Catalogue Search | MBRL
Are you sure you want to remove the book from the shelf?
{{itemTitle}}
967
result(s) for
"Einheit"
Sort by:
Solving the quantum many-body problem with artificial neural networks
The challenge posed by the many-body problem in quantum physics originates from the difficulty of describing the nontrivial correlations encoded in the exponential complexity of the many-body wave function. Here we demonstrate that systematic machine learning of the wave function can reduce this complexity to a tractable computational form for some notable cases of physical interest. We introduce a variational representation of quantum states based on artificial neural networks with a variable number of hidden neurons. A reinforcement-learning scheme we demonstrate is capable of both finding the ground state and describing the unitary time evolution of complex interacting quantum systems. Our approach achieves high accuracy in describing prototypical interacting spins models in one and two dimensions.
Journal Article
Unidirectional rotary motion in a metal–organic framework
Overcrowded alkene-based light-driven molecular motors are able to perform large-amplitude repetitive unidirectional rotations. Their behaviour is well understood in solution. However, Brownian motion precludes the precise positioning at the nanoscale needed to harness cooperative action. Here, we demonstrate molecular motors organized in crystalline metal–organic frameworks (MOFs). The motor unit becomes a part of the organic linker (or strut), and its spatial arrangement is elucidated through powder and single-crystal X-ray analyses and polarized optical and Raman microscopies. We confirm that the light-driven unidirectional rotation of the motor units is retained in the MOF framework and that the motors can operate in the solid state with similar rotary speed (rate of thermal helix inversion) to that in solution. These ‘moto-MOFs’ could in the future be used to control dynamic function in crystalline materials.Overcrowded alkene-based motors are made part of the crystalline structure of a metal–organic framework. It is shown that they retain the same rotatory speed and unidirectionality as in solution.
Journal Article
The knowns and unknowns of neural adaptations to resistance training
The initial increases in force production with resistance training are thought to be primarily underpinned by neural adaptations. This notion is firmly supported by evidence displaying motor unit adaptations following resistance training; however, the precise locus of neural adaptation remains elusive. The purpose of this review is to clarify and critically discuss the literature concerning the site(s) of putative neural adaptations to short-term resistance training. The proliferation of studies employing non-invasive stimulation techniques to investigate evoked responses have yielded variable results, but generally support the notion that resistance training alters intracortical inhibition. Nevertheless, methodological inconsistencies and the limitations of techniques, e.g. limited relation to behavioural outcomes and the inability to measure volitional muscle activity, preclude firm conclusions. Much of the literature has focused on the corticospinal tract; however, preliminary research in non-human primates suggests reticulospinal tract is a potential substrate for neural adaptations to resistance training, though human data is lacking due to methodological constraints. Recent advances in technology have provided substantial evidence of adaptations within a large motor unit population following resistance training. However, their activity represents the transformation of afferent and efferent inputs, making it challenging to establish the source of adaptation. Whilst much has been learned about the nature of neural adaptations to resistance training, the puzzle remains to be solved. Additional analyses of motoneuron firing during different training regimes or coupling with other methodologies (e.g., electroencephalography) may facilitate the estimation of the site(s) of neural adaptations to resistance training in the future.
Journal Article
Shape-encoded dynamic assembly of mobile micromachines
Field-directed and self-propelled colloidal assembly have been used to build micromachines capable of performing complex motions and functions. However, integrating heterogeneous components into micromachines with specified structure, dynamics and function is still challenging. Here, we describe the dynamic self-assembly of mobile micromachines with desired configurations through pre-programmed physical interactions between structural and motor units. The assembly is driven by dielectrophoretic interactions, encoded in the three-dimensional shape of the individual parts. Micromachines assembled from magnetic and self-propelled motor parts exhibit reconfigurable locomotion modes and additional rotational degrees of freedom that are not available to conventional monolithic microrobots. The versatility of this site-selective assembly strategy is demonstrated on different reconfigurable, hierarchical and three-dimensional micromachine assemblies. Our results demonstrate how shape-encoded assembly pathways enable programmable, reconfigurable mobile micromachines. We anticipate that the presented design principle will advance and inspire the development of more sophisticated, modular micromachines and their integration into multiscale hierarchical systems.
