IRIS - Res&Arch Institutional Research Information System Università degli Studi di Perugia
Not a member yet
126129 research outputs found
Sort by
Management of autoscaling serverless functions in edge computing via Q-Learning
No full textServerless computing is a recently introduced deployment model to provide cloud services. The autoscaling of function instances allows adapting allocated resources to workload, so as to reduce latency and improve resource usage efficiency. However, autoscaling mechanisms could be affected by undesired ‘cold starts’ events, causing latency peaks due to spawning of new instances, which can be critical in edge deployments where applications are typically sensitive to latency. In order to regulate autoscaling of functions and mitigate the latency for accessing services, which may hinder the adoption of the serverless model in edge computing, we resort to the usage of reinforcement learning. Our experimental system is based on OpenFaaS, the most popular open-source Kubernetes-based serverless platform. In this system, we introduce a Q-Learning (QL) agent to dynamically configure the Kubernetes Horizontal Pod Autoscaler (HPA). This is accomplished via a QL model state space and a reward function definition that enforce service level agreement (SLA) compliance, in terms of latency, without allocating excessive resources. The agent is trained and tested using real serverless function invocation patterns, made available by Microsoft Azure. The experimental results show the benefits provided by the proposed solution over state-of-the-art in terms of compliance to the SLA, while limiting resource consumption and service request losses
On the instant-controllability of a second order multidimensional differential equation subjected to damping term and impulses
No full textIn this paper the existence of admissible trajectory control-pairs for an impulsive problem driven by a multidimensional differential equation with a nonlinear Balakrishnan-Taylor type damping term, is investigated. This purpose is achieved rewriting the impulsive problem in an abstract form governed by a semilinear second order differential inclusion in which the nonlinear term also depends on the first derivative. The method used leads to a preliminary study of the existence of mild solutions for a non-impulsive multivalued problem on a closed and bounded interval, stating two new results in non reflexive Banach spaces. Then, the mild solution in [0, ∞) for the impulsive abstract multivalued problem is obtained glueing the solutions defined on the bounded intervals. This approach allows to not require the continuity on the impulsive functions. Applying the abstract impulsive multivalued results we achieve the desired existence of admissible trajectory control-pairs for the impulsive phenomena described by the multidimensional differential equation. The paper concludes with the study of an instant-controllability relatively to a suitable functional for the impulsive problem in exam. The established results improve recent theorems present in the literature and obtained in reflexive Banach spaces and
assuming the continuity on the impulsive functions
Pea protein isolates adsorption on phospholipid bilayer interfaces: a quartz crystal microbalance and neutron reflectometry study
No full textMellin Analysis, Transform Theory, and Applications
No full textThis monograph presents the theory of the Mellin transform and the resulting Mellin analysis in a rigorous and unified manner. Often dismissed as a subordinate topic within Fourier and Laplace transform, it is instead demonstrated here that the theory is completely independent, can be studied within a self-contained framework, and exhibits some typical characteristics.
In addition to highlighting the foundations of the theory, the book addresses applications to certain partial differential equations, sampling theory and numerical quadrature. These applications provide methods which are in turn of interest in various areas of mathematics, science, and engineering.
Each chapter is enriched by numerous references to further literature and potential research directions. Researchers working in this field will gain new insights and appreciate the deserved attention for this underrated topic in harmonic analysis
Temperature self compensation for dynamic sensitivity in 3D-printed piezoresistive sensors
No full textTemperature variations induce nonlinear changes in the dynamic sensitivity of 3D-printed piezoresistive sensors, thereby limiting their deployment in thermally variable environments. Conventional temperature-compensation approaches rely on dedicated temperature sensors or extensive calibration matrices, which increase system complexity and reduce measurement reliability. This research investigates the hypothesis that temperature-induced changes in dynamic sensitivity in 3D-printed thermoplastic extrusion technology piezoresistive sensors are directly proportional to the changes in electrical resistance. The hypothesis was tested against twenty-four single-process 3D-printed accelerometers operating from 5 °C to 35 °C in three humidity conditions. Strong linear correlations (R2>0.98) between the relative resistance and the dynamic sensitivity changes were found. The resistance-based temperature self-compensation method does not require additional sensors. It enables accurate sensitivity prediction with errors below 5 % in normal humidity and 12 % in extreme humidity environments. This approach eliminates the need for temperature sensors while maintaining measurement accuracy, enabling the deployment of thermoplastic 3D-printed piezoresistive sensors in smart structures operating across varying environmental conditions
Incidence, Risk Factors, and Management of Conjunctivitis in Atopic Dermatitis Patients Treated With Dupilumab or Tralokinumab: Results From a Multicenter, Observational, Retrospective Study
