Student Posters Repository

This study presents a comprehensive investigation into the influence of poly(N-isopropylacrylamide) (PNIPAM) polymer on the structural dynamics of the intrinsically disordered alpha-Synuclein (α-Syn) protein, exploring the formation and intricate features of the resulting α-Syn/PNIPAM complexes. Using atomistic molecular dynamics simulations, our study analyzes the impact of initial configuration, polymer molecular weight, and temperature on the α-Syn and α-Syn/PNIPAM complex. Long atomistic simulations of the protein/polymer complex reveal crucial insights into molecular interactions, emphasizing a delicate balance of forces governing stability and structural evolution. The study accentuates the critical role of the initial configuration, demonstrating the high affinity for the protein regardless of its position in the initial configurations. The study provides an in-depth analysis of the role of molecular weight and temperature in modulating the kinetics of protein/polymer complex formation. Key findings highlight the intricate interplay between these factors, elucidating the molecular mechanisms governing the formation and stability of the complex. Insights into molecular weight-dependent modulation of electrostatic interactions are highlighted, providing valuable information on the nuanced interaction between polymer characteristics and protein behavior. In particular, the 40mer PNIPAM intermediate chain length emerges as pivotal, influencing structural dynamics.  Temperature-dependent analysis at 298 K, 305 K, and 310 K reveal significant impacts on protein structure and polymer-protein interactions. As a result of the study, it has been observed that the PNIPAM polymer interacts with the α-Syn protein through electrostatic and van der Waals interactions, affecting different regions of the protein that exhibit distinct polarities in various ways. The study underscores the unique response of the NAC (non-amyloid compartment) region to temperature, providing stability to variations. These findings, which delves into polymer-protein interactions, hold promise as potential guidance for therapeutic strategies in various neurodegenerative disorders.

Keywords: Alpha-Synuclein, poly(N-isopropylacrylamide), protein misfolding, and molecular dynamics.

This study concerns a comprehensive investigation in the noise produced in the blades of rotating machines (fans, turbines, compressors, etc), focusing on the trailing-edge noise, as it is an important contributor to the total noise emitted. It is originated by the interaction of the turbulent eddies born in the boundary layer and wake unsteadiness with the airfoil itself. Then, the turbulent eddies are convected past the trailing edge, distorting such eddies and producing acoustic waves. However, the complexity of the underlying origin of such noise mechanisms difficults the understanding of such noise generation. Using high-fidelity simulations (Direct Numerical Simulations - DNS) of the Controlled Diffusion (CD) airfoil on the turbulent flow field and the associated self-noise generation mechanisms is evaluated. The installation effects, which modify the loading on the airfoil, are relevant in the boundary layer development on the airfoil. Therefore, in the current DNS, the airfoil is embedded in the core of a jet mimicking the experimental conditions in the anechoic chamber of École Centrale de Lyon. Two angles of incidence are evaluated, 8º and 15º, corresponding to operating point and pre-stall conditions respectively. The study provides an in-depth analysis of the role of three noise sources: the interaction between the attached turbulent flow and the airfoil trailing edge, the flow separation/reattachment at the suction side leading edge, and a third source in the near wake. In addition, these DNS, together with a set of Large Eddies Simulations (LES) compose a numerical database used to train an Artificial Neural Network to predict the wall-pressure spectrum (WPS) as an alternative to empirical models. The ANN is composed of two blocks, a boundary layer velocity autoencoder to compress and extract the driving parameters of the boundary layer, and a WPS predictor. The results on the exploitation of the database have resulted in the evidence of the existence of a minimum of three boundary layer parameters driving the boundary layer evolution, no matter the nature of the pressure gradient. The WPS prediction captures the effect of the main mid-frequency tonal peak as well as the generalized noise trends in the low, mid, and high-frequency content. The latest findings could lead to a rapid and general WPS model that could be used in early design stages of engineering applications.

The wind turbine (WT) is a complex system that integrates aerodynamic, mechanical, and electrical components. Simulating the interaction of these various physics disciplines is a labor-intensive process, particularly when incorporating finite element method (FEM) modeling. Utilizing high-performance computing (HPC) for such simulations significantly reduces the computation time. Moreover, it allows for the implementation of advanced optimization techniques, enabling more efficient design enhancements for modern wind turbines. Incorporating magnetic bearings (MB) and magnetic gears (MG) into the wind turbine design aims to eliminate vibration noise and frictional issues in the drive-train of traditional turbines. This innovation extends the turbine's lifespan, improves system reliability, and reduces maintenance costs.

