Archivio Istituzionale della Ricerca - Università degli Studi di Pavia
Not a member yet
135341 research outputs found
Sort by
Intrinsic Harnack estimates for singular doubly non-linear equations
No full textWe establish forward, backward and elliptic Harnack inequalities for non-negative solutions
to a class of doubly non-linear parabolic partial differential equations. These Harnack estimates
are established in a proper range of parameters p and q. Such a range is shown to
be optimal for a Harnack estimate to hold. Quantitative boundedness estimates for solutions
and an expansion of positivity result for non-negative super-solutions are instrumental in the
proof
Unveiling Ca2+ dynamics into a model of human cerebrovascular endothelial cell line: implications for neurovascular coupling
No full textSurface nematic uniformity
No full textAn antlike observer confined to a two-dimensional surface traversed by stripes would wonder whether such a striped landscape could be devised in such a way as to appear to be the same wherever they go. Differently stated, this is the problem studied in this paper. In a more technical jargon, we determine all possible uniform nematic fields on a smooth surface. It was already known that for such a field to exist, the surface must have constant negative Gaussian curvature. Here we show that all uniform nematic fields on such a surface are parallel transported (in Levi-Civita’s sense) by special systems of geodesics, which are termed uniform. We prove that, for every geodesic on the surface, there are two systems of uniform geodesics that include it; they are conventionally called right and left, to evoke handedness. We found explicitly all uniform fields for Beltrami’s pseudosphere. Since both geodesics and uniformity are preserved under isometries, by a classical theorem of Minding, the
solution for the pseudosphere carries over all other admissible surfaces, thus providing a general solution to the problem (at least in principle). The proved existence of surface nematic uniform fields suggests the definition of a generalized intrinsic elastic energy for fluid membranes with nematic order, which is but one of the many possible applications of our geometric result
How secure is forgetting? Linking machine unlearning to machine learning attacks
No full textAs Machine Learning (ML) continues to evolve, so does the sophistication of security threats targeting data privacy and model integrity. In response, Machine Unlearning (MU) has emerged as a promising paradigm that enables the selective removal of data influence from trained models. By supporting compliance with privacy regulations (such as the GDPR's right to be forgotten) and facilitating model refinement, MU holds significant practical and legal value. Additionally, the effective deployment of MU introduces new security concerns. In real-world settings, malicious actors may exploit vulnerabilities in MU mechanisms, such as incomplete or inaccurate data removal, to infer deleted information, reintroduce adversarial behavior, or manipulate model updates. These risks highlight the urgency of understanding how classical ML threats relate to the design and operation of MU systems. However, despite its growing relevance, this intersection remains underexplored. In this article, we present a structured analysis of four major attack classes in ML (Backdoor Attacks, Membership Inference Attacks, Adversarial Attacks, and Inversion Attacks) and examine their implications for MU across multiple dimensions: (i) as direct threats targeting MU mechanisms, (ii) as challenges that MU can potentially mitigate, (iii) as evaluation metrics to measure the effectiveness and performance of MU techniques, and (iv) as verification factors to validate the success and completeness of the unlearning process. We note that not all attacks exhibit all these perspectives simultaneously; their relevance varies depending on the attack characteristics and MU scenario. We also propose a novel classification that reflects how these attacks are typically employed in this context. Finally, we identify open challenges, including ethical considerations, and highlight promising directions for future research to advance secure and privacy-preserving Machine Unlearning
Revisiting the Political Opposition of the European Parliament: A Rhetorical Analysis of Parliamentary Speeches
No full textComplementarity, Incompatibility, and Irreversible Disturbance. A Resolution to the Debate Between Bohr and Heisenberg
No full textIn 1927, Heisenberg introduced a heuristic argument, based on the famous γ-ray microscope Gedankenexperiment, to show that in quantum theory there exist operations that irreversibly disturb the systems on which they act. This argument was intended to show the existence of quantities that cannot be simultaneously measured. However, with a deeper understanding of how information can be manipulated and propagated, it is possible to prove that only the converse relation actually holds. In this thesis, exploiting the framework of the framework of Operational Probabilistic Theories (OPTs), we show that the impossibility of performing certain measurements simultaneously implies the existence of operations that irreversibly disturb the systems on which they act. We also present two toy theories, called Minimal Classical Theory (MCT) and Minimal Strongly causal Bilocal Classical Theory (MSBCT), that serve as counterexamples to the converse implication. Even though both theories satisfy full compatibility of observations, they still display irreversibility. Moreover, they are classical, Kochen-Specker, and generalised-noncontextual, yet nonetheless satisfy two quantum no-go theorems: No-Information Without Disturbance (NIWD) and no-broadcasting. This shows that these properties cannot be taken per se as signatures of non-classicality.
