ERC FUNDED PROJECTS

Current control applications necessitate the treatment of systems with multiple interconnected components, rather than the traditional single component paradigm that has been studied extensively. The individual subsystems may need to fulfil different and possibly conflicting specifications in a real-time manner. At the same time, they may need to fulfill coupled constraints that are defined as relations between their states. Towards this end, the need for methods for decentralized control at the continuous level and planning at the task level becomes apparent. We aim here towards unification of these two complementary approaches. Existing solutions rely on a top down centralized approach. We instead consider here a decentralized, bottom-up solution to the problem. The approach relies on three layers of interaction. In the first layer, agents aim at coordinating in order to fulfil their coupled constraints with limited communication exchange of their state information and design of appropriate feedback controllers; in the second layer, agents coordinate in order to mutually satisfy their discrete tasks through exchange of the corresponding plans in the form of automata; in the third and most challenging layer, the communication exchange for coordination now includes both continuous state and discrete plan/abstraction information. The results will be demonstrated in a scenario involving multiple (possibly human) users and multiple robots. The unification will yield a completely decentralized system, in which the bottom up approach to define tasks, the consideration of coupled constraints and their combination towards distributed hybrid control and planning in a coordinated fashion require for new ways of thinking and approaches to analysis and constitute the proposal a beyond the SoA and groundbreaking approach to the fields of control and computer science.

1 498 729 €

Start date: 2015-03-01, End date: 2020-02-29

Dynamic interplay between histone modifications and DNA methylation defines the chromatin structure of the humane genome and serves as a conceptual framework to understand transcriptional regulation in normal development and human disease. The ultimate goal of this research proposal is to study the chromatin architecture during normal and malignant T cell differentiation in order to define how DNA methylation drives oncogenic gene expression as a novel concept in cancer research. T-cell acute lymphoblastic leukemia (T-ALL) accounts for 15% of pediatric and 25% of adult ALL cases and was originally identified as a highly aggressive tumor entity. T-ALL therapy has been intensified leading to gradual improvements in survival. However, 20% of pediatric and 50% of adult T-ALL cases still relapse and ultimately die because of refractory disease. Research efforts have unravelled the complex genetic basis of T-ALL but failed to identify new promising targets for precision therapy. Recent studies have identified a subset of T-ALLs whose transcriptional programs resemble those of early T-cell progenitors (ETPs), myeloid precursors and hematopoietic stem cells. Importantly, these so-called ETP-ALLs are characterized by early treatment failure and an extremely poor prognosis. The unique ETP-ALL gene expression signature suggests that the epigenomic landscape in ETP-ALL is markedly different as compared to other genetic subtypes of human T-ALL. My project aims to identify genome-wide patterns of DNA methylation and histone modifications in genetic subtypes of human T-ALL as a basis for elucidating how DNA methylation drives the expression of critical oncogenes in the context of poor prognostic ETP-ALL. Given that these ETP-ALL patients completely fail current chemotherapy treatment, tackling this completely novel aspect of ETP-ALL genetics will yield new targets for therapeutic intervention in this aggressive haematological malignancy.

958 750 €

Start date: 2015-07-01, End date: 2020-06-30

For the last 15 years, optics has undergone a remarkable evolution towards ever decreasing sizes, better integration in complex systems, and more compact devices readily available to mass markets. Whereas traditional optics is at the centimeter scale, newly developed techniques use nanoscale objects to control, guide, and focus light. From the capability to shape metallic and dielectric nanostructures has emerged the field of nanophotonics. Advances in nanophotonics offer the possibility to control the material’s optical properties to create artificial materials with electromagnetic properties not found in nature. Man-made 3D metamaterials have interesting fundamental aspects and present many advantages with respect to conventional devices. Unexpected effects have led to the development of interesting applications like high resolution lenses and cloaking devices. Inspired by this new technology, we have developed new 2D metamaterials. Our flat metamaterials (metasurfaces) are much simpler to manufacture than their 3D counterparts. By depositing a set of nanostructures at an interface, we can immediately control the light properties; unlike refractive optical components, the wavefront is modified without propagation. As of today, these interfaces are created using metallic nanostructures and work in the infrared. In this ERC, we plan to extend the concept of optical metasurfaces in the visible which is the most important wavelength range for applications. By combining with optically active semiconductors such as InGaAlN, we will add optical gain and modulation capability to the system to create new, efficient optoelectronic devices. The response of the metasurfaces is tunable by changing the environment surrounding the nanostructures. We will use this property to create ultrathin reconfigurable flat devices. Metasurfaces will be integrated with AlN/GaN to modulate light at high frequencies and further exploited to control polariton gases in solid state metasystems.

2 000 000 €

Start date: 2015-09-01, End date: 2020-10-31

Motivated by our recent progress in feedback information theory and its deep relations to stochastic dynamical systems, and inspired by natural phenomena such as bio-molecular interactions and human conversation, this research will explore the fundamental limits of information transfer via simple interaction. In the standard information theoretic framework, the problem of reliable communications is typically studied in an asymptotic unidirectional regime, where optimal performance is attained via complex codes employed over increasingly long time epochs. Here, we will investigate a markedly different paradigm where communicating parties are restricted to use simple finite-state rules to act and react on the fly. We will consider a broad spectrum of models ranging from feedback communications and two-way channels to multiuser setups and large homogeneous networks, and study measures of information transfer and dissipation, their relations to dynamical system contraction factors, and the fundamental tradeoffs between complexity and performance. While prominently theoretic, our investigation is expected to admit important practical applications and a cross-disciplinary impact. In communications, and especially in resource-limited systems such as wireless sensor networks where battery-life is a bottleneck, a breakthrough in the understanding of optimal interaction can lead to a paradigm shift in system design, yielding simpler, cheaper, more robust solutions. In Finance, where market behavior is a cumulative effect of local actions taken by individuals based on limited noisy observations, quantifying interaction and its relation to information propagation can enhance our ability to forecast and explain macro level phenomena. Finally, an information theoretic characterization of interaction in large networks can shed light on the underlying mechanisms governing various biological systems that are empirically amenable to cellular automata modeling.

1 323 875 €

Start date: 2015-03-01, End date: 2021-02-28