Our Fellows

The objective of short-term traffic forecast is to predict traffic conditions (usually street flows and travel times) on the road network and its evolution in the near future (e.g. up to one hour ahead). These forecasts can be used for various traffic analysis and applications, such as traffic signals optimization, routing emergency vehicles, fleet operation optimization, journey planners, etc. based on the prediction of how traffic conditions will evolve in the most probable way. From a spectrum of tools available today, machine learning (ML) techniques are being increasingly used in various research fields, with increasing success. Their effectiveness has recently been enhanced by the possibility of accessing huge amounts of data. The main objective of the project is the validation of techniques of artificial intelligence (AI), specifically ML, for the prediction of the evolution of typical and non-typical conditions in urban traffic. The intention is to exploit big data in transportation, combining traditional data sources, such as detector counts, incidents data, and huge amount of disaggregated data, such as floating car data (FCD) providing explicit vehicle trajectories, to extract spatial-temporal network-wide correlations. Obtaining a set of typical recurring traffic states on which to base the prediction using the historical data, and prediction itself capturing spatial and temporal effects of disturbances in traffic, are some of the main challenges in this research. The outcome of the project will make a foundation for outlining the best direction for the creation of systems that will be capable of providing robust traffic forecasts.

We study and develop an on-site medical tool for diagnosing heart attacks, using additive manufacturing. The cardiovascular disease is one of the worst cause of death all over the world; for example, in the year of 2015, it was responsible for nearly 20 millions of lives lost. Amongst these victims, a little less than half died of coronary heart disease, a major cause of heart attacks. It has been investigated less that concentration of certain proteins, troponin and myoglobin, increase in the blood flow of the people suffering from the heart attack. Using these proteins as indicators, so called biomarkers, a fast, accurate, and portable blood tester will be able to save our lives. Lab-on-a tip refers to a compact and integrated analtical tool which performs required medical and /or chemical processing within a tiny volume of the sample, here a drop of blood taken from the patient. Fine fabrication of channels and reactors with submillimeter dimension for the small amount of the sample can be realized with severeal lithography techniques. They allow us to build and test microchannesl that will selectively collect and accumulate the biomarkers that are swiftly detected by optical readout.

Ultimate understanding of physical phenomena of ionic transport through a solid-state nanopore can open the new pathway to a number of applications. Among them is the detailed mechanism of ion channels conductivity involved in signal transmission within the neuronal network.
Coulter counter scheme that is often used to study ionic transport through a 2D nanoporous membrane does not allow active modulation of the nanopore charge since it is provided by the variation of the solution pH. Recent progress in fabrication of the sub-nanometer pores in 2D materials revealed interesting nonlinear current-voltage characteristics (CVC) exhibiting Coulomb blockade that is defined by nonlinear CVC of MoS2 nanopores revealing the suppressed conductance area around V<0.4V. Suppression becomes stronger with growth of ion valence z but it is also strongly linked to charges hosted inside the pore. ln this proposal, I suggest to use a field effect transistor-based system containing a nanopore in conductive channel of the device to study correlation between the in-plane DC electronic transport and the ionic transport through the pore simultaneously for the sub-nanometer pores for the first time. ln my PhD I used the same concept in the study of conductive properties of the edge channels in a semiconductor nanostructures with charged biomolecules deposited. Changes in lateral and cross-sectional voltages provided information on the biomolecules that gated the edge channels. The proposed method will provide correlation between the in-plane electronic transport and ionic transport thus allowing us to achieve a steady read-out method for characterization of the ionic transport through the nanopore and will serve as a starting point for the experiments with active charge modulation.

Advances in technology enabled the monitoring and collection of real-time traffic and weather data in freeways. However, adaptation of existing research and traffic management to introduction of autonomous vehicles (AVs) has yet to be investigated. The proposed research aims at advancing the cur-rent knowledge by applying advanced statistical models and by considering AVs through simulations. This assumes a number of hypotheses (e.g. AVs will use a dedicated lane) and an extended modification of existing behavioral models. It will also be examined how the introduction of AVs will affect safety of existing traditional road users. To enhance pro-active safety, short-term traffic forecasting will also be used as input to perform road safety evaluation of traditional road users. This approach will also be tested for different AV penetration rates. The ultimate aim is to understand crash mechanism with and without AVs and propose a proactive management system. To achieve the aims, traffic, weather and crash data from Attica Tollway in Athens will be used. The objectives are particularly useful for real-time safety evaluation of freeways not only in Greece but also of other European countries. This is a unique opportunity as it is the first time that AVs are considered in real-time traffic management. Thus, the produced results will offer new and innovative insights for better understanding real-time road safety, but also practical implications by developing a proactive traffic management system based on real-time safety. This will be a great opportunity for the candidate to make an important step towards his professional development, academic career and his improvement of his research knowledge, as it will also enable him to go to proficient important analytical techniques.

