41199 research outputs found
Sort by
Lenti ortocheratologiche e lenti morbide multifocali a confronto per il controllo della progressione miopica
Get PDF- OBIETTIVO PRINCIPALE: Lo scopo di questo elaborato è paragonare le lenti a contatto ortocheratologiche e le lenti a contatto morbide multifocali per il controllo della progressione miopica. Nello specifico, capire le modalità in cui queste tipologie di lenti a contatto agiscono a livello oculare per rallentare la progressione miopica, e così rallentare l’evolversi di eventuali problemi gravi causati da un alto livello di miopia.
- METODI DI RICERCA: In questo elaborato sono stati analizzati articoli scientifici raccolti su siti di divulgazione scientifica e banche dati online, sono stati consultati documenti reperiti online e dispense fornite da docenti. Ho inoltre analizzato un case report reale di un’applicazione di una lente ortocheratologica in soggetto miope. Le parole chiave utilizzate sono state: ortocheratologia, lenti a contatto multifocali, miopia, progressione miopica.
- RISULTATI: Da un’attenta e accurata valutazione degli articoli scientifici, emerge che le lenti a contatto ortocheratologiche e multifocali sono efficaci per il trattamento della progressione miopica. Risultano essere i trattamenti contattologici reversibili più efficaci per la correzione dell’ametropia centrale e il controllo dello stato refrattivo periferico. Inoltre, a supporto dell’efficacia dell’ortocheratologia in termini di correzione della miopia, ho analizzato un case report reale, ottenendo un’ottima correzione della miopia associata ad astigmatismo con una lente a contatto ortocheratologica con zona ottica sferica.
- CONCLUSIONI: Nelle ultime decadi abbiamo assistito ad un aumento significativo dell’incidenza della miopia e della progressione miopica. Sebbene i fattori che influiscono sullo sviluppo e sulla progressione siano noti dal panorama scientifico, si sta ancora cercando di definire una linea guida generale per trattare al meglio i soggetti che presentano una progressione miopica.
Per meglio approfondire il comportamento delle lenti a contatto ortocheratologiche e lenti a contatto morbide multifocali per il controllo della progressione miopica, ho consultato numerosi articoli scientifici di interesse optometrico da me selezionati. Ho compreso che
le lenti a contatto ortocheratologiche e le lenti a contatto morbide multifocali risultano essere efficaci per rallentare la progressione miopica, ma per poter definire delle linee guida univoche sul modo di agire in questo caso, come sottolineano diversi autori, bisognerà poter disporre di più studi randomizzati di maggior durata e con più partecipanti
Effects of traditional and biodynamic farmyard manure application on soil properties
No full textIn this work, we analysed the effects of traditionally composted farmyard manure (FYM) applications with two types(Yarrow and All) of biodynamic additions that had been applied for over 27 years on the same trial soils.
Results showed that: a) repeated manure supplementation increased soil microbial biomass carbon; b) the addition of the six biodynamic preparations to manure counteracted the above effect; c) the addition of the single preparation from yarrow (Achillea millefolium) to manure did not affect the increase of soil microbial biomass carbon due to manure, and its mean value higher than that of manure alone, d) known microbial communications models from quorum sensing and other cell-to-cell signalling notions are consistent with the explanation of how the observed effects can be explained on a quantitative basis
Studio dell'effetto della temperatura della polvere sulla determinazione della luminosità totale lontano-infrarossa di galassie ad alto redshift.
