Project acronym 2D-TOPSENSE
Project Tunable optoelectronic devices by strain engineering of 2D semiconductors
Researcher (PI) Andres CASTELLANOS
Host Institution (HI) AGENCIA ESTATAL CONSEJO SUPERIOR DEINVESTIGACIONES CIENTIFICAS
Call Details Starting Grant (StG), PE7, ERC-2017-STG
Summary The goal of 2D-TOPSENSE is to exploit the remarkable stretchability of two-dimensional semiconductors to fabricate optoelectronic devices where strain is used as an external knob to tune their properties.
While bulk semiconductors tend to break under strains larger than 1.5%, 2D semiconductors (such as MoS2) can withstand deformations of up to 10-20% before rupture. This large breaking strength promises a great potential of 2D semiconductors as ‘straintronic’ materials, whose properties can be adjusted by applying a deformation to their lattice. In fact, recent theoretical works predicted an interesting physical phenomenon: a tensile strain-induced semiconductor-to-metal transition in 2D semiconductors. By tensioning single-layer MoS2 from 0% up to 10%, its electronic band structure is expected to undergo a continuous transition from a wide direct band-gap of 1.8 eV to a metallic behavior. This unprecedented large strain-tunability will undoubtedly have a strong impact in a wide range of optoelectronic applications such as photodetectors whose cut-off wavelength is tuned by varying the applied strain or atomically thin light modulators.
To date, experimental works on strain engineering have been mostly focused on fundamental studies, demonstrating part of the potential of 2D semiconductors in straintronics, but they have failed to exploit strain engineering to add extra functionalities to optoelectronic devices. In 2D-TOPSENSE I will go beyond the state of the art in straintronics by designing and fabricating optoelectronic devices whose properties and performance can be tuned by means of applying strain. 2D-TOPSENSE will focus on photodetectors with a tunable bandwidth and detectivity, light emitting devices whose emission wavelength can be adjusted, light modulators based on 2D semiconductors such as transition metal dichalcogenides or black phosphorus and solar funnels capable of directing the photogenerated charge carriers towards a specific position.
Summary
The goal of 2D-TOPSENSE is to exploit the remarkable stretchability of two-dimensional semiconductors to fabricate optoelectronic devices where strain is used as an external knob to tune their properties.
While bulk semiconductors tend to break under strains larger than 1.5%, 2D semiconductors (such as MoS2) can withstand deformations of up to 10-20% before rupture. This large breaking strength promises a great potential of 2D semiconductors as ‘straintronic’ materials, whose properties can be adjusted by applying a deformation to their lattice. In fact, recent theoretical works predicted an interesting physical phenomenon: a tensile strain-induced semiconductor-to-metal transition in 2D semiconductors. By tensioning single-layer MoS2 from 0% up to 10%, its electronic band structure is expected to undergo a continuous transition from a wide direct band-gap of 1.8 eV to a metallic behavior. This unprecedented large strain-tunability will undoubtedly have a strong impact in a wide range of optoelectronic applications such as photodetectors whose cut-off wavelength is tuned by varying the applied strain or atomically thin light modulators.
To date, experimental works on strain engineering have been mostly focused on fundamental studies, demonstrating part of the potential of 2D semiconductors in straintronics, but they have failed to exploit strain engineering to add extra functionalities to optoelectronic devices. In 2D-TOPSENSE I will go beyond the state of the art in straintronics by designing and fabricating optoelectronic devices whose properties and performance can be tuned by means of applying strain. 2D-TOPSENSE will focus on photodetectors with a tunable bandwidth and detectivity, light emitting devices whose emission wavelength can be adjusted, light modulators based on 2D semiconductors such as transition metal dichalcogenides or black phosphorus and solar funnels capable of directing the photogenerated charge carriers towards a specific position.