Computed tomography is an important tool in various medical, engineering and academic applications. Think for instance about breast cancer research, noninvasive inspection of specialised equipment or materials science. The current standard for image reconstruction from tomographic data is filtered backprojection (FBP). This methods works very well when noise levels are low and the object can be scanned from all angles. However, there is an increasing demand for tomography in applications where this may not be true.
Consider for example medical applications, where there is an ever increasing desire to lower the dosage of x-rays. This leads to higher noise levels and consequently a breakdown of the FBP algorithm. In some science or engineering applications, such as petrochemical extraction for instance, objects can be scanned from limited angles due to heavy-duty holding clamps.
There is therefore an increasing need for reconstruction algorithms that are able to deal with these challenges. They need to be flexible, robust and reliable. We believe that algebraic reconstruction methods fit the bill. These methods are based on the more general linear algebra formulation, and are able to incorporate prior information. Using acceleration techniques, as well as thinking about high-performance computing implementations, we aim to make algebraic reconstruction methods competitive.

A sensor which can detect bacteria efficiently in the source of contamination as well as in physiological fluids of affected patients is key for preliminary diagnosis of pathogen infestation. The conventional techniques for detection of pathogens are time-consuming and require skilled labour for performing complex assays. Biosensing devices are therefore required for continuous monitoring to check for infestation in affected patients as well as in areas important to public health, for example- public water supply systems. The aim of MOFSense is to develop a low cost sensor capable of rapidly detecting and identifying multiple species of pathogens in a drop of sample from a single test. Most of the devices fabricated so far suffer from a number of issues resulting from the necessary modification of the sensor with biomolecules to carry out the sensing activity. The MOFSense project will circumvent this important issue by using Metal Organic Frameworks (MOFs) that will be highly selective towards various species of pathogens. The envisioned biosensor will allow rapid quantitative reading of the pathogen concentration without sample manipulation.

This project falls into the fields of scientific computing and mathematical models applied to biology. The main topic is the bifurcation structure analysis of partial differential equations. Starting from theoretical results in the context of reaction diffusion systems for competitive species and predator-prey interactions, we are interested in cross-diffusion systems obtained as the limit, when suitable parameters tend to zero, of linear diffusion “microscopic” models by the exploitation of different timescales (Quasi Steady State Approximation). With this approach, we end up with models with a reduced number of equations, that naturally capture basic and specific interactions and that are able to reproduce complex phenomena such as spatial segregation of species and pattern formation. The purpose of the project are detailed numerical continuation calculations for partial differential equations arising in biology, in particular we focus on cross-diffusion systems. The aim is to understand how the global structure of equilibrium solutions of the microscopic systems behaves as the time-scale parameter becomes smaller and if it converges to the one of the corresponding limiting system. A key computational tool is the MatLab continuation package pde2path, but also the development of new numerical techniques is going to be an additional interesting aspect of this project. The achieved results are also going to improve the theoretical study of these models, clarifying the role of the cross-diffusion terms as the key ingredients in pattern formation.

Metal halide perovskites have emerged in the last decade as revolutionary semiconductors for optoelectronic devices such as photovoltaics, light-emitting devices, photodetectors or lasers, among others. The astonished evolution of the certified efficiency of perovskite solar cells (PSCs) exceeding 22% power conversion efficiency (PCE) in only one decade after its first implementation in photovoltaics, it raises perovskites above other lights absorbers. The exceptional device performance of PSCs, the low-cost of the precursors and the variety of deposition techniques from solution to vacuum indicate a promising future toward their commercialization. However, this new technology needs to address several concerns: poor material stability under operational conditions, material toxicity due to the presence of the toxic lead and the use of toxic solvents in solution-processing fabrication methods.
The aim of the project “ABSIPERO” is the suppression of toxicity in perovskite-based solar cells, using a non-toxic perovskite composition and processing. To this purpose, alternatives to the toxic lead needs to be explored, being tin and bismuth promising candidates. However, tin-based perovskites are still eco-toxic, being bismuth-based halide perovskites the most adequate alternative for non-toxic perovskite solar cells. In this project vacuum-deposited bismuth based perovskite layers will be fabricated and its characteristics will be studied and evaluated in solar cells. Due to the combination of non-toxic materials and techniques, a step forwards toward non-toxic perovskite-based solar cells will be done.

