Flagship Research Projects

Description

In the last hundred years, General Relativity has passed a plethora of experimental tests on a vast range of energies, curvatures, field strengths and velocities. Nevertheless, since it is not power-counting renormalizable, we know that it must be corrected in the ultraviolet, where it should be replaced by a yet unknown quantum theory of gravity. Moreover, the fact that the universe on large scales cannot be described by General Relativity without introducing a dark sector (dark matter and dark energy/cosmological constant) calls for entertaining at least the possibility that General Relativity may also have to be modified in the infrared.

With this project, we will explore the use of binary pulsar systems to constrain modifications/extensions of General Relativity in both the infrared and ultraviolet limit. We will consider in particular effective field theories of dark energy, which produce deviations from General Relativity on cosmological scales by introducing scalar gravitons, and which reduce to General Relativity in the quasistatic weak field regimes of the solar system thanks to nonlinear screening mechanisms. In the ultraviolet, we will consider gravitational theories that are power counting renormalizable thanks to violations of Lorentz symmetry, thus providing a possible model of quantum gravity. Binary pulsars, thanks to the strong gravitational field inside the stars and to the high precision timing data collected by radio observatories, are the perfect arena to test these gravity theories in a hitherto unexplored regime. In more detail, we will derive theoretical predictions for both the dissipative dynamics (damping of the orbital period) and for the conservative dynamics of binary pulsar systems in both classes of theories (effective field theories of dark energy and Lorentz violating gravity). We will then compare these theoretical predictions to existing binary pulsar data, which we will reprocess to eliminate possible systematics. We will also explore the possibility of using black hole-pulsar systems (when they are discovered) to probe quantum gravity effects near the event horizon using different models of black hole information escape.

Objectives

GUVIRP's goal is to use the most recent pulsar timing data from relativistic binaries to gain insight on the behavior of gravity in both the deep ultraviolet regime (i.e. at high energies, where deviations from GR may appear due to quantum effects) and in the infrared cosmological regime, where deviations from GR are hinted at by the existence of a hitherto unexplained Dark Sector.

Expected or Achieved Results

In Kuntz & Barausse, Phys. Rev. D 109, 124001 (2024), we showed that in scalar-tensor gravity, a time-varying scalar field can change the spin period of an isolated rotating neutron star by violating the equivalence principle through a new effect called angular momentum sensitivity. This effect, computed for realistic nuclear equations of state, may be tested via pulsar timing and has implications for modified gravity and ultralight dark matter models. The latter were explored by Smarra, Kuntz, Barausse et al in Phys. Rev. D 110 (2024) 043033 using the data of the European Pulsar Timing Array. Moreover, in Cayuso, Kuntz, Bezares & Barausse Phys. Rev. D 110, 104071 (2024), we showed that in K-essence theories (an effective theory of dark energy), scalar dipole radiation from compact object binaries is strongly suppressed, but quadrupole radiation remains significant. Thus, future gravitational-wave detectors could constrain these theories using quadrupole emission.

Description

The ARES project aims to investigate how liquid water can persist beneath Mars’ south polar cap, as suggested by data from the MARSIS radar sounder on board the Mars Express orbiter. It combines thermal and dielectric numerical modeling with brine chemistry to explain the phenomenon and support future exploration of Mars and icy worlds.

Objectives

Expected or Achieved Results

The SISSA research unit focuses on dielectric and thermal-transport quantum numerical simulation. We use machine-trained molecular dynamics to investigate how salts influence the dielectric response of water—key to interpreting the radar reflections detected by the MARSIS radar. Our simulations reveal that the reduction in dielectric constant and the shift in absorption peak arise from ion-induced changes in water’s molecular orientation. Current efforts aim to extend this research to the temperature range relevant to the Martian south polar environment. We also plan to model thermal-transport properties to better understand how heat could dissipate from subglacial liquid water reservoirs.

Description

This project develops advanced simulations to accurately place Mg²⁺ and K⁺ ions in cryo-EM maps of RNA, enabling atomic-level insight into RNA structure and catalysis. It focuses on the group II intron ribozyme, with potential applications in genome editing.

Objectives

This project develops advanced simulations to accurately place Mg²⁺ and K⁺ ions in cryo-EM maps of RNA, enabling atomic-level insight into RNA structure and catalysis. It focuses on the group II intron ribozyme, with potential applications in genome editing.

Expected or Achieved Results

A method for integrating molecular simulations and cryo-EM maps has been published [1]. In addition, we participated in the CASP16 competition with a prediction of how Mg²⁺ ions arrange around a ribozyme, with excellent results (method described in [2], results are in a manuscript that is in preparation).