Max ERC Funding
1 930 437 €
Duration
Start date: 2018-03-01, End date: 2023-02-28
Project acronym 2DNANOPTICA
Project Nano-optics on flatland: from quantum nanotechnology to nano-bio-photonics
Researcher (PI) Pablo Alonso-González
Host Institution (HI) UNIVERSIDAD DE OVIEDO
Call Details Starting Grant (StG), PE3, ERC-2016-STG
Summary Ubiquitous in nature, light-matter interactions are of fundamental importance in science and all optical technologies. Understanding and controlling them has been a long-pursued objective in modern physics. However, so far, related experiments have relied on traditional optical schemes where, owing to the classical diffraction limit, control of optical fields to length scales below the wavelength of light is prevented. Importantly, this limitation impedes to exploit the extraordinary fundamental and scaling potentials of nanoscience and nanotechnology. A solution to concentrate optical fields into sub-diffracting volumes is the excitation of surface polaritons –coupled excitations of photons and mobile/bound charges in metals/polar materials (plasmons/phonons)-. However, their initial promises have been hindered by either strong optical losses or lack of electrical control in metals, and difficulties to fabricate high optical quality nanostructures in polar materials.
With the advent of two-dimensional (2D) materials and their extraordinary optical properties, during the last 2-3 years the visualization of both low-loss and electrically tunable (active) plasmons in graphene and high optical quality phonons in monolayer and multilayer h-BN nanostructures have been demonstrated in the mid-infrared spectral range, thus introducing a very encouraging arena for scientifically ground-breaking discoveries in nano-optics. Inspired by these extraordinary prospects, this ERC project aims to make use of our knowledge and unique expertise in 2D nanoplasmonics, and the recent advances in nanophononics, to establish a technological platform that, including coherent sources, waveguides, routers, and efficient detectors, permits an unprecedented active control and manipulation (at room temperature) of light and light-matter interactions on the nanoscale, thus laying experimentally the foundations of a 2D nano-optics field.
Summary
Ubiquitous in nature, light-matter interactions are of fundamental importance in science and all optical technologies. Understanding and controlling them has been a long-pursued objective in modern physics. However, so far, related experiments have relied on traditional optical schemes where, owing to the classical diffraction limit, control of optical fields to length scales below the wavelength of light is prevented. Importantly, this limitation impedes to exploit the extraordinary fundamental and scaling potentials of nanoscience and nanotechnology. A solution to concentrate optical fields into sub-diffracting volumes is the excitation of surface polaritons –coupled excitations of photons and mobile/bound charges in metals/polar materials (plasmons/phonons)-. However, their initial promises have been hindered by either strong optical losses or lack of electrical control in metals, and difficulties to fabricate high optical quality nanostructures in polar materials.
With the advent of two-dimensional (2D) materials and their extraordinary optical properties, during the last 2-3 years the visualization of both low-loss and electrically tunable (active) plasmons in graphene and high optical quality phonons in monolayer and multilayer h-BN nanostructures have been demonstrated in the mid-infrared spectral range, thus introducing a very encouraging arena for scientifically ground-breaking discoveries in nano-optics. Inspired by these extraordinary prospects, this ERC project aims to make use of our knowledge and unique expertise in 2D nanoplasmonics, and the recent advances in nanophononics, to establish a technological platform that, including coherent sources, waveguides, routers, and efficient detectors, permits an unprecedented active control and manipulation (at room temperature) of light and light-matter interactions on the nanoscale, thus laying experimentally the foundations of a 2D nano-optics field.
