosystem. Several governmental and private initiatives were created to detect deforestation in tropical forests through analyses of remote sensing images, which demands skilled labor and different ways to deal with a great amount of data. Citizen Science could be used to mitigate these challenges, as it consists of nonspecialized volunteers collecting, analyzing, and classifying data to solve technical and scientific problems. In this sense, this work proposes the ForestEyes Project 1, which aims to combine citizen science and machine learning for deforestation detection. The volunteers classify remote sensing images, and these data are used as the training set for classification algorithms. The volunteers classified more than 5, 000 tasks from remote sensing images of the Brazilian Legal Amazon, and the results were compared to a groundtruth from the Amazon Deforestation Monitoring Project PRODES. The volunteers achieved good labeling of the remote sensing data, even for recent deforestation tasks, building high-confidence labeled collections as they selected the most relevant samples and discarded noisy segments that might disrupt machine learning techniques. Finally, the proposed methodology is promising, and with improvements, it could be able to generate complementary information to official monitoring programs or even generate information for areas not yet monitored.

Large-scale commercial bioprocesses that manufacture biopharmaceutical products such as monoclonal antibodies generally involve multiple bioreactors operated in parallel. Spectra recorded during in situ monitoring of multiple bioreactors by multiplexed fiber-optic spectroscopies contain not only spectral information of the chemical constituents but also contributions resulting from differences in the optical properties of the probes. Spectra with variations induced by probe differences cannot be efficiently modeled by the commonly used multivariate linear calibration models or effectively removed by popular empirical preprocessing methods. In this study, for the first time, a calibration model is proposed for the analysis of complex spectral data sets arising from multiplexed probes. In the proposed calibration model, the spectral variations introduced by probe differences are explicitly modeled by introducing a multiplicative parameter for each optical probe, and then their detrimental effects are effectively mitigated through a "dual calibration" strategy. The performance of the proposed multiplex calibration model has been tested on two multiplexed spectral data sets (i.e., MIR data of ternary mixtures and NIR data of bioprocesses). Experimental results suggest that the proposed calibration model can effectively mitigate the detrimental effects of probe differences and hence provide much more accurate predictions than commonly used multivariate linear calibration models (such as PLS) with and without empirical data preprocessing methods such as orthogonal signal correction, standard normal variate, or multiplicative signal correction.

There is considerable interest in exploiting the novel physical and biological properties of microbial exopolysaccharides in industry and medicine. For economic and scientific reasons, large scale production under carefully monitored and controlled conditions is required. Producing exopolysaccharides in industrial fermenters poses several complex bioengineering and microbiological challenges relating primarily to the very high viscosities of such culture media, which are often exacerbated by the producing organism's morphology. What these problems are, and the strategies for dealing with them are discussed critically in this review, using pullulan, curdlan, xanthan, and fungal β-glucans as examples of industrially produced microbial exopolysaccharides. The role of fermenter configuration in their production is also examined.

Many of the foods we eat are of microbial origin or contain constituents produced by microbial fermentation. This chapter considers the biotechnological problems associated with growing these producing strains, most of which are aerobic organisms, in large scale fermenters. It discusses how rational fermenter design is a crucial component in providing them with the physical environment for optimal desirable metabolite production, which with the filamentous fungi is often related to their morphological form. It also discusses how some of the problems in achieving this can be ameliorated by innovative fermenter configuration. Carefully selected examples are used to illustrate the practical application on an industrial scale of these philosophies and suggestions are made about how fermentation technology is likely to change in the future to satisfy the increasing demands for highly nutritious and safe foods for human consumption.

A possible approach for modeling two-dimensional convection-diffusion problems in a cell-centered scheme with an unstructured triangular grid is the use of the Circumcenter, that is the center of the circumference that passes through the vertices of the triangular volume. This point is used to calculate all variables involved in the numerical simulation, and a Finite Volume Method was use to discretize the equations of an Incompressible Viscous Flow. This work analyzes classical problems of bidimensional flow, such as the inlet region of a Poiseuille flow, lid-driven cavity, backward-facing step and free convection with Boussinesq approximation. The application of the method has been shown to be a simple and flexible scheme and the results fit the analytical, experimental or numeric data presented in the literature.

Controlling the dissolved oxygen (DO) in the fed-batch culture of the medicinal mushroom Ganoderma lucidum led to a two-fold increase of the maximum biomass productivity compared to uncontrolled DO conditions. By contrast, extracellular polysaccharide (EPS) production was two times higher under oxygen limitation (uncontrolled DO) than under increased oxygen availability (controlled DO). Morphologically, dispersed mycelium was predominant under controlled DO conditions, with highly branched hyphae, consistent with the enhanced culture growth noted under these conditions. While in the uncontrolled DO processes mycelial clumps were the most common morphology throughout the culture. However, in both cultures clamp connections were found. This is an exciting new finding, which widens the applicability of this basidiomycete in submerged fermentation. In rheological terms, broths demonstrated shear-thinning behaviour with a yield stress under both DO conditions. The flow curves were best described by the Herschel-Bulkley model: flow index down to 0.6 and consistency coefficient up to 0.2 and 0.6 Pa sn in uncontrolled and controlled cultures DO, respectively. The pseudoplastic behaviour was entirely due to the fungal biomass, and not to the presence of EPS (rheological analysis of the filtered broth showed Newtonian behaviour). It is clear that dissolved oxygen tension is a critical process parameter that distinctly influences G. lucidum morphology and rheology, affecting the overall performance of the process. This study contributes to an improved understanding of the process physiology of submerged fermentation of G. lucidum.