[1] Posani et al, bioRxiv preprint. 
[2] Tribello et al, https://doi.org/10.1063/5.0251501

Description

Entanglement is the essential trait of quantum mechanics. Pioneering quantum optics experiments, performed at the end of last century with single quantum objects like individual photons and atoms, established the true reality of entanglement, also setting the ground for the second quantum revolution and the application of entanglement as a resource for quantum technologies. While the theory and the methods to measure entanglement in simple quantum systems are nowadays well established, it is much more demanding to take into account the interactions in quantum many-body systems. The characterization and the measurement of entanglement in highly correlated states of matter is a major challenge in quantum physics, requiring a simultaneous attack from two different fronts: theory and experiments. On the one hand, unveiling the buildup of long-range quantum correlations in experiments demands novel theoretical approaches. On the other hand, the fragility of entangled states to environmental couplings calls for a new generation of experiments, where robust many-body systems can be engineered in the lab at the individual constituent level.

Objectives

This project precisely targets these fundamental challenges, with the ambition of strongly advancing our understanding and probing capabilities of highly entangled quantum matter. To this end, we establish a collaborative effort involving a groundbreaking experimental platform and the most effective theoretical methods in a multi-angled, synergistic enterprise.

Expected or Achieved Results

Concerning the Trieste units, many results have already been reached. They concern the theory of the Quantum Mpemba effect and also its experimental verification; the development of space-time duality to access entanglement-related quantities in non-equilibrium quantum matter; the study of quantum correlations in inhomogeneous systems using quantum generalized hydrodynamics; the characterization of full counting statistics in quantum simulators; the quasiparticle picture to describe entanglement Hamiltonian and entropy after a quantum quench.

Description

The recent progress in machine learning is transforming many aspects of science and society. The first wave of successes with deep learning were obtained using supervised learning, a simple paradigm where neural networks are trained to perform a task using large amounts of carefully labeled examples. This approach yields specialized models that perform extremely well on the given task, yet the resulting models easily fail outside of the controlled environment in which they were trained, are unable to acquire new skills without forgetting what was learned previously, and require large amounts of labeled data, which is generally difficult to obtain.

Self-supervised learning (SSL) is an alternative approach to training neural networks. Its key idea is to leverage large amounts of unlabeled data by training the models to predict a hidden part or property of an input from the rest of this input. A classic example is predicting a hidden word in a sentence, a task that only requires a large corpus of text, but no labels.

While scaling up self-supervised learning to enormous data sets laid the foundation for large language models like ChatGPT, what and how neural networks learn during SSL is an open question. This is particularly problematic as the power of SSL to leverage unstructured data is enabling its application to sensitive areas in business and health care, for example.

This project builds on recent advances in analyzing non-linear neural networks trained on structured datasets. Our goal is to open the black box that is self-supervised learning to develop a theory of SSL: to establish the precise conditions under which SSL beats supervised learning in terms of the input statistics, the learning task, and the network architecture; to analyze quantitatively and qualitatively the interplay between symmetries in datasets and the representations learned by self-supervised learning; and to relate SSL to existing unsupervised learning approaches.

Objectives

The goal of this project is to develop a quantitative theory for self-supervised learning, a machine learning technique whose application in natural language processing laid the foundation for the success of chatbots like ChatGPT.

Expected or Achieved Results

We have recently discovered that transformers learn distributions of increasing complexity, providing an intriguing perspective on the dynamics of self-supervised learning on a natural language processing task (Rende, Gerace, Laio & Goldt, NeurIPS 2024). Further works on self-supervised learning are currently on their way.

Description

The nature of dark matter (DM) remains one of the biggest mysteries of modern physics. In order to confront DM with experiments and to reveal its nature, it is crucial to explore various possibilities of DM interactions with the Standard Model particles. This project is dedicated to exploring new avenues of DM interactions. By investigating types of interactions that have not been analyzed in detail, we aim to shed new light on the nature of DM, propose new detection experiments, and explore connections to experimental anomalies. In particular, we focus on the following three research lines: (a) DM with a magnetic charge, (b) DM with a non-minimal coupling to gravity, and (c) DM–flavor physics connections. Each of the research lines are conducted by a team of astrophysicists and particle physicists, who bring together expertise in early universe cosmology, galactic physics, gravitational physics, collider physics, flavor physics, and ultraviolet completion of DM theories.

Objectives

This project is dedicated to exploring new avenues of dark matter interactions. By investigating types of interactions that have not been analyzed in detail, we aim to shed new light on the nature of dark matter, and propose new detection experiments.

Expected or Achieved Results

We discovered a novel mechanism for cosmological phase transitions where the nucleation to the true vacuum is induced by boson stars. This finding also has rich implications for studies of axion DM stars.

For DM with a magnetic charge, we placed novel constraints on the abundance by analyzing its backreaction on intergalactic magnetic fields, and also by using terrestrial cosmic ray experiments. The new results improve existing limits by more than an order of magnitude.

In the framework of DM that is non-minimally coupled to gravity, we investigated the existence and properties of static, spherically symmetric, and nonsingular solutions of the general relativistic field equations.

Description

The project combines methods from statistical mechanics, polymer physics, and machine learning to investigate the forces shaping genome architecture and their consequences for gene regulation, cell memory, differentiation, and disease.

Objectives

Our genome is not just a string of DNA—it is intricately folded in three dimensions inside the cell nucleus. This spatial organization is crucial for cellular function, as it influences how genes are turned on or off. DNA regions that are far apart along the chromosome can interact through physical contact, regulating gene expression in ways we are only beginning to understand.