We also introduce two new classes of operational theories, which we here define and study: Minimal Operational Probabilistic Theories (MOPTs) and Minimal Strongly causal Operational Probabilistic Theories (MSOPTs), of which MCT and MSBCT are representatives, respectively. These theories are characterised by allowing only the minimal possible set of dynamics, with the latter also admitting classical conditioning. We prove that all MOPTs and all MSOPTs whose state spaces contain a spanning set of entangled states necessarily satisfy both NIWD and no-broadcasting.
Finally, we propose an operational definition of Bohr’s complementarity, understood as the existence of properties of physical systems that cannot be simultaneously well-defined. We show that complementarity implies incompatibility and, consequently, irreversibility. In the specific case of quantum theory, however, complementarity and incompatibility are proved to coincide.In 1927, Heisenberg introduced a heuristic argument, based on the famous γ-ray microscope Gedankenexperiment, to show that in quantum theory there exist operations that irreversibly disturb the systems on which they act. This argument was intended to show the existence of quantities that cannot be simultaneously measured. However, with a deeper understanding of how information can be manipulated and propagated, it is possible to prove that only the converse relation actually holds. In this thesis, exploiting the framework of the framework of Operational Probabilistic Theories (OPTs), we show that the impossibility of performing certain measurements simultaneously implies the existence of operations that irreversibly disturb the systems on which they act. We also present two toy theories, called Minimal Classical Theory (MCT) and Minimal Strongly causal Bilocal Classical Theory (MSBCT), that serve as counterexamples to the converse implication. Even though both theories satisfy full compatibility of observations, they still display irreversibility. Moreover, they are classical, Kochen-Specker, and generalised-noncontextual, yet nonetheless satisfy two quantum no-go theorems: No-Information Without Disturbance (NIWD) and no-broadcasting. This shows that these properties cannot be taken per se as signatures of non-classicality.
We also introduce two new classes of operational theories, which we here define and study: Minimal Operational Probabilistic Theories (MOPTs) and Minimal Strongly causal Operational Probabilistic Theories (MSOPTs), of which MCT and MSBCT are representatives, respectively. These theories are characterised by allowing only the minimal possible set of dynamics, with the latter also admitting classical conditioning. We prove that all MOPTs and all MSOPTs whose state spaces contain a spanning set of entangled states necessarily satisfy both NIWD and no-broadcasting.
Finally, we propose an operational definition of Bohr’s complementarity, understood as the existence of properties of physical systems that cannot be simultaneously well-defined. We show that complementarity implies incompatibility and, consequently, irreversibility. In the specific case of quantum theory, however, complementarity and incompatibility are proved to coincide
Analysis, Design, and Characterization of a Class-D Audio Amplifier Integrating Analog and Digital Methodologies
No full textThis thesis presents a novel Class-D audio amplifier designed and implemented using BCD C090D technology, a 90nm process node. This amplifier addresses the growing demand
in the automotive sector for integrated devices capable of delivering high-quality audio performance. The proposed solution aims for extensive digitalization of the structure,
seeking to shift as many blocks as possible in this direction. At the same time, it is based on analog feedback with the goal of reducing the converter’s requirements by
working directly on the loop error signal. The amplifier features high reconfigurability, making it adaptable to various application scenarios. The obtained Class-D is capable of
operating under typical conditions with a carrier frequency of 2.3 MHz, using an ADC that samples at only 4.6 MHz. Furthermore, a design technique is presented that highlights
how the combined knowledge of analog and digital methodologies is essential nowadays to tackle challenges with an innovative approach. Finally, the measurement
results performed on the chip are presented, carried out both in test modes dedicated to evaluating individual blocks and in functional mode. The sources of all identified limitations
are located, and corresponding intervention solutions are proposed. The system exhibits a noise floor of 44 μVrms and a THD+N of 0.023% at 21mW
New Allosteric Modulators of Molecular Chaperone TRAP1 from the Integration of Computational Biology, Medicinal Chemistry, and Biophysics
No full textProtein homeostasis is one of the key mechanisms that determine cellular life, and the Hsp90 family of molecular chaperones plays a key role in it. While Hsp90 dysregulation is a hallmark of numerous diseases, ranging from cancer to neurodegeneration, traditional inhibitors targeting its highly conserved ATPase site have largely failed in the clinic due to off-target toxicity and compensatory stress responses. One of the challenges in drug discovery, as well as in the development of chemical tools to investigate the specific roles of single family members, lies in achieving isoform specificity across the cytoplasm, ER, and mitochondria.
Here, we exploit the intrinsic asymmetry of mitochondrial isoform TRAP1 and combine it with a fragment-design inspired approach to develop new possible TRAP1 targeting leads. We start from the consideration that TRAP1 catalytic cycle relies on a strained, asymmetric dimer conformation that enforces sequential ATP hydrolysis. By integrating advanced computational dynamics with biochemical profiling, we demonstrate that small molecules can be rationally designed to target these transient asymmetric states. Our findings reveal that targeting allosteric, symmetry-breaking interfaces allows for the modulation of TRAP1, offering a novel platform and starting point for next-generation, isoform-specific anticancer therapeutics