Max ERC Funding
1 459 219 €
Duration
Start date: 2017-01-01, End date: 2021-12-31
Project acronym 2DNanoSpec
Project Nanoscale Vibrational Spectroscopy of Sensitive 2D Molecular Materials
Researcher (PI) Renato ZENOBI
Host Institution (HI) EIDGENOESSISCHE TECHNISCHE HOCHSCHULE ZUERICH
Call Details Advanced Grant (AdG), PE4, ERC-2016-ADG
Summary I propose to investigate the nanometer scale organization of delicate 2-dimensional molecular materials using nanoscale vibrational spectroscopy. 2D structures are of great scientific and technological importance, for example as novel materials (graphene, MoS2, WS2, etc.), and in the form of biological membranes and synthetic 2D-polymers. Powerful methods for their analysis and imaging with molecular selectivity and sufficient spatial resolution, however, are lacking. Tip-enhanced Raman spectroscopy (TERS) allows label-free spectroscopic identification of molecular species, with ≈10 nm spatial resolution, and with single molecule sensitivity for strong Raman scatterers. So far, however, TERS is not being carried out in liquids, which is the natural environment for membranes, and its application to poor Raman scatterers such as components of 2D polymers, lipids, or other membrane compounds (proteins, sugars) is difficult. TERS has the potential to overcome the restrictions of other optical/spectroscopic methods to study 2D materials, namely (i) insufficient spatial resolution of diffraction-limited optical methods; (ii) the need for labelling for all methods relying on fluorescence; and (iii) the inability of some methods to work in liquids. I propose to address a number of scientific questions associated with the spatial organization, and the occurrence of defects in sensitive 2D molecular materials. The success of these studies will also rely critically on technical innovations of TERS that notably address the problem of energy dissipation. This will for the first time allow its application to study of complex, delicate 2D molecular systems without photochemical damage.
Summary
I propose to investigate the nanometer scale organization of delicate 2-dimensional molecular materials using nanoscale vibrational spectroscopy. 2D structures are of great scientific and technological importance, for example as novel materials (graphene, MoS2, WS2, etc.), and in the form of biological membranes and synthetic 2D-polymers. Powerful methods for their analysis and imaging with molecular selectivity and sufficient spatial resolution, however, are lacking. Tip-enhanced Raman spectroscopy (TERS) allows label-free spectroscopic identification of molecular species, with ≈10 nm spatial resolution, and with single molecule sensitivity for strong Raman scatterers. So far, however, TERS is not being carried out in liquids, which is the natural environment for membranes, and its application to poor Raman scatterers such as components of 2D polymers, lipids, or other membrane compounds (proteins, sugars) is difficult. TERS has the potential to overcome the restrictions of other optical/spectroscopic methods to study 2D materials, namely (i) insufficient spatial resolution of diffraction-limited optical methods; (ii) the need for labelling for all methods relying on fluorescence; and (iii) the inability of some methods to work in liquids. I propose to address a number of scientific questions associated with the spatial organization, and the occurrence of defects in sensitive 2D molecular materials. The success of these studies will also rely critically on technical innovations of TERS that notably address the problem of energy dissipation. This will for the first time allow its application to study of complex, delicate 2D molecular systems without photochemical damage.
Max ERC Funding
2 311 696 €
Duration
Start date: 2017-09-01, End date: 2022-08-31
Project acronym 2DTHERMS
Project Design of new thermoelectric devices based on layered and field modulated nanostructures of strongly correlated electron systems
Researcher (PI) Jose Francisco Rivadulla Fernandez
Host Institution (HI) UNIVERSIDAD DE SANTIAGO DE COMPOSTELA
Call Details Starting Grant (StG), PE3, ERC-2010-StG_20091028
Summary Design of new thermoelectric devices based on layered and field modulated nanostructures of strongly correlated electron systems
Summary
Design of new thermoelectric devices based on layered and field modulated nanostructures of strongly correlated electron systems
Max ERC Funding
1 427 190 €
Duration
Start date: 2010-11-01, End date: 2015-10-31
Project acronym 3D-FIREFLUC
Project Taming the particle transport in magnetized plasmas via perturbative fields
Researcher (PI) Eleonora VIEZZER
Host Institution (HI) UNIVERSIDAD DE SEVILLA
Call Details Starting Grant (StG), PE2, ERC-2018-STG
Summary Wave-particle interactions are ubiquitous in nature and play a fundamental role in astrophysical and fusion plasmas. In solar plasmas, magnetohydrodynamic (MHD) fluctuations are thought to be responsible for the heating of the solar corona and the generation of the solar wind. In magnetically confined fusion (MCF) devices, enhanced particle transport induced by MHD fluctuations can deteriorate the plasma confinement, and also endanger the device integrity. MCF devices are an ideal testbed to verify current models and develop mitigation / protection techniques.