We are witnessing a massive growth of applications that generate an enormous amount of data.
Social network sites, cloud-based applications, scientific experiments, data warehouses are data- generating phenomena with rapidly growing relevance. The resulting continuous production of structured and unstructured data will have a major impact on today’s society, and has the potential to allow for better decision-making, both for individual organizations, as well as for a general community. It is more important than ever to investigate and to improve the current approaches and solutions for decision-making based on continuous data-flows. Scheduling plays an important role in these processes. A scheduling problem aims to allocate activities to resources and to sequence these activities on the resources in order to achieve a good performance on one or more desirable objectives. Resource allocation, scheduling and sequencing are decision-making processes that are carried out on a permanent basis in most production and service sectors, and not only in data processing, storage or transfer. These processes also play an important role in workforce planning, tactical production planning, operational shop floor, and so on, and because of this practical relevance, scheduling has become one of the most important areas of study within the field of operations research. More specifically, we aim to extend models and techniques for parallel machine scheduling problems. We will also customize the existing models and algorithms to relevant industrial cases that are encountered in real-life applications.

Accidental ingestion of biotoxins present in food constitutes a remarkable public health issue which can be addressed by properly monitoring their levels through the development of analytical methods for their quantification, even at trace levels. Immunochemical methods are suitable for this purpose, indeed many examples have been reported for biotoxins as relevant as ochratoxin A, anatoxin-a or aflatoxins. However, immunoassays for direct detection of those biotoxins in solution based on bioluminescent protein biosensors have not been developed yet. Such assays would have interesting advantages compared to other immunoassays, including the possibility of adapting them to very cheap paper-based devices. In this sense, LUMABS (luminescent antibody sensors) recently introduced by the team of prof. M. Merkx (TU/e) are valuable tools with promising biomedical applications, but also with a high exploitation potential in other fields such as food safety and small molecule detection.
Therefore, this project has two main objectives. The first one is to elicit new LUMABS variants capable of directly detecting small molecules in homogeneous assays, with special focus on the development of ratiometric sensor formats. Once the new sensors have been successfully obtained and their intended purpose has been demonstrated, the second aim will be adapt them for the homogeneous analysis of some of the aforementioned biotoxins, either directly in solution or making use of paper-based devices such as μ-PADs. Different approaches for sensor design will be investigated, whereas sensors will include biotoxin functionalized analogs conjugated by click chemistry with unnatural amino acids and/or covalently linked immunoglobulins by means of the recently reported LASIC technology. Moreover, in collaboration with the team of prof. B. Correia (EPFL), we will try to replace analogs with equivalent peptide motifs to simplify sensor preparation without losing molecular recognition features.

The production of drinking water from sewage can be an important strategy to make a sustainable use of water resources and an important tool to mitigate water scarcity. However, it poses some important challenges, as sewage is a known source of multiple pathogens such as bacteria, viruses, protozoa and helminths that are excreted through feces, urine or skin desquamation.
However, the presence of pathogens in sewage, and particularly of viruses, is a matter of concern as viral removal from sewage is not specifically addressed by conventionally treatments applied in wastewater treatment plants. On countries where wastewater is treated, viruses have been identified as the main cause of water and food-related outbreaks. Traditionally, the study of viruses present in sewage has been limited by the use of specific tests for each concrete viral species that you want to analyze.
If sewage is intended to be used as a water source to produce drinking water, a complete inventory of all possible viral hazards present in sewage that might cause disease must be elaborated. Next Generation Sequencing techniques (NGS) applied on sewage have the potential to detect all human viral pathogens in a single test. Despite its huge potential, this methodology needs some refinement to detect viruses at low concentrations and also to provide information about their infectivity, and hence the hazard that they represent for consumers. These two important limitations need to be addressed if NGS has to be implemented as a routine method to detect viruses in drinkable water.
The present project called METAVIR aims to perfect NGS techniques to overcome their potential limitations in the context of producing drinkable water from sewage. METAVIR will improve the protocols used while allowing the characterization of sewage infectious viruses and providing a complete inventory of viral species. This inventory will be used to study the possible impacts of detected viruses on human health.

