Room temperature superconductors are probably the most desired systems in solid state physics because of their energy implications: superconductors offer conductivity without energy loss. While many materials can reach this state, only those with a sufficiently high critical temperature (Tc) will have technological applications. The recent discovery of hydrogen-rich materials room temperature superconductors has set up a quest in this direction. However, since theoretical models do not offer today sufficient conditions to find such materials most of the new discoveries have been based on trial-and-error (expensive) experiments. Our recent discovery of the networking value [Nat. Comm 12, 5381 (2021)], should allow us to find new superconducting materials through a cheap estimation of Tc from chemical and electronic structure features, thus avoiding expensive electron-phonon coupling calculations. TcPredictor aims at applying this new index in order to find new binary and trinary superconducting materials with industrial and transportation applications. To make this project possible, we have formed an interdisciplinary consortium (chemistry, physics, computer science): specialists in electronic structure and superconductivity to cover the theoretical aspects, computer engineers to search for new methods of acceleration by machine learning and finally, specialists in phase prediction to assemble the results and propose new room temperature superconductors. The project will be naturally structured in the following WPs: WP1 will aim at improving the performance of the networking value by including bi-electronic and anharmonic effects, WP2) will aim at accelerating the calculation to make it accessible for high-throughput screening, and WP3) will use the newly improved and accelerated index to predict new superconductors.

The MUSTER project is a fundamental pilot research project which introduces a new multi-modal framework for the machine-readable representation of meaning. The focus of MUSTER lies on exploiting visual and perceptual input in the form of images and videos coupled with textual modality for building structured multi-modal semantic representations for the recognition of objects and actions, and their spatial and temporal relations. The MUSTER project will investigate whether such novel multi-modal representations will improve the performance of automated understanding of human language. MUSTER starts from the current state-of-the-work platform for human language representation learning known as text embeddings, but introduces the visual modality to provide contextual world knowledge which text-only models lack while humans possess such knowledge when understanding language. MUSTER will propose a new pilot framework for joint representation learning from text and vision data tailored for spatial and temporal language processing. The constructed framework will be evaluated on a series of HLU tasks (i.e., semantic textual similarity and disambiguation, spatial role labeling, zero-shot learning, temporal action ordering) which closely mimic the processes of human language acquisition and understanding. MUSTER will rely on recent advances in multiple research disciplines spanning natural language processing, computer vision, machine learning, representation learning, and human language technologies, working together on building structured machine-readable multi-modal representations of spatial and temporal language phenomena.