The proposed project paves the way for providing active control techniques to tame the MHD induced particle transport in a fusion plasma. A solid understanding of the interaction between energetic particles and MHD instabilities in the presence of electric fields and plasma currents is required to develop such techniques. I will pursue this goal through innovative diagnosis techniques with unprecedented spatio-temporal resolution. Combined with state-of-the-art hybrid MHD codes, a deep insight into the underlying physics mechanism will be gained. The outcome of this research project will have a major impact for next-step MCF devices as I will provide ground-breaking control techniques for mitigating MHD induced particle transport in magnetized plasmas.
The project consists of 3 research lines which follow a bottom-up approach:
(1) Cutting-edge instrumentation, aiming at the new generation of energetic particle and edge current diagnostics.
(2) Unravel the dynamics of energetic particles, electric fields, edge currents and MHD fluctuations.
(3) From lab to space weather: The developed models will revolutionize our understanding of the observed particle acceleration and transport in the solar corona.
Based on this approach, the project represents a gateway between the fusion, astrophysics and space communities opening new avenues for a common basic understanding.
Summary
Wave-particle interactions are ubiquitous in nature and play a fundamental role in astrophysical and fusion plasmas. In solar plasmas, magnetohydrodynamic (MHD) fluctuations are thought to be responsible for the heating of the solar corona and the generation of the solar wind. In magnetically confined fusion (MCF) devices, enhanced particle transport induced by MHD fluctuations can deteriorate the plasma confinement, and also endanger the device integrity. MCF devices are an ideal testbed to verify current models and develop mitigation / protection techniques.
The proposed project paves the way for providing active control techniques to tame the MHD induced particle transport in a fusion plasma. A solid understanding of the interaction between energetic particles and MHD instabilities in the presence of electric fields and plasma currents is required to develop such techniques. I will pursue this goal through innovative diagnosis techniques with unprecedented spatio-temporal resolution. Combined with state-of-the-art hybrid MHD codes, a deep insight into the underlying physics mechanism will be gained. The outcome of this research project will have a major impact for next-step MCF devices as I will provide ground-breaking control techniques for mitigating MHD induced particle transport in magnetized plasmas.
The project consists of 3 research lines which follow a bottom-up approach:
(1) Cutting-edge instrumentation, aiming at the new generation of energetic particle and edge current diagnostics.
(2) Unravel the dynamics of energetic particles, electric fields, edge currents and MHD fluctuations.
(3) From lab to space weather: The developed models will revolutionize our understanding of the observed particle acceleration and transport in the solar corona.
Based on this approach, the project represents a gateway between the fusion, astrophysics and space communities opening new avenues for a common basic understanding.
Max ERC Funding
1 512 250 €
Duration
Start date: 2019-05-01, End date: 2024-04-30
Project acronym 3DNANOMECH
Project Three-dimensional molecular resolution mapping of soft matter-liquid interfaces
Researcher (PI) Ricardo Garcia
Host Institution (HI) AGENCIA ESTATAL CONSEJO SUPERIOR DEINVESTIGACIONES CIENTIFICAS
Call Details Advanced Grant (AdG), PE4, ERC-2013-ADG
Summary Optical, electron and probe microscopes are enabling tools for discoveries and knowledge generation in nanoscale sicence and technology. High resolution –nanoscale or molecular-, noninvasive and label-free imaging of three-dimensional soft matter-liquid interfaces has not been achieved by any microscopy method.