Carbon dioxide is the largest anthropogenic contributor to the global warming, which puts three-fourths of the human population at serious risks. Modelling and simulation analysis reveals that with the current rate of carbon dioxide discharge, the observed changes can exponentially deter our environment’s stability to the point of no-return, by the end of this century. Therefore, it is imperative that essential steps must be taken to reduce the anthropogenic discharge of carbon dioxide in the atmosphere.
Many carbon capture techniques have been investigated, which include absorption, adsorption, cryogenics, membrane-based separations and many more. Of which, absorption via chemical solvents have seen extensive commercial application, because of their prevalence and operational ease. The conventional temperature swing absorption (TSA) technology removes carbon dioxide in the absorber at low temperature using a weak base, like alkanolamines. The absorbed gas is collected in another part of the plant in its pure form, using high temperature. This process results in a high energy penalty. Moreover, alkanolamines have low absorption capacity and their corrosive nature and degradative nature results in equipment loss and fugitive emissions of hazardous chemicals to environment.
Contrarily, systems using pressure-swing principle (where pressure is changed across the system to achieve separation of a gas from its stream, e.g. adsorption) are known for their superior energy savings and almost negligible environmental impact. The same concept is applied to the chemical absorption systems.
In this project, a novel pressure swing absorption (PSA) process using a catalytic biological solvent is proposed for eco-friendly and energy efficient separation of carbon dioxide from process streams. The biological solvent is expected to reduce the toxic emissions to the environment by 50% and provide an energy savings of 10%, as compared to current state-of-the-art absorption systems.

The project TOPPING (TOPological PlasmonIcs oN the edGe of 2D materials) aims to study topologically protected collective excitations in one-dimensional (1D) systems formed at the edge of novel two-dimensional (2D) materials, combining theoretical models and first-principles simulations (i.e. parameter-free quantum mechanical calculations).
Research performed over the past decade has uncovered the deep connection between the topology of the crystal Hamiltonian and the existence of symmetry-protected quantum states. One of the most well known examples are 2D topological insulators, whose most striking manifestation is the emergence of 1D metallic edge states that are topologically protected against disorder and impurities. Their physics, however, has been mainly analyzed at the single-particle level, using simple non-interacting models. TOPPING will advance the established understanding by exploring the physics of many-body interactions, which drive the formation of collective excitations called plasmons, at the edge of 2D materials with non-trivial topologies.
Lots of unexplored questions concerning the interplay of topology and collective plasmonic excitations are still open. For instance, what is the effect of the topological protection on edge plasmonic excitations? Are they more robust (in terms of lifetime and scattering rates) with respect to bulk 2D plasmons or ordinary non-topological materials? How do they couple to external electromagnetic fields? I aim to answer these fundamental questions combining my own knowledge of topological materials with the world-leading expertise of DTU and EPFL supervisors on first-principles simulations.
I envisage several possible applications for topological plasmonics, not least in the context of opto-electronics. Here the coupling of electromagnetic fields to collective edge excitations of topological 2D materials could pave the way to new and innovative recipes for transmitting information in a robust, protected way.

A car-free city center is a valuable solution for decreasing traffic congestion and CO2 emission, and improving active mobility and quality of live. For achieving these goals, one of the biggest challenges is the relocation of parking places scattered in the inner district, which cause cruising for available parking as well as car inflows. A possible solution is a “fringe parking” system, which groups parking into a limited number of spots on the border of the district. However, this may decrease the level of accessibility moving the parking away from the final destinations. The key idea of this research project is to use “accelerated moving walkway (AMW)”, a novel transport system, in combination with a fringe parking system to design car-free city center with a high accessibility. The goal of the project is to identify the optimal configuration of a fringe parking system with AMWs. The main methodology is network design optimization and behavioral demand modeling. The research deals with a multi-layer network of vehicular traffic and pedestrians, which interact in terms of both demand and supply and are linked with each other by parking. The research is mainly dedicated to 1) developing a multi-layer network assignment model, 2) optimizing a fringe parking system with AMWs and 3) applying it to a real city case study. The research is expected to contribute to methodology developments in the traffic assignment and network design fields, as well as urban policy developments. The results could support municipalities and tow planners in designing car-free districts for future cities.

Considering that for the next decades autonomous vehicles and human drivers will be sharing the roads, one of the greatest road safety challenges globally will be the proper traffic operation of both driving behaviors. While autonomous vehicles may drive “perfectly”, their behavior does not al-ways resemble that of human drivers. The potential negative implication of this, is that human drivers might be confused by the unintuitive behavior of the automated vehicles, potentially leading to road safety problems.
The aim of the proposed research is to formulate, develop and validate a Turing test tool indicating whether an autonomous vehicle can imitate (the “positive” properties of) human driving behavior. More specifically a tool will be developed (including a mobile and desktop interface) in which users will be asked to distinguish autonomous/human driving behavior through a Turing test methodology consisting of pairs of provided information (e.g. videos or graphs). The methodological framework consists of the following steps:
– Development of a methodology for generating simulation data regarding both human driving cars and scenarios for autonomous driving through a suitable microscopic simulation tool
– Combined investigation of driving performance of autonomous vehicles and human drivers by advanced statistical analysis techniques
– Establishment of a Turing Test Tool to investigate whether an examined autonomous vehicle is behaving in a way that cannot be distinguished from human behavior
Results will have an immediate impact amongst several stakeholders including policy-makers, re-searchers, automotive and insurance industry while the proposed research will be a significant next step of the personal research and professional activities of the researcher.