This project aims to uncover the physical principles that govern the folding and interaction of chromosomes. By developing and applying advanced computer models, we aim to simulate the behavior of chromosomes in real cells. These models will integrate genetic and epigenetic data to provide a detailed and predictive understanding of genome folding. A key goal is to develop the first comprehensive framework that explains how both passive mechanisms (such as DNA-binding molecules) and active processes (like loop extrusion and DNA supercoiling) organize the genome in three dimensions.

This interdisciplinary project combines expertise in physics, computational biology, and molecular genetics of three different groups based in Naples, Padova, and SISSA, as well as collaborators from experimental labs across Europe and the United States. Our ultimate goal is to bridge fundamental physics with biomedical applications, thereby advancing both our understanding of genome organization and our ability to address genetic diseases.

Expected or Achieved Results

The project as a whole will help clarify how specific genetic mutations can disrupt normal chromosome structure and lead to disease. By predicting how these changes affect DNA contacts and gene regulation, our models can inform new strategies for understanding and potentially treating conditions such as cancer and congenital disorders. The SISSA group will gather results mostly regarding the geometrical and topological shaping of the chromosomes induced by SMC complexes. Particular attention will be paid to emerging concepts and methods that can be transferred to more general soft-matter contexts, such as the characterization and design of topology-based materials.

Description

The ever-increasing demand for high-speed, high-density, and low-power electronic components has stimulated the scientific community to identify new paradigms to achieve information processing and communication. Interesting scenarios have emerged from operating on degrees of freedom of the electrons in materials other than their charge, leading to spintronics, and, more recently, to valleytronics. In the latter, it is the momentum of one of the equivalent pockets (so-called valleys) in the Brillouin zone to which the electron/hole belongs that plays the role of a pseudospin variable. Although early investigations date back to the 1970s, the field has only recently been revolutionized by the discovery of two-dimensional (2D) materials, which offer unprecedented opportunities to control and manipulate valleys. Much of this success arises from spin-valley locking, by which different valleys are endowed with a specific spin polarization, enhancing both spin and valley coherence times. Three recent major breakthroughs have the potential to further expand the range of spin-valleytronic phenomena enabled by 2D materials: the isolation of 2D magnets, the identification of an extensive portfolio of novel monolayers from computational exfoliation, and the combinatorially vast opportunities offered by stacking 2D materials into van der Waals (vdW) heterostructures.

SECSY aims at capitalizing on these developments by combining the internationally recognized expertise of the partners in materials design, theoretical modeling, high-throughput calculations, and efficient device simulations towards the identification of promising materials platforms for spin-valleytronics, addressing in particular the pressing need to achieve a fully electrical control of the locked spin and valley degrees of freedom. Two classes of systems will be investigated: single layer antiferromagnets and magnet-semiconductor-magnet vdW heterostructures. In the first case, spin-valley locking arises either from spin-orbit coupling or from a recently discovered subtle interplay between exchange interactions and crystal fields. We will extend the design principles to identify such altermagnets also in 2D, and explore materials’ databases through accurate first-principles simulations to find viable materials realizations. In the second case, spin and valley are coupled through a valley-Zeeman effect in the semiconductor, further affected by the proximity to the magnetic layers. Materials combinations will be identified by data-mining and high-throughput calculations, eventually validated against many-body effects. In both cases, we envision realistic pathways to tune the spin-valley polarization through a vertical, gate-induced, electric field. The range of possible applications of these materials platforms will be investigated by multiscale device simulations that we plan to accelerate by developing an innovative formulation based on hybrid Wannier functions.

Objectives

SECSY aims to identify and design 2D materials for spin-valleytronic devices by discovering monolayer altermagnets and engineering magnetic van der Waals heterostructures, targeting full electrical control of spin-valley locking via ab initio simulations, high-throughput computing, and device modeling.

Expected or Achieved Results

[To be filled once results are available.]

Description

The ever-increasing demand for high-speed, high-density, and low-power electronic components has stimulated the scientific community to identify new paradigms to achieve information processing and communication. Interesting scenarios have emerged from operating on degrees of freedom of the electrons in materials other than their charge, leading to spintronics, and, more recently, to valleytronics. In the latter, it is the momentum of one of the equivalent pockets (so-called valleys) in the Brillouin zone to which the electron/hole belongs that plays the role of a pseudospin variable. Although early investigations date back to the 1970s, the field has only recently been revolutionized by the discovery of two-dimensional (2D) materials, which offer unprecedented opportunities to control and manipulate valleys. Much of this success arises from spin-valley locking, by which different valleys are endowed with a specific spin polarization, enhancing both spin and valley coherence times. Three recent major breakthroughs have the potential to further expand the range of spin-valleytronic phenomena enabled by 2D materials: the isolation of 2D magnets, the identification of an extensive portfolio of novel monolayers from computational exfoliation, and the combinatorially vast opportunities offered by 2D stacking materials into van der Waals (vdW) heterostructures.