Get PDFL'obiettivo di questa tesi è quantificare l'effetto che l'assunzione sulla distribuzione di temperatura delle
polveri ha sulle luminosità FIR delle galassie, tenendo conto della correzione-K e del redshift cosmologico. Il
metodo è stato applicato ad un campione di galassie normali a 4.5 < z < 6 osservate con ALMA a 158 μm
nell'ambito del progetto ALPINE
Economia circolare: politiche e approcci operativi. Il caso TrenDevice negli smartphone ricondizionati
Get PDF
Influenza di alcune proprietà chimiche e idrauliche sui processi di adsorbimento e movimento di glifosate in suoli di alta e bassa pianura veneta Influence of some chemical and hydraulic proprieties on the process of adsorption and movement of glyphosate in soils of the upper and lower Veneto plain
No full textAmong the ecosystem services delivered by the soil, the soil filtering against potential contaminants of the water resource -such as plant protection products- is of primary importance. Nevertheless, the multitude of exogenous substances on the market and their different physicochemical characteristics, as well as the strong site-specific variability of the soils, makes it difficult to accurately predict the environmental fate of such molecules that are released into the environment. With this thesis work, some hydraulic and chemical properties of different agricultural soils of the Veneto region were analyzed, with the aim of understanding the dynamics that regulate adsorption and transport phenomena of glyphosate, and determine the susceptibility of the territory to subsurface water contamination. Two soils were sampled in Colnù (Conegliano) and Settolo (Valdobbiadene) sites in the northern Veneto plain, and compared with a Cambisol soil, typical of the low-lying venetian plain and located at the Experimental Farm of the University of Padua in Legnaro (Padua), which usually has a shallow water table around 1 m depth. Chemical analyses at different depths were performed to determine Al and Fe oxide and hydroxide contents, whose quantification is crucial to better understand the adsorption of glyphosate to the soil. Also, the adsorption coefficients were experimentally determined and described according to Freundlich isotherms. As for the soil hydraulic properties, undisturbed samples at different depths were analyzed by determining the saturated water conductivity and the water retention curve. Finally, the HYDRUS-1D hydrological model was fed with the experimental data previously obtained and used to simulate the environmental fate of glyphosate across the soil in each site-specific context. Each simulation involved the application of glyphosate in pre-sowing conditions on a maize crop, repeated in three consecutive years. Results showed that Fe and Al oxides and hydroxides increased with the depth in Legnaro soil, where the extraction method with oxalate returned the greatest quantity of both Fe and Al oxides. In contrast, results from Colnù and Settolo did not follow the same trend as per Legnaro. In particular, the oxalate extraction method results in extracting greater quantities of oxides and hydroxides only in the Conegliano sites; in Valdobbiadene the oxides most extracted are those chelated to the organic matter. extracted with pyrophosphate. As regards the Kf, the highest values were found in Conegliano. Correlation analysis showed that the content of fine soil particles –silt and clay– and the SOM-chelated oxides determined an increase in Kf. Numerical simulations highlighted the pivotal role of a shallow water table in promoting the movement of glyphosate which, despite its affinity for the soil phase, partly moved towards the deep layers favored by high soil moisture and water conductivity. On the other hand, in high plain soils, or where the aquifer is deep, glyphosate tends to slow down its movement, to be adsorbed to soil particles, and therefore degrade before reaching the aquifer. In-depth site-specific investigations are required to provide to policy makers and local authorities useful
4
tools to define more precisely the areas more or less susceptible to possible environmental contamination, both for the surface and sub-surface waters, so as to identify safeguard and protection actions
An experimental setup for the detection of e+e- pairs in the decay of 8Be*
Get PDFThe existence of a light neutral boson, possible mediator of a new fundamental interaction, has been recently proposed by A.J. Krasznahorkay and collaborators. A series of experiments was carried out at the Atomki laboratory (Debrecen,Hungary) measuring electron-positron pairs emitted in the decay of excited 8Be and 4He. Anomalies in the e+e− angular correlation distributions with respect to the Internal Pair Creation (IPC) process have been reported. The first results concern the decay of the 18.15 MeV 1+ state in 8Be, while the group later focused on the ≈20 MeV excited states in 4He. In order to explain the reported anomaly several interpretation have been developed. The approach proposed by X. Zhang and G. Miller aims to describe the anomalies in an almost conventional nuclear physics framework, improving the reaction modeling. Conversely, a possible interpretation given by J. Feng and collaborators involves the appearance of a new gauge vector boson, called X17, that could possibly mediate