Mucin glycoproteins provide lubrication and hydration to protect wet epithelia in the human body from tribological stress by reducing wear and friction. Recently, it was shown that purified mucins applied as aqueous lubricants could greatly improve the tribological properties of artificial surfaces. The putative protective potential of mucin-based multi-layer coatings has, however, not been studied yet. A key goal of this project will be to generate such mucin-based multi-layer coatings on different substrates and to study their tribological performance. For this purpose, dopamine, a molecule which serves as an important neurotransmitter in the human body but can also adsorb to a broad range of synthetic materials, will be employed to develop dopamine-mucin multi-layer films using a layer-by-layer approach. Stainless steel, Al2O3 and polydimethylsiloxane will be chosen as representatives for a metal, ceramic and polymeric substrate, respectively. The adsorption properties and multi-layer formation of the mucin-based films on different substrates will be determined using a combination of quartz-crystal-microbalance, and tribological performance both at nanoscale and macroscale will be compared using an atomic force microscopy, a rheometer-based tribometer and a mini-traction machine, repectively. The wear rates and wear mechanisms will be studied using a light profilometer as well as a scanning electron microscope, respectively. By introducing a protective multi-layer coating generated from two biomolecules, this project will be able to provide an innovative technological solution to the area of health and bioengineering by increasing the longevity of synthetic materials which are permanently brought into the human body, e.g. contact lenses, catheters and implants.

Microscopic hair like structures called cilia are nearly ubiquitously present in the human body, and are important for normal functioning of organs. Synchronous motion of motile-cilia induces flow in the surrounding viscous fluid that functions as vital particle/ nutrient transporting medium. During development, for example, cilia present on the human embryonic node generate a flow that determines the left-right asymmetry of our organs. Structural alterations in cilia, mostly caused by genetic mutations, results into a number of severe diseases, called Ciliopathies, such as misplacement of organs in the body, chronic bronchitis, sinusitis, collapsing lungs, and infertility. However, there is still a lack of knowledge about the origin and consequences of ciliopathies, because of the absence of good model systems that allow studying ciliary (dys)function mechanistically. In vivo testing using animal models is problematic since control and measurement of separate parameters are severely hampered. In vitro models are entirely absent or very basic. In this project, I will develop a lab-on-chip model of cilia towards in-vitro analysis against controlled physical parameters. The model first re-quires development of a novel microfabricaton process to produce magnetic artificial cilia mimicking the biological cilium structure as closely as possible. Advanced femtosecond laser machining will be accustomed for incorporating local modifications in the structures, to represent ciliopathies. A new magnetic cilia actuation system, to re-enacting complex natural beating profiles, will be used to artificially reproduce motile cilia motions. As a proof-of-concept, I will create an artificial embryonic node and study the interaction between cilia structures and surrounding fluid at microscopic scales to understand the cilium functioning in their normal and defected forms.

The EU commission established a road-map to decarbonise the transport sector by 2050. A main focus is on urban mobility, which represents 29% of the sector emissions and is over 90% reliant on oil. Biofuels, hydrogen/fuel cells and battery/hybrid vehicles present different characteristics (infrastructure,maturity and driving range). They are foreseen to play a role in the medium and long terms to ensure energy security, address price volatility issues and reduce emissions.
However, compared to fossil fuels, biofuels and hydrogen are not yet commercially competitive because of higher capital costs and possibly low energy efficiency in the production phase. Battery/hybrid vehicle powertrains are heavier and more complex to design. There is a need for increasing the energy efficiency and cost-effectiveness of these alternative fuels and propulsion systems, developing novel processes and optimizing the current ones. This is the aim of this two-folded project.
Firstly, we will investigate how to design and optimize biofuel/hydrogen production routes and hybrid powertrains, developing process models and performing energetic and economic analyses. Secondly, we will analyse the market integration and environmental impact of these fuels and vehicles, considering uncertainties related to the driving behaviour and prices. This project will pave the way for a low-carbon transport sector: innovative and sustainable technologies will be developed in economically-acceptable conditions.

Driven by the need for more sustainable industrial processes, this project sets out to isolate a series of novel aluminium double-bonded complexes and examine their reactivity to the synthesis of value-added products. These are expected to be highly reactive compounds due to the low-valent Lewis-acidic aluminium centre, and as aluminium is the most abundant metal found within the Earth’s crust it is a prime candidate to use in catalytic processes typically dominated by non-sustainable, expensive transition metals. To fully exploit the chemistry available to aluminium a fundamental understanding of its chemical bonding properties is required and will be investigated through the targeted synthesis of heterodiatomic double-bonded complexes of Al = M, where E = p-, d-, or f- block elements.

