223 research outputs found
Metamaterial-based Toraldo pupils for super-resolution at millimetre wavelengths
No full textUsing the long-established Cardiff metal-mesh filter technology, we have exploited our ability to artificially manipulate the phase of a wavefront across a device in order to produce a dielectric-based Toraldo pupil working at millimeter wavelengths. The use of a Toraldo pupil to push the angular resolution of an optical imaging system beyond the classical diffraction limit is yet to be realized in the millimeter regime, but is an exciting prospect. Here we present the design and measured performance of a prototype Toraldo pupil, based on a 5 annuli design
Dipole-flow disturbed by a circular inclusion of conductivity different from the background: From deterministic to a self-consistent analytical solution
No full textSteady dipole-flow through a porous medium, and disturbed by a circular inclusion Omega(0) of conductivity different from the background, is solved analytically. The solution is achieved by means of the circle theorem, which is reformulated to account for the entry/leave of mass and energy through the boundary partial derivative Omega(0). It is shown that the governing potential is that which one would consider in absence of the disturbance supplemented with an ad hoc (fictitious) dipole laying inside Omega(0). Besides the theoretical interest, the analytical solution is used to compute the effective conductivity K-eff, by means of the self-consistent approximation. Overall, K-eff is found to depend upon the flow configuration, and therefore it cannot be sought as a medium's property (nonlocality). In particular, K(eff )depends upon the joint probability density function f of the conductivity and the distribution/size of the inclusions. Results, analyzed for a fairly general model of f, demonstrate that the coefficient of correlation rho between the involved random fields is the key parameter characterizing the structure of K-eff. Indeed, the latter results larger or smaller than that of the background, depending on whether rho is negative or positive, respectively. For rho = 0, the effective conductivity is a local property and, in this case, one can apply the superposition principle with the homogeneous conductivity replaced by the geometric mean
Optimization techniques for large scale finite sum problems
No full textWith the explosion of machine learning and artificial intelligence applications, the need for optimization methods specialized in the training of such models has been steadily growing for the last 10-20 years. Indeed, given the big data regime and the special structure of the optimization problems to be solved in these settings, a number of new, efficient optimization methods have been developed. A large amount of these new methods strongly rely on the finite sum structure of the objective function to be minimized, where the indices i=1,...,N often refer to the availability of N input-output pairs on which the model should be trained, i.e. the training set. Nevertheless, this is not the only application where a finite sum structure of the objective function appears. Indeed, beyond the training of Neural Networks (NN) and Support Vector Machines (SVM), which depend by definition on a dataset of input-output pairs, a finite sum structure can also be recognized in Reinforcement Learning (RL) applications, due to the need of estimating expected values by sample approximation. In all these cases, N is usually huge, in the order of millions, or even billions, therefore making the exact computation of the function and gradient infeasible for many real life applications. This is one of the reasons why the field has seen a flourishing of publications from the most diverse communities, beyond the operations research one, for example the dynamical control, computer science, stochastic optimization ones. Many new methods have been developed by these communities, both deterministic and stochastic algorithms, although their comparison is made difficult by the different approaches coming from the different communities the new algorithms belong to. Due to the above considerations, the focus of this dissertation is on how to solve optimization problems where the function is structured as a finite sum of component functions. In this finite sum setting, a function fi can be referred to as a component function, and its gradient Ñfi as a component gradient. In particular, a deep investigation of the algorithms developed so far to solve such problems is carried on, with a specific interest in showing the similarities and differences of the convergence analysis when it is developed in the deterministic vs stochastic cases. The target of the investigation is the case when the component gradients are continuously differentiable, and easily computable, like in many machine learning settings (e.g., neural networks training). In this framework, dynamic minibatching schemes are addressed. These are employed to determine the size of the sample to be used during the optimization process, especially in gradient-based methods, when the gradient is estimated by subsampling the component gradients, namely, when it is estimated based on a subset of the indices 1,...,N. The aim of dynamic minibatching schemes is to dynamically test the quality of the gradient approximation, and consequently suggest if the sample size should grow or not. A new technique is proposed, based on statistical analysis of the gradient estimates. The new technique is based on the well-known Analysis of Variance (ANOVA) test, and the convergence of a subsampled gradient-based method is proved when such technique is employed. Numerical experiments are reported on standard machine learning tasks, like (nonlinear) regression and binary classification. Then, the derivative free setting is explored, i.e. the setting where the component functions come from a black-box-like process and the component gradients are not directly available. An example of such setting is policy optimization for reinforcement learning, where only sample approximations of the stochastic reward function are available. Therefore, in literature, Derivative Free Optimization (DFO) methods have been applied to solve this problem, in particular by trying to estimate the gradient by computing only sample approximations of the function. An analysis of the convergence guaran4 tees of stochastic optimization methods in this setting is performed, showing that approximating the gradient by only computing sample-based estimates of the function brings a further approximation error, leading to poorer theoretical results. The special case of policy optimization for reinforcement learning is analysed, showing that such application is even harder, since the sample approximation of the function, in general, does not have continuity guarantees. Finally, a new class of distributed algorithms is introduced to solve linearly constrained, convex problems, with potential application to the dual formulation of the support vector machines training problem. This employs augmented Lagrangian and primal-dual theory to develop a simple, distributable and parallelizable class of algorithms to solve convex problems with simple bound and hard (i.e. coupling all the variables), linear constraints. Such class of algorithms is of particular interest for training support vector machines, since it allows to fully distribute the data, i.e. the input-output pairs, to the available parallel processes, simplifying the (often infeasible) storage of such large amount of data
In tema di strutture reticolari per coperture di grandi luci nel costruito storico e monumentale
No full textLe strutture reticolari sono un esempio di “storia nella storia” dell’evoluzione dei metodi di calcolo delle strutture. Non a caso G. Colonnetti dedicava l’intero secondo volume su “La storia delle Costruzioni” (1932) alle travature reticolari isostatiche e iperstatiche, con tale ampiezza e profondità da farne, più che un’applicazione tra le altre, il modello privilegiato di ogni applicazione strutturale.
