13675 research outputs found
Sort by
Integration of Computational Techniques for System Modeling and Data Assimilation in Industrial and Life Science Problems
No full textThis thesis presents a series of innovative studies demonstrating the development and application of advanced computational and machine learning methodologies to address real-world, three-dimensional problems with significant practical relevance. Unlike conventional research, which often relies on idealized or simplified scenarios, this work directly tackles complex, practical challenges found in everyday engineering and biomedical contexts.
The first part of the thesis focuses on the development of a comprehensive system model for a washer-dryer, incorporating both analytical and machine learning approaches to predict the work cycle of a fan/heater module.
The second part investigates the wake flow fields of wind turbines, a key factor in optimizing renewable energy generation in wind farms. Through advanced simulations and dimensionality reduction techniques, the analysis explores the effects of ice accretion, structural flexibility, and turbulence, offering actionable insights for enhancing wind farm layouts and turbine performance.
The third and fourth chapters introduce a novel approach to cooling system optimization, leveraging neural networks for flow prediction and deep reinforcement learning for intelligent control.
Finally, the thesis addresses a critical biomedical challenge: predicting boundary conditions in cardiovascular flows using data assimilation. By integrating the full Navier-Stokes equations with observational data, the research achieves unprecedented accuracy in modeling patient-specific blood flow dynamics, with direct implications for diagnostic and therapeutic applications.
Collectively, this thesis advances the state-of-the-art in computational modeling by applying cutting-edge techniques to practical, high-dimensional, and non-linear problems in diverse domains. The methodologies and insights presented herein establish a foundation for tackling complex real-world problems
Developmental Trajectories Predict Dendritic Remodeling After Injury
No full textNeurons in the adult central nervous system exhibit limited regenerative capacity, yet certain retinal ganglion cell subtypes exhibit greater resilience. We tested whether the timing of dendritic maturation shapes subtype-specific responses to injury. Reconstruction of over 1,000 retinal ganglion cells shows that ON-sustained (sONα) and ON-transient (tONα) cells follow distinct developmental trajectories: tONα cells reach peak dendritic size by postnatal day 10, while sONα cells mature by day 14. Post-injury, both subtypes undergo dendritic shrinkage; however, sONα cells remodel more rapidly and stabilize earlier. Computational modeling indicated that injury-induced morphologies resemble earlier developmental stages. Deletion of PTEN and SOCS3, which promotes axon regeneration, led to increased dendritic regression. These findings suggest that developmental timing constrains structural remodeling after injury and that axonal regeneration occurs at the expense of dendritic stability, highlighting a trade-off between axon growth and maintenance of dendritic architecture in adult retinal ganglion cells
StAGE: Stellar Archaeology-driven Galaxy Evolution
No full textWe build a semiempirical framework of galaxy evolution (dubbed StAGE) firmly grounded on stellar archaeology. The latter provides data-driven prescriptions that, on a population statistical ground, allow us to define the age and the star formation history for the progenitors of quiescent galaxies (QGs). We exploit StAGE to compute the cosmic star formation rate (SFR) density contributed by the progenitors of local QGs, and show it to remarkably agree with that estimated for high-z dusty star-forming galaxies which are faint/dark in the near-infrared (NIR), so pointing toward a direct progenitor-descendant connection among these galaxy populations. Furthermore, we argue that by appropriately correcting the observed stellar mass density by the contribution of such NIR-dark progenitors, StAGE recovers a SFR density which is consistent with direct determinations from UV/IR/radio surveys, so substantially alleviating a longstanding tension. Relatedly, we also show how StAGE can provide the average mass and metal assembly history of QGs, and their redshift-dependent statistics. Focusing on the supermassive black holes (BHs) hosted by massive QGs, we exploit StAGE to reconstruct the average BH mass assembly history, the cosmic BH accretion rate density as a function of redshift, and the evolution of the Magorrian-like relationship between the relic stellar and BH masses. All in all, StAGE may constitute a valuable tool to understand via a data-driven, easily expandable, and computationally low-cost approach the coevolution of QGs and of their hosted supermassive BHs across cosmic times
Cosmic Magnification of High-Redshift Submillimeter Galaxies
No full textWeak lensing magnification probes the correlation between galaxies and the underlying matter field in a similar fashion to galaxy–galaxy lensing shear. Although it has long been sidelined in favor of the latter on the grounds of poorer performance in terms of statistical significance, the provision of a large sample of high-redshift submillimeter galaxies by the Herschel observatory has transformed the landscape of cosmic magnification due to their optimal physical properties for magnification analyses. This review aims to summarize the core principles and unique advantages of the cosmic magnification of high-redshift submillimeter galaxies and discuss recent results applied to cosmological inference. The outlook and challenges of this observable are also outlined, with a focus on the ample scope for exploration and its potential to emerge as a competitive independent cosmological probe
MDRefine: A Python package for refining molecular dynamics trajectories with experimental data