Single cell analysis methods including microscopy, flow cytometry, and single cell transcriptomic profiling (scRNA-seq) have each contributed to our understanding of cellular identity and are increasingly used in diagnostics. However, each of these technologies currently functions as a stand-alone approach. Here, we reasoned that combining these methods into one device would yield a powerful tool to probe cellular heterogeneity and perform unbiased diagnostics. The aim of this proposal is to engineer a high-throughput, robotic approach integrating the power of microscopy, flow cytometry and molecular processing of single-cells. The core of the Machine-vision Aided Single-cell Processing (MASP) platform will be based on machine-vision systems and augmented with sophisticated imaging capacities. The platform will allow to relate cellular phenotypes to the molecular fingerprint of a cell by leveraging the strengths of high-quality imaging and scRNA-seq. We aim to obtain predictive cellular and pathologic classifiers with the goal of increasing our understanding of cell function and improving diagnostics from liquid biopsies. By combining imaging, machine-learning and microfluidics, MASP will provide the opportunity to develop an interdisciplinary approach for single-cell manipulation, morphological and molecular profiling, opening a new avenue to advance diagnostics and to tackle a plethora of unanswered biological questions, not addressable with available technologies.

The state of the art of filtering and membrane processes evidences the existence of limitations, at theoretical and practical levels, related to their efficient performance. Techniques to analyze the behavior of fluids flowing through membranes and the performances of the membrane itself, are essentially based on microscopic models with experimental justifications, or on ad hoc models derived by merging different theories. Membrane design still proceeds through trials and errors.
The purpose of this project is the design of membranes characterized by the concept of modularity, nowadays one of the fundamental needs in technological adaptation. They must also be more energy and space-saving than classical ones. A multi-scale hierarchical thin membrane is designed as a result of these needs.
In order to satisfy energy saving and downsizing, the new membrane must be thin along the direction of filtration (in contrast to the most used devices which are thick and require more operating energy). To address modularity the membrane owns heterogeneities at different scales, as represented in figure 1, where each scale corresponds to a filtration level. Since Darcy’s law, generally used to simulate flows across thick membranes, is in principle unjustified with the present design, a theoretical and numerical framework to analyze thin membranes, based on homogenization theory, is developed starting from first principles.
Resources-consuming direct numerical simulations of fluid flows and particle transportation across the studied membranes will be calculated, to validate the multi-scale model developed. As final result, an easy-to-use and computationally light reduced order model to analyze the dynamics of fluid solvent and solutes through membranes will be delivered to the scientific community, which will be able, via an inverse formulation, to identify the optimal parameters in membrane processes, in order to give some new insights into the paradigms of industrial production.

Mobility issues in Europe and all over the world are a major socio-economic problem. Even without considering physical constraints, conventional mobility solutions are not capable of meeting EU’s goals in terms of emissions and pollution. The INTERMODE (INtegrating Transportation sErvices thRough Multi-mOdal Demand Estimation) framework aims at developing a new support decision tool to help policymakers designing innovative mobility solutions. The scheme involves two phases: first, collecting deep information for a sample of the population in terms of preferences (mode of transport, departure time, activity agenda) and socio-demographic characteristics (car ownership, household composition). Starting from this sample, aggregate data (such as mobile phone network data) will be used to estimate mobility preferences for the entire population, identifying the potential demand for new services. There are two innovative elements in this project. First, while conventional approaches mostly estimate the mobility demand, we expect to also capture its evolution. Once that the sample data are available, aggregate data are used to evaluate the evolution of the demand without collecting new data at a user level, which is expensive and time-consuming. Second, the INTERMODE scheme will allow measuring the impact of new mobility solutions (such as shared mobility and integrated mobility services) in an efficient way, allowing the modeller to perform cost-benefit analysis before introducing a new service. The ultimate goal of this project is to develop a new framework to represent the mobility demand, by taking into account flexible user behaviour and the need of reducing the overall energy footprint of transport systems

