20 research outputs found
Theano: new features and speed improvements
Get PDFAbstract Theano is a linear algebra compiler that optimizes a user's symbolically-specified mathematical computations to produce efficient low-level implementations. In this paper, we present new features and efficiency improvements to Theano, and benchmarks demonstrating Theano's performance relative to Torch7, a recently introduced machine learning library, and to RNNLM, a C++ library targeted at recurrent neural networks
The application of differentiable programming frameworks to computational fluid dynamics
No full textIn recent years many automatic differentiable programming frameworks have been developed in which numerical programs can be differentiated through automatic differentiation (AD). Examples of these frameworks are Theano, TensorFlow and Pytorch. These frameworks are widely used in Machine Learning. AD also finds applications in the field of computational fluid dynamics (CFD). It is used to develop discrete adjoint CFD code for research concerning for instance sensitivity analysis, data assimilation and design optimization. However, the use of the automatic differentiable programming frameworks in the field of CFD is limited. One can find some examples in the literature on how to find a numerical solution to an initial value problem using a differentiable programming framework. In this work it will be clarified how one can implement an semiimplicittime integration scheme for a staggered grid to simulate the propagation of long waves in water with a free surface in TensorFlow. A main advantage of the automatic differentiable programming frameworks is the user friendly applicationprogramming interface (API) for AD. No research has been conducted to use this API in the field of CFD. In this work an example will be given how one can use TensorFlow for research concerning sensitivity analysis. AD requires a significant allocation of memory on a CPU/GPU when working with fine meshes and/or long simulations and since CPU/GPU memory is finite, the method checkpointing isproposed to make it feasible to perform sensitivity analysis when working with fine meshes and/or long simulations. Another main advantage of the differentiable programming framework TensorFlow is the use of compute unified device architecture (CUDA) of a NVIDIA GPU in order to perform computations in parallel, which results in a significant reduction in computation time. A Benchmark will be given that indicates the computational efficiency of TensorFlow compared to a loop over grid implementation in NumPy and a Fortran CPU scalar implementation.Applied Mathematic
Tensile mechanical performance of electron-beam welded joints from aluminum alloy (Al-Mg-Si) 6156
No full textAbstractThe mechanical behavior of both, reference and electron beam welded aluminum alloy 6156 specimens was experimentally investigated. Sheets of AA6156 were artificially aged before and after the welding process and tensile specimens were machined from the welded sheets according to the ASTM E8 standard. The specimens were artificially aged at 170°C for different times that corresponded to all precipitation-hardening conditions, namely under-ageing (UA), peak-ageing (PA) and over-ageing (OA). The results showed that the effect of welding without any heat treatment (condition T4) decreases by, about 100 MPa, the yield stress and the yield strength, while the remaining elongation at fracture hardly exceeds 4 %. It was also shown that artificial ageing before welding increases the tensile ductility (almost 50 % joint efficiency in deformation) while the artificial ageing post to welding significantly increases the strength properties (more than 75 % joint efficiency in strength)
Effect of artificial aging on the mechanical performance of (Al-Cu) 2024 and (Al-Cu-Li) 2198 aluminum alloys
No full textAbstractAl-Cu-Li-Mg based alloys exhibit an excellent combination of low density, high elastic modulus and high specific strength and have already being used as structural materials for aerospace applications. The mechanical properties of these alloys are often associated with the addition of Li which enables the formation of several strengthening precipitates including δ΄ (Al3Li), θ΄ (Al2Cu), δ (AlLi) and T1 (Al2CuLi). Other precipitates have also been reported in these alloys that include GP zones, θ (Al2Cu), Ω (Al2Cu), S΄ (Al2CuMg), and β΄ (Al3Zr), e.g. [1-3]. Nevertheless, the quasi-static mechanical properties of AA2198 are scarcely reported in the open literature while quite few for the fracture toughness. For example, Chen et al. [4] performed tests on two different heat treated AA2198 (namely T351 and T851) and investigated their plastic and fracture behavior. Steglich et al. [5, 6] investigated experimentally and analytically the anisotropic deformation of AA2198-T8 occurred during mechanical loading with and without the presence of artificial notches. In a recent publication [7], a combination of transmission electron microscopy, atom probe tomography and high-energy X-ray diffraction was employed to investigate the influence of local microstructural changes on strengthening in AA2198 in different aging