Force microscopy (AFM) is considered the second most relevant advance in materials science since 1960. Despite its impressive range of applications, the technique has some key limitations. Force microscopy has not three dimensional depth. What lies above or in the subsurface is not readily characterized.
3DNanoMech proposes to design, build and operate a high speed force-based method for the three-dimensional characterization soft matter-liquid interfaces (3D AFM). The microscope will combine a detection method based on force perturbations, adaptive algorithms, high speed piezo actuators and quantitative-oriented multifrequency approaches. The development of the microscope cannot be separated from its applications: imaging the error-free DNA repair and to understand the relationship existing between the nanomechanical properties and the malignancy of cancer cells. Those problems encompass the different spatial –molecular-nano-mesoscopic- and time –milli to seconds- scales of the instrument.
In short, 3DNanoMech aims to image, map and measure with picoNewton, millisecond and angstrom resolution soft matter surfaces and interfaces in liquid. The long-term vision of 3DNanoMech is to replace models or computer animations of bimolecular-liquid interfaces by real time, molecular resolution maps of properties and processes.
Summary
Optical, electron and probe microscopes are enabling tools for discoveries and knowledge generation in nanoscale sicence and technology. High resolution –nanoscale or molecular-, noninvasive and label-free imaging of three-dimensional soft matter-liquid interfaces has not been achieved by any microscopy method.
Force microscopy (AFM) is considered the second most relevant advance in materials science since 1960. Despite its impressive range of applications, the technique has some key limitations. Force microscopy has not three dimensional depth. What lies above or in the subsurface is not readily characterized.
3DNanoMech proposes to design, build and operate a high speed force-based method for the three-dimensional characterization soft matter-liquid interfaces (3D AFM). The microscope will combine a detection method based on force perturbations, adaptive algorithms, high speed piezo actuators and quantitative-oriented multifrequency approaches. The development of the microscope cannot be separated from its applications: imaging the error-free DNA repair and to understand the relationship existing between the nanomechanical properties and the malignancy of cancer cells. Those problems encompass the different spatial –molecular-nano-mesoscopic- and time –milli to seconds- scales of the instrument.
In short, 3DNanoMech aims to image, map and measure with picoNewton, millisecond and angstrom resolution soft matter surfaces and interfaces in liquid. The long-term vision of 3DNanoMech is to replace models or computer animations of bimolecular-liquid interfaces by real time, molecular resolution maps of properties and processes.
Max ERC Funding
2 499 928 €
Duration
Start date: 2014-02-01, End date: 2019-01-31
Project acronym 4SUNS
Project 4-Colours/2-Junctions of III-V semiconductors on Si to use in electronics devices and solar cells
Researcher (PI) María Nair LOPEZ MARTINEZ
Host Institution (HI) UNIVERSIDAD AUTONOMA DE MADRID
Call Details Starting Grant (StG), PE7, ERC-2017-STG
Summary It was early predicted by M. Green and coeval colleagues that dividing the solar spectrum into narrow ranges of colours is the most efficient manner to convert solar energy into electrical power. Multijunction solar cells are the current solution to this challenge, which have reached over 30% conversion efficiencies by stacking 3 junctions together. However, the large fabrication costs and time hinders their use in everyday life. It has been shown that highly mismatched alloy (HMA) materials provide a powerful playground to achieve at least 3 different colour absorption regions that enable optimised energy conversion with just one junction. Combining HMA-based junctions with standard Silicon solar cells will rocket solar conversion efficiency at a reduced price. To turn this ambition into marketable devices, several efforts are still needed and few challenges must be overcome.
4SUNS is a revolutionary approach for the development of HMA materials on Silicon technology, which will bring highly efficient multi-colour solar cells costs below current multijunction devices. The project will develop the technology of HMA materials on Silicon via material synthesis opening a new technology for the future. The understanding and optimization of highly mismatched alloy materials-using GaAsNP alloy- will provide building blocks for the fabrication of laboratory-size 4-colours/2-junctions solar cells.