No full textBackground: Conjunctivitis is among the most frequent adverse events (AEs) emerged in clinical trials for all biologic drugs approved for atopic dermatitis (AD). However, real-world comparative data on the incidence, risk factors, and management of conjunctivitis remain limited. Objective: We aimed to compare the incidence, clinical features, and management of conjunctivitis in patients with moderate-to-severe AD treated with dupilumab or tralokinumab in a real-life setting. Patients and Methods: A multicenter, retrospective, observational study including adult patients with moderate-to-severe AD treated with dupilumab or tralokinumab for at least 16 weeks was carried out. Demographic, clinical, and therapeutic data were collected from 35 dermatological referral centers across Italy. Conjunctivitis incidence, severity, time to onset, and ophthalmologic management were analyzed and compared between treatment groups. Results: A total of 6668 patients were included (5899 on dupilumab and 769 on tralokinumab). Conjunctivitis occurred in 10.76% of dupilumab-treated and 12.61% of tralokinumab-treated patients, with no statistically significant difference in overall incidence. However, time to onset was significantly shorter with tralokinumab than with dupilumab (15.3 ± 14.5 weeks vs. 35.5 ± 45.2 weeks, respectively; p < 0.0001). Ophthalmologic management strategies were similar between groups, mainly involving lubricants and corticosteroid-based eye drops. Dupilumab-treated patients more frequently discontinued or switched treatment due to conjunctivitis than tralokinumab patients (25.4% vs. 14.4%, respectively; p < 0.05). Conclusions: Conjunctivitis represents a relatively frequent AE in patients with AD receiving dupilumab or tralokinumab. However, earlier onset of conjunctivitis with tralokinumab and higher discontinuation rates with dupilumab suggest differing tolerability profiles. Early recognition of ocular symptoms is essential, and dermatologists should promptly initiate supportive eye care and refer to ophthalmologists when appropriate to avoid unnecessary treatment interruptions
Existence and nonexistence of solutions for elliptic inequalities involving gradient terms and weights
No full textIn this paper we prove existence and nonexistence theorems for positive so-
lutions of elliptic inequalities for general quasilinear operators, including m-Laplacian,
mean curvature and generalized mean curvature operator, in the entire RN with a re-
action involving power type gradient terms and positive weights, possibly singular or
degenerate. A complete picture for the exponents involved is given. The proof technique
is based on cumbersome integral a priori estimates, in the spirit of the nonlinear capacity
method. No maximum principle or growth conditions at infinity for the solutions are
required
The Gut Microbiome in Early-Onset Colorectal Cancer: Distinct Signatures, Targeted Prevention and Therapeutic Strategies
No full textPhylogenetic Position of Hungarian Grey Cattle Breed Based on Total-Representation Sample
No full textThe Hungarian Grey (HG) cattle breed was almost extinct after WW2; only 200 cows and six bulls survived. Despite the historical significance of the HG, no comprehensive genomic analysis has been conducted to clarify its genetic diversity and evolutionary history. Previous studies have relied on random or limited pedigree sampling, lacking a fully representative dataset determining genetic and conservation status. Here, the founder sampling of 110 individuals and the analysis of their mitochondrial DNA (mtDNA) sequence variation aim to investigate the phylogenetic placement of the breed using, for the first time, a fully representative sample. All identified haplogroups belong to the taurine T macro-haplogroup, with a predominance of T3 (89.1%), followed by T2 (4.5%), T1 (3.6%), and T1'2'3 (2.7%). The phylogenetic analysis confirms the absence of ancient haplogroups derived from European aurochs, suggesting a purely taurine origin for the HG breed. The high haplotype diversity (Hd = 0.94) and the genetic similarity to other Podolian breeds, particularly Maremmana cattle, indicate a preserved genetic background despite centuries of selective breeding. The lack of intensive crossbreeding practices has maintained the original beef production purpose of the breed, distinguishing it from the crossbred Podolian cattle used for dual-purpose or dairy production
A Novel Computational Framework for Visual Snow Syndrome
No full textVisual Snow Syndrome (VSS) is a neurological condition that impairs the perceived image by causing the appearance of numerous constantly flickering dots, similar to those seen on a television screen that does not receive the antenna signal properly. This article presents two innovative applications using emerging technologies to help people suffering from VSS. One is based on Augmented and Virtual Reality to simulate the visual perception experienced by people affected by such syndrome in real time on the image sequence acquired by a smartphone's camera. The other is based on a web platform based on a Virtual Reality engine, designed to create exercises parametrically to improve the UX and the usability of an experimental therapy that aims to alleviate the syndrome's symptoms. We present the architecture of the two systems, which is characterised by a very simple yet very effective approach and use, as documented by the comments made by users. The results of an evaluation questionnaire submitted to volunteers who tested and scored the mobile App and eventually expressed opinions and suggestions are reported. The questionnaire results show users' positive appreciation, as they outline in their comments how important it is for patients to have simulation tools that allow them to better explain their pathologies