Objectives

Our goal is to introduce novel approaches to solve hard optimization problems, where traditional Quantum Annealing fails. Hybrid algorithms like the Quantum Approximate Optimization Algorithm are ideal for that purpose. Applications to frustrated spin models are currently pursued.

Expected or Achieved Results

[To be filled once results are available.]

Description

The project developed an end-to-end platform for the joint inference of the properties of the first galaxies, physical cosmology, and instrumental systematic effects, applied to 21cm tomography of the Cosmic Dawn and Epoch of Reionization (EoR). The main objectives were: modeling astrophysical and cosmological uncertainties and their effects on the 21cm signal; integrating uncertainties from instruments and foregrounds; and developing a Bayesian pipeline for joint data analysis, in preparation for upcoming experiments like SKA, EDGES, and HERA. The project included the development of realistic physical models for the global 21cm signal, new emulation tools for fast inference, and machine learning methods for robust estimation of photometric redshifts.

Objectives

Maximize the cosmological and astrophysical information extractable from 21cm tomography by developing a statistical inference pipeline that integrates physical models, efficient simulations, instrumental uncertainties, and advanced machine learning methods.

Expected or Achieved Results

The project produced three main results:

1. Exclusion of a radio background excess during the cosmic dawn: In Cang et al. 2024, a realistic physical model for the global 21cm signal including an extra radio background from Pop III star-forming galaxies was developed. Bayesian comparison with EDGES data showed that models predicting an anomalous signal are strongly disfavored, contradicting previous interpretations based on approximate signal shapes. The associated code and emulator have been publicly released.

2. Improvement in photometric redshift estimation in the presence of selection bias: In Moretti et al. 2025, StratLearn was applied to photo-z estimation, demonstrating robustness against covariate shift effects compared to standard algorithms (e.g., GPz), achieving a significant reduction in catastrophic errors and up to an order-of-magnitude improvement in bias. The Julia code has been made available as open source.

3. Development of a fast 21cm simulation emulator: Breitman et al. 2024 introduced 21cmEMU, capable of emulating six 21cm observables with sub-percent accuracy, reducing computational times by over 10⁴ times. The emulator was integrated into 21CMMC and used to study the constraining power of various datasets and to produce forecasts for HERA.

These results provide concrete tools for the statistical analysis of SKA/HERA data and support evidence-based decisions for experimental design, strengthening the role of the Italian community in the SKA project.

Expected Results

The SKA Data Challenge aims to prepare the radio astronomy community for the unprecedented nature of data from the future SKA telescope, providing realistic synthetic datasets of the 21cm power spectrum generated by two simulators: 21cmFAST (fast, CPU-based) and C2Ray (more accurate but much more computationally expensive, GPU-based). Within the project, three teams were formed to participate in the SKA Data Challenge with different approaches to inferring the 21cm signal for studying the Epoch of Reionization. The following were generated:

• A dataset of 50,000 21cmFAST simulations to train a fast emulator.
• A C2Ray dataset of 1,800 simulations (targeting 5,000) to correct biases from the first emulator.

The emulator enabled high-confidence reconstruction of the reionization history and identification of the simulator used in the datasets. Soon it is expected to achieve:

• Completion of debiasing to estimate expected results from C2Ray,
• Finalization of the ToSKA-explicit submission,
• Progress in SBI submissions with multiple simulations for more robust inference.

Thanks to the complementary and advanced nature of the three planned submissions, excellent results are expected in the Data Challenge.

Description

Geometric Analysis is by nature a field at the intersection of different branches of mathematics. As such, it attracts the interest of specialists with diverse expertise. Our group reflects this fact, as it includes colleagues with backgrounds ranging from analysis on non-smooth spaces to geometric flows, from Lorentzian geometry to optimal control problems.

These complementary perspectives share a common research interest: the desire to study singular geometric structures as a means to better understand more regular ones. Here, “structure,” “singular,” and “regular” are purposely vague terms, as our interests span broad areas of knowledge. For instance, we study metric measure spaces with weak curvature bounds as tools to understand the degeneration of smooth Riemannian metrics; as another example, geometric evolutions tend to develop singularities, which we study with the aim of capturing the geometry of evolving shapes.

Objectives

[To be filled once objectives are defined or summarized.]

Expected or Achieved Results

E. Minguzzi and S. Suhr. Lorentzian metric spaces and their Gromov-Hausdorff convergence. Lett. Math. Phys., 114 (2024) 73. arXiv:2209.14384. 
A. Bykov and E. Minguzzi and S. Suhr. Lorentzian metric spaces and GH-convergence: the unbounded case. arXiv:2412.04311. 
R. A. Hounnonkpe and E. Minguzzi. Horizon classification via Riemannian flows arXiv:2410.07231 
A. Bykov and E. Minguzzi. Global hyperbolicity and manifold topology from the Lorentzian distance. arXiv:2503.04382 
 

Description

The present proposal aims to study nonlinear systems originating from classical and quantum physics that have the property of being exactly solvable or integrable in some asymptotic regime. For 50 years, integrability has continuously staked claims in an ever-growing range of applications in mathematics and physics, and integrable or near-to-integrable models keep emerging in important areas of Classical and Quantum Physics such as Condensed Matter and Plasma Physics, Optics, Gravity, Statistical Physics, and String Theories.