conditions.As AA2198 is supposed to replace AA2024 in aerostructures designed with the damage tolerance philosophy, the authors in the present work report and compare their tensile mechanical and fracture toughness behavior under different aging conditions to simulate the natural aging parameter. To this end, a comparison of both alloys of their tensile mechanical behavior is reported for different stages of aging, including conditions of under-aging (UA), peak-aging (PA) and over-aging (OA). Typical results of yield stress as well as elongation at fracture for the case of AA2024 can be seen in the diagrams of Figure 1. The effect of artificial aging on the fracture toughness is also assessed for both alloys. Although the peak-aged temper of AA2198 is of application interest, a detailed investigation of the structure-property relationships of other temper states is attempted since it provides the basis for understanding the influence of process-induced microstructural changes on the post-processing properties of the alloy. To this end, structural characterization as well as relationship between yield stress, elongation at fracture and fracture toughness is reported and discussed in the manuscript for all investigated aging conditions
'This last farewell to Cooke-ham here i give': The politics of home and nonhome in Aemilia Lanyer's 'the description of Cooke-ham'
No full textThis article examines some of the potentially feminist dimensions to Aemilia Lanyer's poetic narrative and her critique of patronage and social inequalities by paying particular attention to the politics of home and nonhome in 'The Description of Cooke-ham'. More specifically, it argues that the poet's ambivalent attitude towards Margaret and Anne Clifford and her legitimization of her poetic subjectivity are shaped by her mapping of what Theano Terkenli refers to in her critical discussion, 'Home As A Region', as 'the dialectical relationship between home and nonhome': this may be detected to lurk beneath the surface of the poem, a relationship that both forces and enables the poet to establish her gendered, social, and artistic identity in textual terms. Lanyer presents home, the lived experience of sharing Cookeham with the Cliffords, as a construct of memory which may be seen to have been formulated from the textual perspective of 'nonhome' in the aftermath of her departure from the Clifford estate. Finally, this article treats the poet's conviction that her true home is the text of her poem, since it is through poetry that Lanyer proposes a haven in which self-expression and spiritual growth are nurtured. Thus, although she may be materially dispossessed, she will never be rendered spiritually homeless. © The Author 2010. Published by Oxford University Press on behalf of the English Association; all rights reserved
Improving the odds of survival: transgenerational effects of infections
No full textRecent studies argue for a novel concept of the role of chromatin as a carrier of epigenetic memory through cellular and organismal generations, defining and coordinating gene activity states and physiological functions. Environmental insults, such as exposures to unhealthy diets, smoking, toxic compounds, and infections, can epigenetically reprogram germ-line cells and influence offspring phenotypes. This review focuses on intergenerational and transgenerational epigenetic inheritance in different plants, animal species and humans, presenting the up-to-date evidence and arguments for such effects in light of Darwinian and Lamarckian evolutionary theories. An overview of the epigenetic changes induced by infection or other immune challenges is presented, and how these changes, known as epimutations, contribute to shaping offspring phenotypes. The mechanisms that mediate the transmission of epigenetic alterations via the germline are also discussed. Understanding the relationship between environmental fluctuations, epigenetic changes, resistance, and susceptibility to diseases is critical for unraveling disease etiology and adaptive evolution. © The Author(s) 2025
Aristophanes and Euripides: A Palimpsestuous Relationship
Get PDFAristophanes allows Euripides to interrupt constantly. In Athenian comedy of the fifth century they are on stage together, both literally and figuratively. Despite Aristophanes’ comedies having a meaning of their own, Euripides’ lines are so clearly visible underneath them that they can only be described as the verbal equivalent of a palimpsest. The Oxford English Dictionary defines a palimpsest as a manuscript or piece of writing on which later writing has superimposed or effaced earlier writing, or something reused or altered but still bearing visible traces of its earlier form. It is clear that a palimpsest is the product of layering that results in something as new, whilst still bearing traces of the original. Dillon describes the palimpsest as “...an involuted phenomenon where otherwise unrelated texts are involved and entangled, intricately interwoven, interrupting and inhabiting each other”. Aristophanes takes texts, particularly those of Euripides, which may otherwise have been unrelated, and weaves them together to form something new.