Active turbulence is the chaotic multi-scale flow observed in dense suspensions of active matter. Examples include microbial suspensions, actin and microtubule networks powered by molecular motors, cellular monolayers, synthetic Janus particles, flocks of birds and schools of fish. Unlike classical turbulence, powered by an external input of energy, active turbulence is driven locally by energy injected at the level of its single components. These systems have an intrinsic tendency to self-assemble and self-organize in large spatio-temporal evolving structures. As a result, the computational study of their properties is very demanding, because one needs to address the multiscale nature of the phenomena of relevance.
This project aims at improving our understanding of active turbulence by focusing on the role and effects of confinement provided by solid walls and interfaces. This will reveal how these systems can be used as building-blocks for designing active soft materials, a new class of materials with novel qualitative features and potential. Examples range from a two-fluid emulsion encapsulating active matter used for drug delivery to biomimetic materials, such as a soft tissue made up of highly-packed active droplets capable to resist to intense deformations, or the propulsion of small motors embedded in active fluid suspensions.
We will perform our investigation by developing a highly scalable numerical code that integrates the active nematic equations and represents all the relevant physical scales of the problem by taking full advantage of High Performance Computing architectures. The development of cutting-edge computing tools and the collaboration with experimentalists will enable us to assess the validity of current theoretical models for active turbulence and improve on them.

The proposed Smart Aging research will be an integrated system to recognize ADL and forecast the wellness of an individual. The main objective is to fill the knowledge gap of two contextual observations (i.e. day and time) in the frequent behavior modeling and limitations with predefined activity sets for ambient assisted living (AAL) of service users. Behaviors’ may also change according to other contextual observations including seasonal, weather (or temperature) and social interaction. The metrics in performance evaluation of existing analysis models are completely derived from the machine learning domain, and they do not concern the explicit requirements of activity of daily living (ADLs) analysis such as timestamp, season and frequency of occurrence. Some of the latest activity learning models for behavioral analysis use a predefined set of ADLs and ignore the other activity functions performed by study subjects within the smart home monitoring environment. In the present research, I propose to perform research and redesign of the machine learning model by adding behavioral observations. AAL data sets would be applied to other most appropriate existing techniques Hidden Markov Model and Conditional Random Field to evaluate the performance of redefined wellness behavioral analysis model proposed by me. This objective presents the challenges of sensor data fusion and decision-making. Additionally, the research would be conducted to discover the possible data analytics solution to address the noisy sensor activation data. Two types of data noise would be filtered, anomaly behavioral patterns generated by visitors and anomaly sensor activation (data outlier) that will be generated by faulty sensors as well as the noises generated by the environment.

Driven by the enormous growth of the Internet, Internet of Things, and cloud services, the global data traffic has experienced an explosive growth that has led the optical networks to operate close to their capacity limit. Therefore, to support future capacity demands, a paradigm shift in the design of the next generation optical network is crucial. One viable solution to address this problem of high societal importance is by exploring spatial-division-multiplexing (SDM), by means of multicore and multimode optical fibers. However, exploring SDM is challenging due to a large number of degrees of freedom and complex signal interactions. Thus, aiming at address those challenges, this proposal will explore the latest advances in machine learning to enable SDM-based future optical net-works. Inspired by Google’s DeepMind AlphaGo algorithm, the main objective of this proposal is to perform research into intelligent optical networks that are able to actively learn and provide optimum performance in terms of energy efficiency and information throughput. Specific objectives are the development and experimental verifications of novel machine learning algorithms that are able to predict data traffic evolution and perform self-optimization to satisfy data traffic demands and energy efficiency.

By multi-model coupling, we refer to the surface coupling of a high-fidelity model to a low-fidelity model in a single simulation resulting in a spatial model adaptivity. Using the high-fidelity model for the complete simulation is computationally too expensive, while the low-fidelity model is not accurate enough. Therefore, multi-model coupling is essential for many complex applications, such as climate prediction or nuclear fusion. The fundamental challenge in multi-model coupling is, that the models are in many cases disparate, e.g. they may have different dimensionalities or different numbers of state variables. To tackle the inherent complexity, researchers in academia and industry seek strongly after a usable and scalable coupling software that is able to work with legacy and community codes. The objective of this proposal is to develop generally applicable methods for multi-model coupling and to provide them in a usable and scalable form in the coupling library preCICE.