The huge progress achieved in the manipulation of quantum systems is opening novel routes towards the generation of realistic quantum-based technology. Notably many counterintuitive manifestations of quantum mechanics are turning to be key features for next generation devices, whose performances will beat those of classical machines. Atom interferometry is a hallmark example of that. According to quantum mechanics particles can behave like waves, showing interference as well as light does. In addition, they are very sensitive to the surrounding environment and they have mass, which make of them extremely powerful sensors for measuring linear accelerations and rotations. Implementing reliable atom interferometers for practical applications is however still challenging. State-of-the-art devices are based on atomic samples which are manipulated while they fall due to gravity inside a vacuum apparatus. These interferometers are currently reaching their ultimate performances being limited by technical issues. Their ultimate sensitivity depends in turn on the time available for the interrogation and on the finite atom number. An immediate solution to improve the sensitivity consists in enlarging the interrogation area, at the expenses of the size of the device, and increasing the atom number, at the expenses of the spatial resolution of the atomic probe. To obtain high sensitivity while maintaining the devices compact, a new generation of interferometers based on trapped and guided atoms is emerging. These devices have several advantages: the atoms do not fall and the interrogation time can be long, the use of BECs guarantees micrometrical spatial resolution, and interatomic interactions allow for the preparation of entangled states surpassing the standard quantum limit set by the finite atom number. New challenges also arise: the effects of the confining potentials and interatomic interactions must be controlled at a metrological level. The proposed project aims at realizing novel BEC-based quantum sensors which will be able to surpass the limitations of current trapped and guided interferometers by combining some of the most powerful manipulation techniques currently available in the field of ultracold atoms (and beyond). The two key elements are the accurate tailoring of the optical potentials by a spatial light modulator, and the control of the interactions. This exceptional experimental control will be assisted by theoretical optimization such as short-cut-to-adiabaticity and optimal control techniques. In most atom interferometers to date, the beam splitters are realized by pulsing two laser beams in Bragg or Raman configuration. We will instead engineer innovative splitters directly integrated into the optical waveguides which confine the atoms. They can operate continuously and without the need of extra laser beams. All the elements of the interferometer (beam splitter, phase accumulation and recombiner) will be integrated into the same device by properly sculpturing one single laser beam. First, a complete Mach-Zehnder operation will be performed with a condensate with tunable interactions. A negligible or weakly attractive value of the interactions will be used to suppress interaction-induced decoherence or create dispersionless wavepackets. As a result, high sensitivities are expected for such interferometer. In a second phase of the project, we will demonstrate a Sagnac-like interferometer with non-interacting condensates propagating in a close circuit. This will realize a guided atom gyroscope whose achievement has been a long-standing goal, and which finds an important application in inertial navigation. Finally, we will generate mesoscopic optical tweezers for realizing a dynamical double-well potential for Mach-Zehnder interferometry. By moving the tweezers apart we will control the coupling between the two wells, and by setting strong repulsive interactions we will produce optimally spin-squeezed states.

The MUSTER project is a fundamental pilot research project which introduces a new multi-modal framework for the machine-readable representation of meaning. The focus of MUSTER lies on exploiting visual and perceptual input in the form of images and videos coupled with textual modality for building structured multi-modal semantic representations for the recognition of objects and actions, and their spatial and temporal relations. The MUSTER project will investigate whether such novel multi-modal representations will improve the performance of automated understanding of human language. MUSTER starts from the current state-of-the-work platform for human language representation learning known as text embeddings, but introduces the visual modality to provide contextual world knowledge which text-only models lack while humans possess such knowledge when understanding language. MUSTER will propose a new pilot framework for joint representation learning from text and vision data tailored for spatial and temporal language processing. The constructed framework will be evaluated on a series of HLU tasks (i.e., semantic textual similarity and disambiguation, spatial role labeling, zero-shot learning, temporal action ordering) which closely mimic the processes of human language acquisition and understanding. MUSTER will rely on recent advances in multiple research disciplines spanning natural language processing, computer vision, machine learning, representation learning, and human language technologies, working together on building structured machine-readable multi-modal representations of spatial and temporal language phenomena.

This project focuses on implications for linguistic theory of a phenomenon found in some Basque dialects known as allocutivity—morphological agreement with non-participant addressees. Current syntactic literature is besotted with this phenomenon for the privileged access it offers into syntactic representations of Speaker and Hearer speech act roles, which are encoded syntactically in all human languages. Basque is special as a laboratory variety for exploring the syntax of speech act roles in that no other language so far described approximates Basque in terms of (i) the direct morphological evidence that it offers about the relationship among allocutivity, thematic addressees and vocative expressions, and (ii) its rich patterns of cross-dialectal/cross-speaker variation revealing loci of formal variation. The principal goal of this project is to describe three facets of this phenomenon wholly ignored in extant formal and psycholinguistic descriptions: (i) the nature of cross-dialectal variation and change in the syntax of allocutivity; (ii) the relationship of allocutivity to vocative expressions across dialects; and (iii) differences in grammatical constraints across levels of politeness marking. The project team —spanning diverse fields of expertise and from several different institutions— will gather two kinds of data to address these issues: survey data through a web-based application (N ?500), and interview data with a smaller set of participants (N ?120). Data will be made publicly accessible through an online data dashboard and through summaries in peer-reviewed publications.