I will show that in a number of cases Aristophanes offers scenes that have already been performed in Euripides’ plays but lays his own plot over the tragedian’s, whilst at the same time drawing the audiences’ attention to the original. The nature of this borrowing overwrites Kristeva’s theory of ‘intertextuality’ and provides a new and more apposite name for the permutation of texts in which the geno-text corresponds to infinite possibilities of palimpsestuous textuality (and the pheno-text to a singular text, which contains echoes of what it could have been). The plurality of Euripides’ texts, whilst engendering those of Aristophanes, constantly interrupts them. Through the consideration of ancient and modern literary theory and by a close analysis of Aristophanes’ and Euripides’ plays, this thesis sets out to offer a new reading of the relationship between these two poets. It shows that they were engaged in a dialogue of reciprocal influence that came to a head at the end of the Peloponnesian War
Nitrogen dioxide spatiotemporal variations in the complex urban environment of Athens, Greece
No full textIn this study, five years of total, tropospheric and near-surface nitrogen dioxide, ΝΟ2, observations from a Pandora spectrometer system routinely operating in the city center of Athens, Greece, are presented and compared with space-borne observations from the S5P/TROPOMI sensor and in-situ air quality monitoring station measurements from the Greek National Air Pollution Monitoring Network. Each of the three different systems exhibits its own monitoring capabilities and restrictions, based on the spatial representativeness and temporal resolution of its measurements. However, they all reveal a clear weekly pattern with low NO2 concentrations on Sundays while a diurnal cycle with higher levels of NO2 in the morning is observed from the ground due to the high NOx emissions from heavy traffic in the urban environment. A seasonal pattern is further demonstrated by both space- and ground-based remote sensing instruments with enhanced NO2 loadings in wintertime and decreased levels in summertime. Even though, as expected, S5P/TROPOMI appears to underestimate the tropospheric NO2 load, the correlation between the ground-based and space-borne data is relatively high, with corresponding correlation coefficients ranging between 0.73 in spring and 0.80 in winter. High correlations are also found between the relatively new near-surface NO2 product of Pandora and the in situ observations, ranging from 0.91 to 0.99. However, an underestimation is observed by the Pandora instrument compared to those in situ monitoring stations which are highly affected by the increased NOx emissions in the city center. For the Pandora near-surface NO2 at 0° (north) azimuthal viewing angle, the best agreement is found with the urban background in situ stations, with a negative bias at 0.99 ± 1.60 × 1011 molecules cm−3, whereas the Pandora retrievals at 39° (north-east) are closer to the urban in situ NO2 levels, with a bias of −1.10 ± 2.02 × 1011 molecules cm−3 on average. © 2023 The Author
Considerations for the value of three-dimensional printed (3DP) versus cadaveric specimens for anatomy education
No full textWe enjoyed reading the Brumpt et al. paper, which showed that a three-dimensional printed model (3DPM) of the ear was more effective than conventional cadaveric models for teaching anatomy. We would like to comment on the findings of this exciting study. In this case, the 3DPM of the ear was compared with dried bone models but not with a cadaveric specimen (with all adjacent soft tissues). The better results after the first test of students who used the 3DPMs were probably attributed to the optimized 3D representation of the ear anatomy. Also, the educational outcomes will likely be better if a more complex 3DPM is used, as it permits better visualization of the structures compared to the dried bone specimens. We certainly agree that 3DPMs have a remarkable ability to represent anatomy. Still, their effectiveness has not been