The modern theory of integrable systems grew around the study of the Korteweg-de Vries (KdV) equation, with origins in the seminal work of Kruskal and Zabusky on the recurrence behavior of solutions, the discovery of the Lax pair, multi-soliton solutions, bi-Hamiltonian structures, and the infinite number of conservation laws. Later, surprising connections showed that integrable systems like the KdV equation and the Toda lattice appear in combinatorial models, random matrices, and the geometry of moduli spaces of algebraic curves.

In general, integrability provides a route to an explicit description of behavior and phenomena that turn out to be ubiquitous in nonlinear systems. The idea that integrable behavior persists in non-integrable systems, together with the combination of cutting-edge geometrical and analytical techniques in the theory of Hamiltonian equations, is the leitmotif of this research project, whose main aims can be summarized as follows:

1) To study the theory of soliton gases, and substantiate the features of the “unlikely marriage” of integrability and randomness motivated by the complexity of many nonlinear wave phenomena;
2) To give new insights into the fundamental class of wave-induced dynamics in the theory of stratified fluids and vorticity by linking their dimensionally reduced models to the theory of integrable PDEs;
3) To use spectral methods and Riemann-Hilbert methods to describe solutions for physically relevant initial/boundary value problems of nonlinear PDEs;
4) To study nonlinear wave phenomena, such as the formation and dynamics of singularities, blow-ups, wave breaking, shocks, elliptic/hyperbolic regime transitions, dispersive/dissipative regularization;
5) To study geometric and algebraic properties of integrable Hamiltonian and bi-Hamiltonian systems arising in mathematical and physical models, together with their hierarchies of symmetries/conserved quantities.

Objectives

The present proposal aims to study nonlinear systems originating from classical and quantum physics that have the property of being exactly solvable or integrable in some asymptotic regime. The outcome is a very explicit description of the behavior of models originating from physics and geometry.

Expected or Achieved Results

We have studied nonlinear dispersive integrable equations with a large number of solitons. Solitons are the strongly nonlinear component of the initial data that survive in the long-time asymptotics. In the strongly nonlinear regime, when the initial data is not known, it is well approximated by a large set of random solitons. We obtained properties of solutions of nonlinear dispersive equations in the limit of an infinite number of solitons and derived a central limit theorem in the case of a random soliton gas. We obtained similar results for a gas of breathers.

In the setting of hydrodynamic equations, namely quasi-linear hyperbolic equations, we have obtained a generalized hodograph method for non-diagonalizable integrable systems of hydrodynamic type.

Description

Turbulence is an ubiquitous phenomenon in nature, experienced daily in many situations, from waves breaking on a shore to dust settling. Despite its pervasiveness, its mathematical interpretation can be described as “a riddle wrapped in a mystery inside an enigma.” This elusiveness lies in the very structure of the main hydrodynamical models, which are nonlinear equations that give rise to very rich and nontrivial dynamics. Despite these evident limitations, there are limited settings in which we can describe turbulent phenomena, or rule them out. These can mostly be formalized as results of stability and its negative counterpart, instability, in mathematical terms.

The main aim of this research proposal is to provide several new settings describing stable and unstable configurations in incompressible fluid dynamics. The results we propose to investigate will be studied using a mix of techniques recently developed in outstanding scientific publications and new implementations by members of the research teams. Specifically, we wish to methodically bridge techniques originally developed in the context of dynamical systems (normal forms, KAM theory, pseudodifferential calculus) to practical problems of fluid dynamics, and vice versa. All of this will be implemented using sophisticated tools of harmonic and complex analysis, such as pseudo- and para-differential calculus and other tools originally developed in nonlinear wave theory. These methods have already been successfully applied in the past 20 years, experiencing a renaissance in the classical topic of incompressible fluid dynamics.

Each member participating in the project is under 40 years of age and holds a position at an Italian institution. This designs a cooperation among the main members of the research program that spans deeply into the future, potentially laying the foundations for developing a school of incompressible fluid dynamics indigenous to the Italian university system. The proposal is thus conceived to jump-start this cooperation and enhance interactions among the various members of the group.

Objectives

[To be filled once objectives are defined or summarized.]

Expected or Achieved Results

[To be filled once results are available.]

Description

Current progress in engineering fields, such as material science and robotics, is achieved through the study and understanding of biological systems. These embody some form of intelligence, due to the effectiveness of natural selection in pursuing optimality principles. Capturing such principles is crucial for the success of bioinspired design. Building upon these premises, the research project aims at exploring the behavior of plants to extract the principles that have driven their evolution and underline their functioning, and to transfer them to synthetic biomimetic replicas based on soft active materials, such as polymer gels and liquid crystal elastomers. To achieve this ambitious goal, the research program encompasses modeling, computational, and experimental activities, carried out at the SAMBA and BioMat laboratories of SISSA.

Objectives

By combining mathematical modeling, scientific computing, and physical experiments, the research project aims at unveiling the biophysical principles underlying the functioning of plants and transferring them to the design of innovative synthetic replicas based on soft active materials, driving innovation in soft robotics.