Le prime trattazioni semplificate degli ingegneri tedeschi J.W. Schwelder e A. Ritter, il grande trattato di Müller-Breslau, la celebre opera di C. Giudi, come le Abbondlunggen di Mohr, sono prove di dibattito profondo, animato dalle circostanze del progresso e delle costruzioni ardite dell’epoca (1851-1932). Sino a cinquanta-sessanta anni fa, nei testi di scienza e tecnica delle costruzioni più diffusi il tema della travatura reticolare campeggiava come argomento di preminente e sovrastante interesse
Italian normative data for the original version of the Tower of London test: a bivariate analysis on speed and accuracy scores.
No full textThe Tower of London (ToL) test is traditionally used to assess strategical reasoning, problem-solving, and mental planning in clinical populations. Here, we provide the Italian standardization norms for the original, 12-problem version of the ToL test. The performance of 216 Italian individuals ranging 18 to 89 in age was scored in terms of both Time (Speed) and Accuracy—the time, and the number of attempts, necessary to find a solution. We performed univariate analyses on separate Time and Accuracy scores, using Age in years, Education in years, and Sex (male vs. female) as predictors. z scores
and equivalent scores were provided. Moreover, we performed a bivariate analysis for the assessment of individuals’ performance in terms of Time and Accuracy simultaneously. This standardization allows clinicians to use the original, most widespread version of ToL with the Italian population, thus optimizing comparability with other clinical and experimental research worldwide. Critically, this article offers a new statistical perspective on how Time and Accuracy scores, which are typically related to each other, can be combined to obtain a single, consistent clinical categorization that captures most of
the information contained in the patient’s performance
TRAVELLING COMPANIONS. ADOLFO NATALINI, CRISTIANO TORALDO DI FRANCIA AND SUPERSTUDIO
No full textThe essay deals with the figures of Adolfo Natalini and Cristiano Toraldo of France and their role in the foundation and construction of the Florentine avant-garde group Superstudio. The text is included in the catalogue of the exhibition Superstudio Migrazioni, organized by CIVA in Brussels in 2020, of which the author was a member of the scientific committee
COMPAGNONS DE VOYAGE: ADOLFO NATALINI, CRISTIANO TORALDO DI FRANCIA ET SUPERSTUDIO
No full textThe essay deals with the figures of Adolfo Natalini and Cristiano Toraldo of France and their role in the foundation and construction of the Florentine avant-garde group Superstudio. The text is included in the catalogue of the exhibition Superstudio Migrazioni, organized by CIVA in Brussels in 2020, of which the author was a member of the scientific committee
Easy quantitative methodology to assess visual-motor skills
No full textMatteo Chiappedi,1 Alessio Toraldo,2 Silvia Mandrini,3 Federica Scarpina,2 Melissa Aquino,2 Francesca Giulia Magnani,2 Maurizio Bejor31Don Carlo Gnocchi ONLUS Foundation, Milan, Italy; 2University of Pavia, Department of Psychology, Pavia, Italy; 3University of Pavia, Department of Surgical, Resuscitative, Rehabilitative and Transplant Sciences, Pavia, ItalyIntroduction: Visual-motor skills are the basis for a great number of daily activities. To define a correct rehabilitation program for neurological patients who have impairment in these skills, there is a need for simple and cost-effective tools to determine which of the visual-motor system levels of organization are compromised by neurological lesions. In their 1995 book, The Visual Brain in Action (Oxford: Oxford University Press), AD Milner and MA Goodale proposed the existence of two pathways for the processing of visual information, the “ventral stream” and “dorsal stream,” that interact in movement planning and programming. Beginning with this model, our study aimed to validate a method to quantify the role of the ventral and dorsal streams in perceptual and visual-motor skills.Subjects and methods: Nineteen right-handed healthy subjects (mean age 22.8 years ± 3.18) with normal or corrected-to-normal vision were recruited. We proposed that a delayed pointing task, a distance reproduction task, and a delayed anti-pointing task could be used to assess the ventral stream, while the dorsal stream could be evaluated with a grasping task and an immediate pointing task. Performance was recorded and processed with the video-analysis software Dartfish ProSuite.Results: Results showed the expected pattern of predominance of attention for the superior left visual field, predominance of the flexor tone in proximal peri-personal space arm movements, tendency toward overestimation of short distances, and underestimation of long distances.Conclusion: We believe that our method is advantageous as it is simple and easily transported, but needs further testing in neurologically compromised patients.Keywords: dorsal stream, ventral stream, visual-motor skills, rehabilitation, neurological disorder
109. Coppola G., Auricchio G, Romagnolo G, Licciardi F, Toraldo C, Pascotto A. Levetiracetam as add-on therapy in children, adolescents, and young adults with refractory epilepsy: an open trial.
No full text- …