A demanda constante para melhorar a qualidade de previsões numéricas do tempo obriga o uso de computadores progressivamente mais potentes. Com a popularização de processadores “multicore”, o uso de sistemas com muitas centenas de processadores tornou-se economicamente viável. Este trabalho investiga como escalar o paralelismo de uma aplicação operacional para a previsão do tempo que executa eficientemente em muitas dezenas de processadores para máquinas com muitas centenas de processadores. A investigação determinou as limitações da aplicação, suas surpreendentes causas e permitiu desenvolver uma solução eficiente que atinge a escala de paralelismo desejado.

Publisher Summary This chapter evaluates recent advances in submerged liquid cultivation of the vegetative or mycelial forms of those members of the “higher fungi” that produce macroscopic spore-bearing structures, namely the macrofungi. The chapter focuses on those producing potentially useful natural products and novel compounds with interesting biological activities by means of submerged culture fermentation processes. The taxonomy of the fungi is in a state of rapid flux arising from the widespread use of molecular genetics to characterize and classify the organisms, and many of earlier ideas on their phylogenetic relationships must be revised. Fungi can be defined as being eukaryotic, characteristically mycelial organisms, with chitin-based cell walls, an absorptive nutrition, and which reproduce by production of spores. Some of the macrofungi are edible, and some also have biotechnological and environmental applications. Edible macrofungi have long been consumed both as valuable protein and energy sources, and also for improving human health and longevity.

This work aims at validating the CATT-BRAMS model which can simulate emissions and transport of aerosol and trace gases from forest fires occurred in the Iberian Peninsula on the period from 7 to 12 August 2003. The validation of the generated aerosol concentrations values was made by comparison with AERONET (Aerosol Robotic Network) aerosol optical thickness (AOT) measurements at two different wavelengths, which is a quantity that is related to the aerosol load in the atmospheric column.

The Pichia pastoris expression system is being used successfully for the production of various recombinant heterologous proteins. Recent developments with respect to the Pichia expression system have had an impact on not only the expression levels that can be achieved, but also the bioactivity of various heterologous proteins. We review here some of these recent developments, as well as strategies for reducing proteolytic degradation of the expressed recombinant protein at cultivation, cellular and protein levels. The problems associated with post-translational modifications performed on recombinant proteins by P. pastoris are discussed, including the effects on bioactivity and function of these proteins, and some engineering strategies for minimizing unwanted glycosylations. We pay particular attention to the importance of optimizing the physicochemical environment for efficient and maximal recombinant protein production in bioreactors and the role of process control in optimizing protein production is reviewed. Finally, future aspects of the use of the P. pastoris expression system are discussed with regard to the production of complex membrane proteins, such as G protein-coupled receptors, and the industrial and clinical importance of these proteins.

Background: Novel analytical tools, which shorten the long and costly development cycles of biopharmaceuticals are essential. Metabolic flux analysis (MFA) shows great promise in improving our understanding of the metabolism of cell factories in bioreactors, but currently only provides information post-process using conventional off-line methods. MFA combined with real time multianalyte process monitoring techniques provides a valuable platform technology allowing real time insights into metabolic responses of cell factories in bioreactors. This could have a major impact in the bioprocessing industry, ultimately improving product consistency, productivity and shortening development cycles. Results: This is the first investigation using Near Infrared Spectroscopy (NIRS) in situ combined with metabolic flux modelling which is both a significant challenge and considerable extension of these techniques. We investigated the feasibility of our approach using the industrial workhorse Pichia pastoris in a simplified model system. A parental P. pastoris strain (i.e. which does not synthesize recombinant protein) was used to allow definition of distinct metabolic states focusing solely upon the prediction of intracellular fluxes in central carbon metabolism. Extracellular fluxes were determined using off-line conventional reference methods and on-line NIR predictions (calculated by multivariate analysis using the partial least squares algorithm, PLS). The results showed that the PLS-NIRS models for biomass and glycerol were accurate: correlation coefficients, R 2 , above 0.90 and the root mean square error of prediction, RMSEP, of 1.17 and 2.90 g/L, respectively. The analytical quality of the NIR models was demonstrated by direct comparison with the standard error of the laboratory (SEL), which showed that performance of the NIR models was suitable for quantifying biomass and glycerol for calculating extracellular metabolite rates and used as independent inputs for the MFA (RMSEP lower than 1.5 × SEL). Furthermore, the results for the MFA from both datasets passed consistency tests performed for each steady state, showing that the precision of on-line NIRS is equivalent to that obtained by the off-line measurements. Conclusions: The findings of this study show for the first time the potential of NIRS as an input generating for MFA models, contributing to the optimization of cell factory metabolism in real-time.