Expected or Achieved Results

Building upon recent findings by the research group on plant shoot morphogenesis, the activities of DISCOVER have mainly focused on studying the mechanics of active filaments based on ionic polymer gels. Suitable mathematical models, based on morphoelasticity theory, have been developed and combined with physical experiments on small-scale prototypes to show that the exploitation of mechanical instabilities in soft biomimetic structures (e.g., artificial cilia and robotic swimmers) allows for their untethered control under minimal actuation. This represents a new avenue for innovation in the fields of micro-robotics and micro-fluidics, and the current research activities aim at extending this approach to complex morphoelastic structures based on active materials and soft pneumatic actuators, with applications in a variety of technological fields.

Team: Alessandro Giuliani (Roma 3), Vieri Mastropietro (Sapienza), Marcello Porta (SISSA)

Description

The project is devoted to the rigorous study of lattice quantum field theory (QFT) models. On one hand, these are relevant for condensed matter physics, where the lattice is naturally provided by the crystalline structure of solid-state systems, and the interest lies in the large-scale properties of the system at fixed lattice spacing. On the other hand, lattice quantum field theories are important for high-energy physics, as effective descriptions of more fundamental theories. The project aims at developing new mathematical methods for lattice QFTs, with relevance for both condensed matter and high-energy physics.

Objectives

The project aims to introduce rigorous analytical methods to study lattice gauge theories and to prove their renormalizability. Another objective is the non-perturbative construction of low-dimensional QFTs. On the condensed matter side, the project focuses on the study of topological and semimetallic states of matter, and on the effect of interactions and disorder.

Results

The project achieved the following results:

- Computation of edge transport coefficients and edge correlation functions for Haldane-like models in the presence of quasi-periodic disorder.

- Derivation of the large-scale dynamical response of 1D systems and topological edge modes of 2D systems, emergent bosonization, and validation of linear response theory.

- Characterization of the ground state of the 2D Z2-gauge theory coupled to fermionic matter, emergence of the pi-flux phase, and proof of semimetallic behavior.

- Rigorous derivation of the anomalous exponents and scaling operators for an interacting fermionic QFT with a non-trivial renormalization group fixed point.

- Construction of the scaling limit for the boundary correlations of non-integrable perturbations of the 2D Ising model.

- Determination of the critical Hall conductivity for the interacting Haldane model.

A next goal of the project is to apply the methods developed for the analysis of interacting edge modes to rigorously understand the lattice regularization of chiral gauge theories.

Description

The FaReX project aims at developing new numerical methods for the efficient simulation of complex multiscale and multiphysics problems, such as fluid-structure interaction or phase transitions, using non-conforming (non-matching) modeling techniques, immersed methods, and model reduction techniques, including machine learning approaches. The involved units (SISSA, PoliTO, UniBS) work jointly on numerical analysis, complexity reduction, and open-source software development.

Objectives

The goal of the FaReX project is to reduce computational complexity and enable efficient simulations of complex problems, with applications in cardiovascular medicine, optimization of manufacturing processes, environmental remediation, and geothermal energy production.

Expected Results

The main expected results include the development of new numerical methods for complex geometric domains, the reduction of computational complexity through reduced-order models and machine learning techniques, and the development of open-source software (for example, through the extension of libraries such as ITHACA-FV, RBniCS, and PyGeM). A dissemination effort of the results within the scientific community is also planned.

Description

Neurological disorders represent a leading cause of disability worldwide, with a significant impact on patients’ quality of life. Recent evidence indicates that extracellular vesicles (EVs) derived from stem cells play a key role as mediators of neuroprotection and neuroregeneration, providing a promising avenue for developing new therapies for neurological diseases. The EXAMSPAIN project aims to investigate the molecular mechanisms through which EVs exert their protective effects on neural cells, and to identify specific EV-derived factors responsible for these actions. By combining advanced molecular biology, electrophysiology, and imaging techniques, the project seeks to elucidate the potential of EVs as therapeutic tools for neuroprotection and to explore their use in experimental models of neurological disorders.

Objectives

The project is organized into three work packages (WPs):
- WP1: Isolation, characterization, and engineering of extracellular vesicles from stem cells.
- WP2: Investigation of the neuroprotective and regenerative properties of EVs in in vitro and in vivo models of neurological diseases.
- WP3: Identification of EV-associated molecular factors responsible for their therapeutic effects and exploration of their potential as biomarkers or drug candidates.

Expected or Achieved Results

The project is expected to provide new insights into the neuroprotective role of stem cell-derived EVs, identify specific bioactive molecules within EVs with therapeutic potential, and lay the groundwork for innovative treatments targeting neurological disorders. The research will contribute to developing advanced strategies for neuroprotection and regeneration, with the ultimate goal of improving outcomes for patients affected by neurodegenerative and neuroinflammatory diseases.