Using a molecular beam epitaxy system, 4SUNS will grow 4-colours/2-junctions structure as well as it will manufacture the final devices. Structural and optoelectronic characterizations will carry out to determine the quality of the materials and the solar cells characteristic to obtain a competitive product. These new solar cells are competitive products to breakthrough on the solar energy sector solar cells and allowing Europe to take leadership on high efficiency solar cells.
Summary
It was early predicted by M. Green and coeval colleagues that dividing the solar spectrum into narrow ranges of colours is the most efficient manner to convert solar energy into electrical power. Multijunction solar cells are the current solution to this challenge, which have reached over 30% conversion efficiencies by stacking 3 junctions together. However, the large fabrication costs and time hinders their use in everyday life. It has been shown that highly mismatched alloy (HMA) materials provide a powerful playground to achieve at least 3 different colour absorption regions that enable optimised energy conversion with just one junction. Combining HMA-based junctions with standard Silicon solar cells will rocket solar conversion efficiency at a reduced price. To turn this ambition into marketable devices, several efforts are still needed and few challenges must be overcome.
4SUNS is a revolutionary approach for the development of HMA materials on Silicon technology, which will bring highly efficient multi-colour solar cells costs below current multijunction devices. The project will develop the technology of HMA materials on Silicon via material synthesis opening a new technology for the future. The understanding and optimization of highly mismatched alloy materials-using GaAsNP alloy- will provide building blocks for the fabrication of laboratory-size 4-colours/2-junctions solar cells.
Using a molecular beam epitaxy system, 4SUNS will grow 4-colours/2-junctions structure as well as it will manufacture the final devices. Structural and optoelectronic characterizations will carry out to determine the quality of the materials and the solar cells characteristic to obtain a competitive product. These new solar cells are competitive products to breakthrough on the solar energy sector solar cells and allowing Europe to take leadership on high efficiency solar cells.
Max ERC Funding
1 499 719 €
Duration
Start date: 2018-02-01, End date: 2023-01-31
Project acronym AGRISCENTS
Project Scents and sensibility in agriculture: exploiting specificity in herbivore- and pathogen-induced plant volatiles for real-time crop monitoring
Researcher (PI) Theodoor Turlings
Host Institution (HI) UNIVERSITE DE NEUCHATEL
Call Details Advanced Grant (AdG), LS9, ERC-2017-ADG
Summary Plants typically release large quantities of volatiles in response to attack by herbivores or pathogens. I may claim to have contributed to various breakthroughs in this research field, including the discovery that the volatile blends induced by different attackers are astonishingly specific, resulting in characteristic, readily distinguishable odour blends. Using maize as our model plant, I wish to take several leaps forward in our understanding of this signal specificity and use this knowledge to develop sensors for the real-time detection of crop pests and diseases. For this, three interconnected work-packages will aim to:
• Develop chemical analytical techniques and statistical models to decipher the odorous vocabulary of plants, and to create a complete inventory of “odour-prints” for a wide range of herbivore-plant and pathogen-plant combinations, including simultaneous infestations.
• Develop and optimize nano-mechanical sensors for the detection of specific plant volatile mixtures. For this, we will initially adapt a prototype sensor that has been successfully developed for the detection of cancer-related volatiles in human breath.
• Genetically manipulate maize plants to release a unique blend of root-produced volatiles upon herbivory. For this, we will engineer gene cassettes that combine recently identified P450 (CYP) genes from poplar with inducible, root-specific promoters from maize. This will result in maize plants that, in response to pest attack, release easy-to-detect aldoximes and nitriles from their roots.
In short, by investigating and manipulating the specificity of inducible odour blends we will generate the necessary knowhow to develop a novel odour-detection device. The envisioned sensor technology will permit real-time monitoring of the pests and enable farmers to apply crop protection treatments at the right time and in the right place.
