Integrated Calendar

Human memory is a multi-stage process that in real life cannot be easily quantified let alone predicted by any kind of mathematical model. Cognitive psychologists developed experimental paradigms to overcome the first problem by using randomly assembled lists of words or other items for recognition and recall. Results of these experiments can be precisely characterized, and we recently proposed a set of mathematical models that are based on simple assumptions that can be analytically solved and provide surprisingly accurate predictions tested on Amazon Mechanical Turk internet platform. The main innovation of this approach to modeling memory is that (i) it is based on a very small set of basic principles and has little to no free parameters and (ii) assumes deterministic processes underlying memory. In particular, our recall model results in the prediction with not a single free parameter, indicating full universality of this memory component. Our model for forgetting has one free integer parameter, and indeed our experiments show that different types of items exhibit different rate of forgetting. The most ambitious part of this project is to generalize the quantitative approach to memory to more meaningful material such as narratives. We are designing quantitative measures of performance in these experiments. Our preliminary results indicate interesting features of performance for meaningful material, in particular the recall is more structured and uniform across subjects. We believe that better understanding of memory processes with meaningful material will allow the future AI systems to achieve a better and more ‘human’ level of processing of natural language.

We are interested in how functionality emerges from interactions between biomolecules, and subsequently how these functions can be incorporated into materials.1 Instead of using sequences known in biological systems, we use unbiased computational2 and experimental3 approaches to search and map the peptide sequence space for specific interactions and functions, with a focus on side chain, instead of backbone interactions. The talk will explore how to program molecular order and disorder through side chain interactions in short peptides4, and how the conformations adopted by these peptides can be exploited to regulate interfacial assembly properties, and liquid-liquid phase separation. We will discuss chemo-mechanical peptide-crystals with connected soft and stiff domains, that change their properties upon changes in hydration states.5 The last part of the talk will focus on our progress in holistic study of mixtures of molecules that individually are simple and non-functional, but as components of complex interacting systems, however, they give rise to self-organization patterns that are dictated by the environmental conditions.6 Collectively, we expect to identify insights that allow the repurposing of nature's molecules to design new functions that currently are not known in biology.

Nuclear magnetic resonance (NMR) spectroscopy is a versatile tool for probing structure, dynamics, folding, and interactions at atomic resolution. While naturally occurring magnetically active isotopes, such as 1H, 13C, or 15N, are most commonly used in biomolecular NMR, with 15N and 13C isotopic labeling routinely employed at the present time, 19F is a very attractive and sensitive alternative nucleus, which offers rich information on biomolecules in solution and in the solid state. This presentation will summarize the unique benefits of solution and solid-state 19F NMR spectroscopy for the study of biomolecular systems. Particular focus will be placed on the most recent studies and on unique and important potential applications of fluorine NMR methodology.

I will describe the photonic approach to quantum computation, which is the only technology that has been originally designed to reach the massive scaling required for fault- tolerant universal computation (> 106 physical qubits). It combines topological error correction and measurement-based quantum computation, with the leading effort relying on massive-scale silicon photonics.
I will then describe how cavity-QED with single atoms allows deterministic photon-atom two qubit gates, which in turn can drastically simplify the road towards fault-tolerant photonic quantum computing and improve its scaling to even larger numbers of physical qubits.

Molecular crystals are supramolecules “par excellence"1 as they are macroscopic objects composed by millions of molecules periodically disposed and held together by non-covalent interactions with specific physico-chemical properties dictated by their architectures. This offers a vibrant solid-state chemistry playground to build organic solids with tailored functionalities, such as novel luminescent materials,2 solid-state molecular machines,3 and multicomponent crystals with complex topologies.4 The inherent dynamic nature of the weak intermolecular forces that are driving organic crystals self-assembly is also conferring adaptive responsiveness, e.g., mechanical reconfiguration and shape-memory effect, to this class of materials making them ideal building blocks for the design and synthesis of multifunctional crystalline systems that can be exploited as actuators, flexible single-crystalline optoelectronic devices, and self-healing materials.5