Description

“AICult: Artificial Intelligence with Cultured Neuronal Networks” is a multidisciplinary project at the intersection of neurodevelopmental genetics, neurophysiology, and in silico modeling of neuronal network dynamics, implemented at CNR-Pisa, SNS-Pisa, and SISSA-Trieste. The project aims to exploit real biological neurons as building blocks for advanced artificial computation.

Objectives

The aim of this project was to develop solutions for the use of real neuronal cells, either normal or carrying mutations associated with cognitive diseases, for AI purposes. The goal of the SISSA team was to unravel the control of neuronal integrative properties by the Foxg1 transcription factor gene.

Expected or Achieved Results

First, using primary cultures of murine pallial cells engineered with recombinant lentiviruses and analyzed with a genetic calcium sensor, the SISSA team revealed the impact of subtle modulation of Foxg1-mRNA levels on activity profiles of these cultures, focusing on neuronal excitability and refractoriness.

Second, through comparative transcriptomic and proteomic profiling of these cultures, the team identified selected gene sets potentially mediating Foxg1's impact on neuronal activity.

Third, using TRAP2;Ai9 mouse reporters of neuronal activity engineered with recombinant adeno-associated vectors to subtly perturb in vivo Foxg1 activation levels, the SISSA team uncovered Foxg1’s role in controlling neuron recruitment into real engrams associated with specific fear experiences.

Altogether, these results highlight Foxg1 as a powerful tool for controlling neuronal activity, exploitable for the fine modulation of artificial computation embodied in cultures of biological neurons.

Description

The project aims to investigate the role of hippocampus-dependent memory processes in visual perceptual learning. By combining advanced neuroimaging techniques, computational modeling, and behavioral experiments, the research seeks to elucidate how memory mechanisms supported by the hippocampus contribute to improvements in visual discrimination tasks and perceptual decision-making. Understanding these processes has significant implications for cognitive neuroscience, particularly regarding learning, memory consolidation, and rehabilitation strategies for visual impairments.

Objectives

To identify and characterize the neural and cognitive mechanisms through which hippocampus-dependent memory influences visual perceptual learning, and to develop predictive computational models of these interactions.

Expected or Achieved Results

The project is expected to reveal the contributions of hippocampal activity patterns to performance improvements in visual tasks, to map brain regions engaged during perceptual learning, and to provide insights into how memory systems interact with sensory processing networks. These findings could inform novel training protocols and therapeutic approaches for enhancing visual perceptual abilities.

Description

This project investigates how individual differences shape brain plasticity in the context of both healthy and pathological aging. Although age-related cognitive decline is well documented, its manifestation varies widely across individuals. For example, some older adults without evident cognitive impairment meet post-mortem criteria for Alzheimer’s disease, while others with mild-to-moderate neuropathology do not develop dementia. Brain plasticity refers to the brain’s capacity to reorganize, adapt, and form new neural connections, enabling adaptive responses to learning, environmental changes, injury, and cognitive or sensory challenges. While plasticity occurs throughout the lifespan, the factors determining its extent and variability across individuals—and their interactions—remain poorly understood.

Objectives

The aim of this project is to examine how individual differences in cognitive reserve, personality traits (e.g., Big Five), socio-emotional skills (e.g., emotion regulation, empathy), and domain-specific expertise (e.g., mathematics) contribute to neuroplasticity and shape cognitive outcomes in healthy and pathological ageing. Using neuropsychological assessments and neuroimaging (e.g., EEG), the project investigates how cognitive, numerical, and social functioning are shaped by these individual-level variables across different groups. By adopting a lifespan and transdiagnostic approach, the project aims to identify shared and condition-specific mechanisms of neural adaptation, ultimately defining transdiagnostic profiles of resilience that could support early identification of individuals at risk for accelerated cognitive decline and inform personalized prevention strategies.

Expected or Achieved Results

A first study (Montemurro et al., 2025) showed that higher cognitive reserve mitigates the impact of structural brain changes on cognitive performance, allowing individuals with higher reserve to maintain executive and memory functions over time. Building on these findings, the project includes four additional studies targeting different populations: brain tumor patients, older adults with type 2 diabetes, young adults with autism spectrum disorder, and healthy older adults with or without mathematical expertise. Papers related to the project include:

• Montemurro, S., Rumiati, R.I., Pucci, V., Nucci, M. & Mondini, S. (2025). Cognitive Reserve can impact Trajectories in Ageing: A Longitudinal Study. Aging Clinical and Experimental Research, 37(1):93. doi: 10.1007/s40520-025-03000-z.
• Pisanu, E., Zanon, M., Stea, C., Collari, L., Riolo, A., Rumiati, R.I. (in preparation). Does ASD Individuals’ Personality Affect the Decoding of Mental States?
• Sichani, M.H. & Rumiati, R.I. (in preparation). Integrating Numerical and Cognitive Abilities as Proxies for Cognitive Reserve in the Elderly.

Description

Current efforts to reconnect nerve structures or integrate neural bioelectronic interfaces often fail due to poor tissue adherence, significantly impacting the restoration of pulse conduction along peripheral nerves after neuropathies. This has profound socioeconomic effects on quality of life and productivity. The proposal aims to develop an innovative conductive and adhesive substrate to enhance charge transfer between electronic interfaces and neural tissue. We hypothesize that nontoxic Knoevenagel-type indole (KtI) derivatives with highly delocalized electrons, dispersed in biocompatible adhesives, can facilitate signal conduction.