Summary
Plants typically release large quantities of volatiles in response to attack by herbivores or pathogens. I may claim to have contributed to various breakthroughs in this research field, including the discovery that the volatile blends induced by different attackers are astonishingly specific, resulting in characteristic, readily distinguishable odour blends. Using maize as our model plant, I wish to take several leaps forward in our understanding of this signal specificity and use this knowledge to develop sensors for the real-time detection of crop pests and diseases. For this, three interconnected work-packages will aim to:
• Develop chemical analytical techniques and statistical models to decipher the odorous vocabulary of plants, and to create a complete inventory of “odour-prints” for a wide range of herbivore-plant and pathogen-plant combinations, including simultaneous infestations.
• Develop and optimize nano-mechanical sensors for the detection of specific plant volatile mixtures. For this, we will initially adapt a prototype sensor that has been successfully developed for the detection of cancer-related volatiles in human breath.
• Genetically manipulate maize plants to release a unique blend of root-produced volatiles upon herbivory. For this, we will engineer gene cassettes that combine recently identified P450 (CYP) genes from poplar with inducible, root-specific promoters from maize. This will result in maize plants that, in response to pest attack, release easy-to-detect aldoximes and nitriles from their roots.
In short, by investigating and manipulating the specificity of inducible odour blends we will generate the necessary knowhow to develop a novel odour-detection device. The envisioned sensor technology will permit real-time monitoring of the pests and enable farmers to apply crop protection treatments at the right time and in the right place.
Max ERC Funding
2 498 086 €
Duration
Start date: 2018-09-01, End date: 2023-08-31
Project acronym ALGILE
Project Foundations of Algebraic and Dynamic Data Management Systems
Researcher (PI) Christoph Koch
Host Institution (HI) ECOLE POLYTECHNIQUE FEDERALE DE LAUSANNE
Call Details Starting Grant (StG), PE6, ERC-2011-StG_20101014
Summary "Contemporary database query languages are ultimately founded on logic and feature an additive operation – usually a form of (multi)set union or disjunction – that is asymmetric in that additions or updates do not always have an inverse. This asymmetry puts a greater part of the machinery of abstract algebra for equation solving outside the reach of databases. However, such equation solving would be a key functionality that problems such as query equivalence testing and data integration could be reduced to: In the current scenario of the presence of an asymmetric additive operation they are undecidable. Moreover, query languages with a symmetric additive operation (i.e., which has an inverse and is thus based on ring theory) would open up databases for a large range of new scientific and mathematical applications.
The goal of the proposed project is to reinvent database management systems with a foundation in abstract algebra and specifically in ring theory. The presence of an additive inverse allows to cleanly define differences between queries. This gives rise to a database analog of differential calculus that leads to radically new incremental and adaptive query evaluation algorithms that substantially outperform the state of the art techniques. These algorithms enable a new class of systems which I call Dynamic Data Management Systems. Such systems can maintain continuously fresh query views at extremely high update rates and have important applications in interactive Large-scale Data Analysis. There is a natural connection between differences and updates, motivating the group theoretic study of updates that will lead to better ways of creating out-of-core data processing algorithms for new storage devices. Basing queries on ring theory leads to a new class of systems, Algebraic Data Management Systems, which herald a convergence of database systems and computer algebra systems."
Summary
"Contemporary database query languages are ultimately founded on logic and feature an additive operation – usually a form of (multi)set union or disjunction – that is asymmetric in that additions or updates do not always have an inverse. This asymmetry puts a greater part of the machinery of abstract algebra for equation solving outside the reach of databases. However, such equation solving would be a key functionality that problems such as query equivalence testing and data integration could be reduced to: In the current scenario of the presence of an asymmetric additive operation they are undecidable. Moreover, query languages with a symmetric additive operation (i.e., which has an inverse and is thus based on ring theory) would open up databases for a large range of new scientific and mathematical applications.