Epithelial cancers are highly heterogeneous making this cancer sub-type incredible difficult to treat. Of significant concern are pancreatic ductal adenocarcinoma, triple negative and metastatic breast cancers which remain essentially untreatable. Transcription factor Myc is amplified in these epithelial cancers, stimulating tumourigenesis through the promotion of cell proliferation and reprogramming of cancer metabolism. The 14-3-3 adaptor protein has an essential role in homeostasis. Of particular interest is the 14-3-3σ isoform which suppresses tumourigenesis through Myc binding. This protein-protein interaction (PPI) initiates degradation of Myc via polyubiquitination. The 14-3-3σ/Myc PPI is characterised by the binding of a highly conserved phospho-peptide located on the partner protein. During complexing of 14-3-3σ with partner proteins a small ‘drug size’ pocket is formed. Small molecule (SM) binding to this pocket has been shown to improve binding affinity of the complex and has afforded therapeutic outcomes. Research efforts by Ottmann et al have resulted in multiple SM stabilisers in preclinical evaluation. Based upon the aforementioned we hypothesise that SM stabilisers of the 14-3-3σ/Myc will drive Myc degradation, preventing cancer cell reprogramming and that targeting this PPI will improve epithelial cancer patients’ outcomes. Our three objectives are to: A) design and synthesise the first SM stabilizer of the 14-3-3σ/Myc PPI; B) demonstrate 14-3-3σ/Myc stabilizers specifically target Myc and suppress metabolic reprogramming of cancer; and C) assess 14-3-3σ/Myc stabilizers as anticancer agents in human cancers. This research will result in the development of a valuable chemical probe for further oncogenic research and a novel potential therapeutic option for cancer patients.

The project aims to develop a numerical model that encompasses source generation, sound propagation and sound perception, with a focus on musical instrument, propagation in an indoor space and perception of relevant descriptors to musical and room acoustics. The model is to be built with a bottom-up approach, first by coupling a numerical piano model with psychoacoustic models, before the addition of a state-of-the-art room acoustic model based on discontinuous Galerkin (DG) approach. The latter step necessitates the use of high performance computing for the increased degree-of-freedom (from 2D vibration of soundboard to 3D propagation in space) and in increased scale (3m² soundboard to a room of 300m³). A benchmarking exercise with colleagues at DTU, who employs a different room acoustic model, is envisioned, to identify strengths and weaknesses in the DG model for further improvement and refinement.

The objective of this project is to develop novel metamaterial noise barriers to combat environmental noise. The research project aims to improve the low-frequency range (20-200 Hz) performance of conventional barriers by 5 dB, achieving a surface absorption coefficient of 0.8 dB and keeping a base width of maximum 0.2 m. The obtained barrier would be of high interest for application near highways, railways and around festival areas. For the successful accomplishment of the project, metamaterials with the target properties will be developed and integrated into numerical models to optimize the barrier shape and implement it in real outdoor propagation scenarios including meteorological conditions.

Changes in metabolic activity determine the functional status of immune cells such that activated cells show a different metabolism than resting cells. While several methods exist to study cellular metabolism ex vivo, the ‘in situ immunometabolism project’ (iSIMS) aims to assess cellular metabolism in situ by taking snapshots of immune cells residing in their local microenvironment. We will combine electron microscopy and secondary ion mass spectrometry to highly zoom into tissue sections into the energy producing powerhouses termed mitochondria of individual cells, and visualize the metabolic breakdown of cellular fuels such as glucose and amino acids. We will then use immunological methods combined with pharmacological treatments to gain mechanistic insight into the early molecular events of T cell differentiation into effector and memory T cells, which differ in metabolic state and function. Our goal is to understand how metabolism determines immune cell fate and functional status to further delineate how adaptive immunity develops and can be manipulated for therapeutic purposes.

Bacterial toxins are a family of cytotoxic protein secreted by a variety of pathogenic bacterium. These bacterial toxins aid the infection of pathogens through disrupting the normal functions of host cells, and disrupting the neuromuscular transmission and critical biochemical signaling pathways. In some cases, poisoning of bacterial toxins is even lethal factors via serious tissue disruption or neural paralysis. Therefore, it is of great significance to develop vaccines against bacterial toxins, which can active specific humoral immune response to secret specific antibodies and neutralize bacterial toxins.

Although initial attempts in developing vaccines against toxin have verified their great potential against bacterium infection, existing designs still suffer intolerable low efficacy because of unacceptable levels of immunogenicity and safety issues.

In our project, we aim to develop a smart nanogel vaccine against diverse bacterial toxins with enhanced safety and clinical efficacy. We attempt to overcome the outstanding obstacles by developing a responsive, carrier-free delivery system, in which the native bacterial toxins are chemically crosslinked in a reversible way to generate nanogels (NGs) with stimuli-responsive linkers. The NGs designed for lymph node targeted delivery restrain the toxins in inactive and nontoxic status but facilitate the release of intact toxins in response to the internal or external triggers for faithful antigen presentation. We will evaluate the safety and efficacy of this new vaccine in mouse models. The new design will potentially provide a safe, efficient, and versatile vaccine platform against some mostly concerned biotoxin threats, such as Anthrax. The applicant will greatly benefit from working in such a highly interdisciplinary project and obtain unique opportunity to broaden his education and research experience by working in two universities.