Objectives

The project requires a multidisciplinary team, incorporating advanced chemical synthesis, formulation, and detailed biophysical, histological, and toxicological analyses, to achieve the following objectives:

• Synthesize KtI derivatives with high electron motility.
• Formulate an adhesive substrate to cross-link or disperse KtI derivatives.
• Assess pulse conduction along the implant.
• Electrochemically characterize the new materials.
• Mechanically characterize the adhesive under loading conditions.
• Measure the implant's effects on nerve health and microanatomy through histological analysis.
• Determine the biocompatibility of the new materials in the peripheral and central nervous systems.

Expected or Achieved Results

Electrophysiological in vitro recordings have shown proof of concept for restoring signals across transected nerves using a conductive adhesive implant. Collected data suggests the potential for integrating electronics with neural tissue using a KtI conductive adhesive implant. The findings of the project could lead to a clinical study for peripheral nerve lesion recovery and the patenting of the first conductive adhesive biomaterials for the surgical sealants market.

Description

This project investigates the sensitivity of sensory systems to the statistical properties of natural stimuli, exploring how such sensitivity emerges through evolutionary mechanisms, postnatal development, and the existence of critical periods that shape efficient neural coding. By understanding these factors, the project aims to clarify how brains become finely tuned to process the specific regularities of sensory inputs encountered in natural environments, optimizing perceptual performance and cognitive functions.

Objectives

The project seeks to identify the biological and computational mechanisms that enable the neural circuits of visual and auditory systems to adapt to the statistics of natural scenes and sounds.

Expected Results

Description

The project focuses on fluid-structure interaction (FSI) in the cardiovascular system, aiming to develop innovative numerical methods supported by machine learning techniques to achieve fast and accurate solutions. It tackles three main challenges: the stability of weakly coupled numerical schemes, the reduction of computational complexity through reduced-order models (ROM), and the efficient solution of inverse problems (such as shape optimization and parameter calibration) relevant to clinical practice.

Objectives

The project aims to improve the efficiency and accuracy of cardiovascular simulations, with applications such as predicting the formation of arterial plaques and aneurysms, and supporting surgical planning (e.g., for bypass procedures).

Expected Results

The project aims to develop mathematical and computational tools for use in real clinical contexts, including:

• Numerical strategies to improve the stability of weakly coupled FSI models.
• Machine learning techniques for automatic calibration of critical parameters in numerical models.
• Reduced-order models enabling real-time simulation capabilities.

Other expected outcomes include the efficient solution of inverse problems and the release of open-source software (e.g., through extensions of libraries like lifeX, ITHACA-FV, RBniCS). The project also includes dissemination of the results within the scientific community.

Description

Academic resilience is especially important for low-income students who often face systemic challenges such as financial hardship, limited access to educational resources, unstable living conditions, and psychosocial stressors. Despite these obstacles, many students show remarkable ability to thrive academically and emotionally.

Self-determination theory (SDT) offers a framework for understanding how such resilience can be cultivated, highlighting three key psychological needs: autonomy, competence, and relatedness. Supporting these needs through meaningful choices, constructive feedback, and strong relationships can enhance motivation and engagement, even in disadvantaged contexts.

This project aims to understand how these psychological needs affect academic resilience, performance, and well-being in low-income students, challenging deficit-based narratives by emphasizing strengths and adaptive strategies. It includes two complementary studies:

• A cross-sectional study involving over 1,000 Italian students to identify malleable motivational characteristics that support resilience.
• A longitudinal study following more than 200 low-income students over time to analyze how motivational variables evolve and how parents, teachers, and peers influence these changes.

The findings aim to inform sustainable policies and interventions to foster inclusion and well-being among disadvantaged youth.

Related Publications

Bersier, N., Arbula, & Rumiati, R.I. (2025). Understanding sex differences in mental rotation through the role of strategies. Submitted to Journal of Experimental Psychology: Human Perception and Performance (PRIN-PNRR).

Lunardon, M., Cerni, T., & Rumiati, R.I. (2025). The longitudinal relationship between math anxiety and numeracy in higher education (submitted) (PRIN-PNRR).

Zacchilli, M., Fornaro, S., Raimondi, G., Dawe, J., Manganelli, S., Lucidi F., Cavicchiolo, E., Rumiati, R.I., & Alivernini, F. (2025). Flourishing in the Moment: validation of the EduFlow-2 Scale for Assessing Adolescent Flow in the Italian Context (submitted) (PRIN-PNRR).

Fornaro, S., Cavicchiolo, E., Rumiati, R.I. (2025). Resilience, buoyancy and coping in education: a systematic review of formal constructs and measures (in prep.) (PRIN-PNRR).

Fornaro, S., Lunardon, M., Cavicchiolo, E., & Rumiati, R.I. (2025). The role of socio-economic and socio-demographic factors in academic resilience of university students (in prep) (PRIN-PNRR).