Over the past few decades, cosmological simulations have become invaluable tools that are helping us to reach a deeper comprehension of astrophysics, particularly about the formation and evolution of galaxies in the early Universe (Springel et al., Nature, 2005). The intrinsic multi-scale nature of astrophysical processes renders the task of comprehending the functioning of the Universe a challenging endeavor. This is extremely evident in cosmological simulations, as the reproduction of the behavior of the Universe necessitates the inclusion of phenomena ranging from atomic physics to galactic scales.
The advent of sophisticated numerical techniques and an enhanced computing capacity have enabled modern simulations to follow every fundamental component of the Cosmos, such as dark matter, dark energy and baryonic matter (Plack Collaboration, A&A, 2016). The latter, more than the others, is the most challenging to add in a numeric hydrodynamic simulation, because baryonic matter is involved in many physical processes. Nevertheless, its presence in a simulation is indispensable for the formation of stars and black holes, which represent the main characters in the evolution of the Universe.
The concept of high-performance computing has facilitated the development of increasingly sophisticated hydrodynamic simulations. Indeed, while the desire to create more realistic simulations is understandable, it is also important to consider the time required to obtain results. The incorporation of additional physics necessitates an increase in the computational time. The parallel computation has resulted in a significant acceleration of cosmological hydrodynamic simulations. In our simulations, we employ a combination of shared memory, message passing techniques and GPUs to reduce the time-consuming nature of our codes.
Our research group at Sapienza University of Rome is interested in the formation channels of supermassive black holes (Event Horizon Telescope Collaboration, ApJL, 2019), which are black holes with masses between 10⁸ ÷ 10¹⁰ M_Θ situated at the center of galaxies. The formation of a black hole can occur in several ways, however the maximum mass estimated is approximately 10⁴ M_Θ (Omukai K., ApJ, 2001; Costa et al., MNRAS, 2023). Consequently, the key question is how they grow. Furthermore, the modern space-born James Webb telescope has discovered numerous supermassive black holes at a few million years after the Big Bang (Inayoshi et al., ARA&A, 2020). This challenges our understanding of black hole physics, as the growth of the first black holes should have been extremely rapid.
In my current PhD research, I am utilizing the hydrodynamic galaxy evolution code based on the SPH formalism called dustyGadget (Graziani et al. MNRAS, 2020) to investigate the pristine environments where the first stars are formed, with the goal of elucidating the conditions under which the first stellar black holes will be created. Subsequently, the way these “seeds” will assemble to give rise to the first supermassive black holes will be studied. The final steps will how these massive objects will coevolve with their own host galaxy. Indeed, it remains unclear whether the winds and matter accretion of SMBHs could play an important role in defining the final morphology of their host galaxy.

This poster presents "INNA: Implementation of Non-Equilibrium Applications" code, focusing on the modeling and simulation of spacecraft entry into planetary atmospheres, which involves high-speed entries leading to significant thermal and physico-chemical processes. The project addresses strong thermo-chemical non-equilibrium effects related to high-speed flows. It includes the development and limitations of multi-temperature models, emphasizing the need for accurate models to describe internal population distributions, such as vibrational state-to-state kinetics that capture non-Boltzmann effects.

The physical models described encompass gasdynamics, vibrational non-equilibrium, kinetic and transport models. The gasdynamics model includes mass, momentum, and energy conservation equations with specific initial conditions and relaxation processes. Vibrational non-equilibrium models are detailed with multi-temperature and state-to-state approaches. The kinetic model incorporates vibrational level treatments, source terms for non-equilibrium processes, and iterative methods for computing vibrational temperatures. The computational cost of state-to-state approaches is highlighted.

The simulation results demonstrate the comparison between multi-temperature and state-to-state models, showcasing flow conditions and vibrational descriptions. Applications for one-dimensional shock relaxation and axisymmetric simulations are presented.