No full textMolecular dynamics (MD) simulations play a crucial role in resolving the underlying conformational dynamics of molecular systems. However, their capability to correctly reproduce and predict dynamics in agreement with experiments is limited by the accuracy of the force-field model. This capability can be improved by refining the structural ensembles or the force-field parameters. Furthermore, discrepancies with experimental data can be due to imprecise forward models, namely, functions mapping simulated structures to experimental observables. Here, we introduce MDRefine, a Python package aimed at implementing the refinement of the ensemble, the force field, and/or the forward model by comparing MD-generated trajectories with the experimental data. The software consists of several tools that can be employed separately from each other or combined together in different ways, providing a seamless interpolation between these three different types of refinement. We use some benchmark cases to show that the combined approach is superior to separately applied refinements. MDRefine has been released as an open-source package under the LGPLv2+ license. Source code, documentation, and examples are available at https://pypi.org/project/MDRefine and https://github.com/bussilab/MDRefine
The circular disc made of linear elastic incompressible material and the 'bathyscaphe lesson'
No full textA linear elastic circular disc is analysed under a self-equilibrated system of loads applied along its boundary. A distinctive feature of the investigation, conducted using complex variable analysis, is the assumption that the material is incompressible (in its linearized approximation), rendering the governing equations formally identical to those of Stokes flow in viscous fluids. After deriving a general solution to the problem, an isoperimetric constraint is introduced at the boundary to enforce inextensibility. This effect can be physically realized, for example, by attaching an inextensible elastic rod with negligible bending stiffness to the perimeter. Although the combined imposition of material incompressibility and boundary inextensibility theoretically prevents any deformation of the disc, it is shown that the problem still admits non-trivial solutions. This apparent paradox is resolved by recognizing the approximations inherent in the linearized theory, as confirmed by a geometrically nonlinear numerical analysis. Nonetheless, the linear solution retains significance: it may represent a valid stress distribution within a rigid system and can identify critical conditions of interest for design applications
Conforming virtual element method for nondivergence form linear elliptic equations with Cordes coefficients
No full textWe propose and analyze an H2-conforming virtual element method (VEM) for the simplest linear elliptic PDEs in nondivergence form with Cordes coefficients. The VEM hinges on a hierarchical construction valid for any dimension d ≥ 2. The analysis relies on the continuous Miranda-Talenti estimate for convex domains ω and is rather elementary. We prove stability and error estimates in H2(ω), including the effect of quadrature, under minimal regularity of the data. Numerical experiments illustrate the interplay of coefficient regularity and convergence rates in H2(ω)
Automatic feature selection and weighting in molecular systems using Differentiable Information Imbalance
No full textFeature selection is essential in the analysis of molecular systems and many other fields, but several uncertainties remain: What is the optimal number of features for a simplified, interpretable model that retains essential information? How should features with different units be aligned, and how should their relative importance be weighted? Here, we introduce the Differentiable Information Imbalance (DII), an automated method to rank information content between sets of features. Using distances in a ground truth feature space, DII identifies a low-dimensional subset of features that best preserves these relationships. Each feature is scaled by a weight, which is optimized by minimizing the DII through gradient descent. This allows simultaneously performing unit alignment and relative importance scaling, while preserving interpretability. DII can also produce sparse solutions and determine the optimal size of the reduced feature space. We demonstrate the usefulness of this approach on two benchmark molecular problems: (1) identifying collective variables that describe conformations of a biomolecule, and (2) selecting features for training a machine-learning force field. These results show the potential of DII in addressing feature selection challenges and optimizing dimensionality in various applications. The method is available in the Python library DADApy
Exploring RNA destabilization mechanisms in biomolecular condensates through atomistic simulations
No full text: Biomolecular condensates are currently recognized to play a key role in organizing cellular space and in orchestrating biochemical processes. Despite an increasing interest in characterizing their internal organization at the molecular scale, not much is known about how the densely crowded environment within these condensates affects the structural properties of recruited macromolecules. Here, we adopted explicit-solvent all-atom simulations based on a combination of enhanced sampling approaches to investigate how the conformational ensemble of an RNA hairpin is reshaped in a highly concentrated peptide solution that mimics the interior of a biomolecular condensate. Our simulations indicate that RNA structure is greatly perturbed by this distinctive physico-chemical environment, which weakens RNA secondary structure and promotes extended nonnative conformations. The resulting high-resolution picture reveals that RNA unfolding is driven by the effective solvation of nucleobases through hydrogen bonding and stacking interactions with surrounding peptides. This solvent effect can be modulated by the amino acid composition of the model condensate as proven by the differential RNA behavior observed in the case of arginine-rich and lysine-rich peptides