the interaction between the SM and the dark sector. The Atomki results and the related interpretations triggered a renewed interest in the spectroscopy of light ions by Internal Pair Creation. At this purpose, a new experimental setup is being developed at the INFN Legnaro National Laboratories of INFN (Legnaro, Padova, Italy). The first goal is to provide an independent replica of the Hungarian experiment. This thesis work focuses on a complete simulation of the proposed setup and reports about the first experimental characterization of the prototype detectors. The text is organized as follows: after a short introduction on the physics of electromagnetic transitions and internal pair production in nuclei, the state of the art regarding the 8Be and the 4He anomalies will be discussed, explaining the roots of the X-boson or X17 case. Afterwards, the simulation and the experimental characterization works will be reported
Detecting out of context sentences with a sentence encoder
No full textThe main goal of this thesis is to build a machine learning algorithm with
the capability to solve a Natural Language Processing (NLP) problem (namely a semantic retrieval problem): the detection of “out of context” sentences inside a text. Here is a brief presentation of the chapters which compose this thesis. First, in Chapter 1 we’ll introduce the NLP problem. We’ll make a very brief historical introduction, explaining how the growing amount of available data led this branch to be one of the most studied ones nowadays. Then we’ll move to describe the technical tools we used for this work, describing the idea of building an encoder who produces vector embedding. After that, we dedicated Chapter 2 to the description of the pre-trained encoder we used to build our algorithm. This encoder is an evolution of the Universal Sentence Encoder, the Multilingual Universal Sentence Encoder. The power of this model is the ability to encode a whatever long sentence into a vector belonging to a suitably chosen space (that is R^512), and it has the property that similar sentences are mapped to similar vectors. In Chapter 2, we’ll describe the structure of the neural network which stands behind it. Then, in Chapter 3 we’ll describe the two experiments we made in order to explore this model capabilities. In the first case we replied an experiment suggested by Multilingual Universal Sentence Encoder authors for semantic retrieval. They performed this experiment for English language; we translated the STS benchmark dataset in Italian using the Google Translate API provided by the Python package deep_translator, and then we replied the experiment obtaining the same good results achieved by Multilingual Universal Sentence Encoder authors. In the second case we performed an analysis over synonyms, with the aim to see if the encoder outputs are more similar (using the cosine similarity as similarity measure) for synonyms than for two unrelated words. The fact is to think of a word as an “atomic” sentence of length 1, and in this sense we tried to analyze the model outputs for single words. Then, in Chapter 4 we reported the main contribution of the thesis, which is the building of the algorithm performing the “out of context” sentences detection. First we tried to produce a measure which can detect “out of context” sentences inside a text. We proposed three ways to find the outlier sentences, and with a simple experiment (which provides the text pollution with the insertion of a random out of context sentence into a
text). We highlighted that 2 of these 3 methods give good result, while the first one performs much worse. Then we chose one of these method to perform deeper analysis in order to show that that algorithm is capable to find “out of context” sentences inside a text. In particular we focused on repeating the same experiment over different lengths series, and we found a measure which seems to not suffer the text length, i.e. capable of producing the same performance for both long and short texts
Simulating the emergence of symbolic numerical representations: role of perceptual constraints in a deep reinforcement learning agent
No full textMathematical abilities, as any other task that requires capacity
of abstraction, are considered a distinctive human trait. Complex
mathematical concepts are built upon one apparently simple and basic
operation: counting. Children in all schools of the world learn to count,
and our prehistorical ancestors invented this behavior probably to solve
practical problems. I have built a computational model of counting
behavior, trying to investigate how it might emerge as a consequence of
computational constraints imposed on the input, learning signal and model
architecture. In particular, I have trained two deep reinforcement learning
models in two different experimental settings, consisting in learning
environments where the items to be enumerated were presented either
spatially or temporally to the RL agent. The experimental evidence
collected under the first experimental setting revealed its intrinsic
limitations as a cognitive model of human learning. The second experimental setting led to more relevant findings, which can be read under the light of recent discussions about the emergence of symbolic behavior in artificial intelligence models