proven superior to cadaveric specimens teaching complex anatomy. In conclusion, although we agree that 3DPMs have a high educational potential and can contribute to complex anatomy teaching, those models were not proven significantly more effective than cadaveric specimens in the Brumpt et al. study. The better effectiveness of 3DPMs compared to dried bone specimens (at the first test) does not mean those models are superior to specimens with retained soft tissues. Such cadaveric specimens permit visualization of complex structures and have proven valuable for teaching complex anatomy. Currently, the literature does not support the educational superiority of 3DPMs to those cadaveric specimens. © The Author(s), under exclusive licence to Springer-Verlag France SAS, part of Springer Nature 2024
Python for probability, statistics, and machine learning
No full textIntroduction -- Part 1 Getting Started with Scientific Python -- Installation and Setup -- Numpy -- Matplotlib -- Ipython -- Jupyter Notebook -- Scipy -- Pandas -- Sympy -- Interfacing with Compiled Libraries -- Integrated Development Environments -- Quick Guide to Performance and Parallel Programming -- Other Resources -- Part 2 Probability -- Introduction -- Projection Methods -- Conditional Expectation as Projection -- Conditional Expectation and Mean Squared Error -- Worked Examples of Conditional Expectation and Mean Square Error Optimization -- Useful Distributions -- Information Entropy -- Moment Generating Functions -- Monte Carlo Sampling Methods -- Useful Inequalities -- Part 3 Statistics -- Python Modules for Statistics -- Types of Convergence -- Estimation Using Maximum Likelihood -- Hypothesis Testing and P-Values -- Confidence Intervals -- Linear Regression -- Maximum A-Posteriori -- Robust Statistics -- Bootstrapping -- Gauss Markov -- Nonparametric Methods -- Survival Analysis -- Part 4 Machine Learning -- Introduction -- Python Machine Learning Modules -- Theory of Learning -- Decision Trees -- Boosting Trees -- Logistic Regression -- Generalized Linear Models -- Regularization -- Support Vector Machines -- Dimensionality Reduction -- Clustering -- Ensemble Methods -- Deep Learning -- Notation -- References -- IndexThis book, fully updated for Python version 3.6+, covers the key ideas that link probability, statistics, and machine learning illustrated using Python modules in these areas. All the figures and numerical results are reproducible using the Python codes provided. The author develops key intuitions in machine learning by working meaningful examples using multiple analytical methods and Python codes, thereby connecting theoretical concepts to concrete implementations. Detailed proofs for certain important results are also provided. Modern Python modules like Pandas, Sympy, Scikit-learn, Tensorflow, and Keras are applied to simulate and visualize important machine learning concepts like the bias/variance trade-off, cross-validation, and regularization. Many abstract mathematical ideas, such as convergence in probability theory, are developed and illustrated with numerical examples. This updated edition now includes the Fisher Exact Test and the Mann-Whitney-Wilcoxon Test.A new section on survival analysis has been included as well as substantial development of Generalized Linear Models. The new deep learning section for image processing includes an in-depth discussion of gradient descent methods that underpin all deep learning algorithms. As with the prior edition, there are new and updated *Programming Tips* that the illustrate effective Python modules and methods for scientific programming and machine learning. There are 445 run-able code blocks with corresponding outputs that have been tested for accuracy. Over 158 graphical visualizations (almost all generated using Python) illustrate the concepts that are developed both in code and in mathematics. We also discuss and use key Python modules such as Numpy, Scikit-learn, Sympy, Scipy, Lifelines, CvxPy, Theano, Matplotlib, Pandas, Tensorflow, Statsmodels, and Keras