In conclusion, the study emphasizes the importance of accurate vibrational state-to-state kinetics, efficient coupling with computational fluid dynamics (CFD), and the influence on thermo-chemical relaxation. Algorithmic considerations include full implicitation, domain partitioning, parallelization, and the introduction of machine learning/artificial intelligence (ML/AI) for model enhancement and computational speedup.

The Elongation (ELG) method has been developed as an efficient and accurate electronic structure calculation method for macromolecules. To further improve efficiency, a novel alternating multi-directional ELG method is introduced, enabling the automated property optimization (popt) process to proceed alternately in multiple directions. This alternating popt approach was employed to optimize the (hyper)polarizabilities of donor-acceptor-substituted polydiacetylenes (PDAs) aligned along the z-axis. During each cycle of these popt processes, four types of donor (methoxy, amino, hydroxyl, and sulfhydryl)-substituted diacetylene (DA) monomers and four types of acceptor (cyano, fluoro, nitro, and formyl)-substituted DA monomers were alternately selected and attached to the two terminals of the PDA. The popt processes aimed at maximizing the PDAs’ (hyper)polarizability consistently led to notably faster enhancements in these physical quantities than processes aimed at minimizing them, demonstrating the new alternating popt process's capability in designing systems with the expected properties. The computational cost for the multi-directional popt process using the new alternating ELG method was theoretically compared with that required for the popt process using the existing simultaneous ELG method, by which monomers are simultaneously attached in both elongation directions. When considering n types of donor and n types of acceptor groups, popt using the simultaneous ELG method requires calculating n² combinations for selecting each pair of DA monomers. In contrast, employing the alternating ELG method only requires calculating 2n combinations, which can lead to significant efficiency improvements. Future works will focus on further enhancing popt process’s efficiency through multi-node parallel ELG calculations and the application of machine learning techniques. 

The interstellar medium is the ensemble of gas and dust found among stars inside a galaxy, and it is the place where stars and planets are formed. During their life, stars produce elements heavier than helium, called “metals”, and enrich the surrounding gas.

In the early universe, stars did not have yet the time to enrich the interstellar medium, which then had a so-called low “metallicity”, meaning a low amount of metals present. This different chemical composition has a number of consequences on the chemistry processes taking places in the interstellar medium, impacting its structure and evolution. As today’s most powerful telescopes enable us to have a look at the very early universe, there is the need to understand these primitive environments by means of simulations.

In this work we employ SILCC simulations, which are based on a modified version of the adaptive mesh refinement code FLASH, to model the star formation process, the stellar feedback, and the structure and evolution of the interstellar medium in these environments. Our code is divided into several modules that take care of each physical process alone, while communicating with the other modules, however all of them can run in parallel following different timesteps depending on the process.Our code uses OpenMP to run on different nodes on a supercomputer.We develop in this work two new methods to compute heating mechanisms in the interstellar medium.

Gravitational wave detections of black holes (BHs) in the pair-instability mass gap have sparked interest in dynamical formation channels. In my work, I have explored the process of hierarchical BH mergers in active galactic nuclei (AGNs), which stand out with respect to other dynamical environments for three main reasons: enhanced binary formation due to migration traps, accelerated binary inspiral due to gas hardening and high merger remnant retention due to the deep gravitational potential of the AGN. In this poster presentation, I will show the main results of my new semi-analytical model, which allows me to effectively explore the parameter space while capturing all the main physical processes involved. I will show that gas hardening dramatically impacts the properties of binary BH (BBH) mergers. Continuous gas accretion on binaries during their inspiral greatly enhances the efficiency of hierarchical mergers, leading to the formation of intermediate-mass BHs (up to ten thousand solar masses).

Studying protein-protein interactions (PPIs) is a central pursuit in chemical biology, with significant implications for drug discovery. While the inhibition of PPIs has been extensively explored, the challenge lies in identifying the small molecules that can stabilize these interactions. It has been experimentally observed that dissociation rates (k_off) are a determinant factor in predicting the efficacy of drug-target interactions, specifically in their impact on complex dissociation. Atomistic simulations allow us to observe insights into the molecular forces influencing PPI stabilization, which are not easily accessible through experiments.  However, there remains a gap in how the dissociation rates of PPI complexes in the presence of stabilizers can be proactively targeted in drug design, as few computational studies have taken on this challenging task.