The goal of the proposed project is to reinvent database management systems with a foundation in abstract algebra and specifically in ring theory. The presence of an additive inverse allows to cleanly define differences between queries. This gives rise to a database analog of differential calculus that leads to radically new incremental and adaptive query evaluation algorithms that substantially outperform the state of the art techniques. These algorithms enable a new class of systems which I call Dynamic Data Management Systems. Such systems can maintain continuously fresh query views at extremely high update rates and have important applications in interactive Large-scale Data Analysis. There is a natural connection between differences and updates, motivating the group theoretic study of updates that will lead to better ways of creating out-of-core data processing algorithms for new storage devices. Basing queries on ring theory leads to a new class of systems, Algebraic Data Management Systems, which herald a convergence of database systems and computer algebra systems."
Max ERC Funding
1 480 548 €
Duration
Start date: 2012-01-01, End date: 2016-12-31
Project acronym AlgoRNN
Project Recurrent Neural Networks and Related Machines That Learn Algorithms
Researcher (PI) Juergen Schmidhuber
Host Institution (HI) UNIVERSITA DELLA SVIZZERA ITALIANA
Call Details Advanced Grant (AdG), PE6, ERC-2016-ADG
Summary Recurrent neural networks (RNNs) are general parallel-sequential computers. Some learn their programs or weights. Our supervised Long Short-Term Memory (LSTM) RNNs were the first to win pattern recognition contests, and recently enabled best known results in speech and handwriting recognition, machine translation, etc. They are now available to billions of users through the world's most valuable public companies including Google and Apple. Nevertheless, in lots of real-world tasks RNNs do not yet live up to their full potential. Although universal in theory, in practice they fail to learn important types of algorithms. This ERC project will go far beyond today's best RNNs through novel RNN-like systems that address some of the biggest open RNN problems and hottest RNN research topics: (1) How can RNNs learn to control (through internal spotlights of attention) separate large short-memory structures such as sub-networks with fast weights, to improve performance on many natural short-term memory-intensive tasks which are currently hard to learn by RNNs, such as answering detailed questions on recently observed videos? (2) How can such RNN-like systems metalearn entire learning algorithms that outperform the original learning algorithms? (3) How to achieve efficient transfer learning from one RNN-learned set of problem-solving programs to new RNN programs solving new tasks? In other words, how can one RNN-like system actively learn to exploit algorithmic information contained in the programs running on another? We will test our systems existing benchmarks, and create new, more challenging multi-task benchmarks. This will be supported by a rather cheap, GPU-based mini-brain for implementing large RNNs.
Summary
Recurrent neural networks (RNNs) are general parallel-sequential computers. Some learn their programs or weights. Our supervised Long Short-Term Memory (LSTM) RNNs were the first to win pattern recognition contests, and recently enabled best known results in speech and handwriting recognition, machine translation, etc. They are now available to billions of users through the world's most valuable public companies including Google and Apple. Nevertheless, in lots of real-world tasks RNNs do not yet live up to their full potential. Although universal in theory, in practice they fail to learn important types of algorithms. This ERC project will go far beyond today's best RNNs through novel RNN-like systems that address some of the biggest open RNN problems and hottest RNN research topics: (1) How can RNNs learn to control (through internal spotlights of attention) separate large short-memory structures such as sub-networks with fast weights, to improve performance on many natural short-term memory-intensive tasks which are currently hard to learn by RNNs, such as answering detailed questions on recently observed videos? (2) How can such RNN-like systems metalearn entire learning algorithms that outperform the original learning algorithms? (3) How to achieve efficient transfer learning from one RNN-learned set of problem-solving programs to new RNN programs solving new tasks? In other words, how can one RNN-like system actively learn to exploit algorithmic information contained in the programs running on another? We will test our systems existing benchmarks, and create new, more challenging multi-task benchmarks. This will be supported by a rather cheap, GPU-based mini-brain for implementing large RNNs.
Max ERC Funding
2 500 000 €
Duration
Start date: 2017-10-01, End date: 2022-09-30