Journal Article
Model-Based Reinforcement Learning Variable Impedance Control for Human-Robot Collaboration
Industry 4.0
is taking
human-robot collaboration
at the center of the production environment. Collaborative robots enhance productivity and flexibility while reducing human’s fatigue and the risk of injuries, exploiting advanced control methodologies. However, there is a lack of real-time model-based controllers accounting for the complex human-robot interaction dynamics. With this aim, this paper proposes a
Model-Based Reinforcement Learning
(MBRL) variable impedance controller to assist human operators in collaborative tasks. More in details, an ensemble of Artificial Neural Networks (ANNs) is used to learn a human-robot interaction dynamic model, capturing uncertainties. Such a learned model is kept updated during collaborative tasks execution. In addition, the learned model is used by a
Model Predictive Controller
(MPC) with
Cross-Entropy Method
(CEM). The aim of the MPC+CEM is to online optimize the stiffness and damping
impedance control
parameters minimizing the human effort (i.e, minimizing the human-robot interaction forces). The proposed approach has been validated through an experimental procedure. A lifting task has been considered as the reference validation application (weight of the manipulated part: 10 kg unknown to the robot controller). A KUKA LBR iiwa 14 R820 has been used as a test platform. Qualitative performance (i.e, questionnaire on perceived collaboration) have been evaluated. Achieved results have been compared with previous developed offline model-free optimized controllers and with the robot manual guidance controller. The proposed MBRL variable impedance controller shows improved human-robot collaboration. The proposed controller is capable to actively assist the human in the target task, compensating for the unknown part weight. The human-robot interaction dynamic model has been trained with a few initial experiments (30 initial experiments). In addition, the possibility to keep the learning of the human-robot interaction dynamics active allows accounting for the adaptation of human motor system.
Journal Article
Second messenger–mediated tactile response by a bacterial rotary motor
When bacteria encounter surfaces, they respond with surface colonization and virulence induction. The mechanisms of bacterial mechanosensation and downstream signaling remain poorly understood. Here, we describe a tactile sensing cascade in Caulobacter crescentus in which the flagellar motor acts as sensor. Surface-induced motor interference stimulated the production of the second messenger cyclic diguanylate by the motor-associated diguanylate cyclase DgcB. This led to the allosteric activation of the glycosyltransferase HfsJ to promote rapid synthesis of a polysaccharide adhesin and surface anchoring. Although the membrane-embedded motor unit was essential for surface sensing, mutants that lack external flagellar structures were hypersensitive to mechanical stimuli. Thus, the bacterial flagellar motor acts as a tetherless sensor reminiscent of mechanosensitive channels.
Journal Article
THE ECONOMICS OF DENSITY: EVIDENCE FROM THE BERLIN WALL
This paper develops a quantitative model of internal city structure that features agglomeration and dispersion forces and an arbitrary number of heterogeneous city blocks. The model remains tractable and amenable to empirical analysis because of stochastic shocks to commuting decisions, which yield a gravity equation for commuting flows. To structurally estimate agglomeration and dispersion forces, we use data on thousands of city blocks in Berlin for 1936, 1986, and 2006 and exogenous variation from the city's division and reunification. We estimate substantial and highly localized production and residential externalities. We show that the model with the estimated agglomeration parameters can account both qualitatively and quantitatively for the observed changes in city structure. We show how our quantitative framework can be used to undertake counterfactuals for changes in the organization of economic activity within cities in response, for example, to changes in the transport network.
Journal Article
Contextual Wavefunction collapse: an integrated theory of quantum measurement
This paper is an in depth implementation of the proposal (Ellis 2012 Ann. Phys. NY 327 1890-932) that the quantum measurement issue can be resolved by carefully looking at top-down contextual effects within realistic measurement contexts. The specific setup of the measurement apparatus determines the possible events that can take place. The interaction of local heat baths with a quantum system plays a key role in the process. In contrast to the usual attempts to explain quantum measurement by decoherence, we argue that the heat bath follows unitary time evolution only over limited length and time scales (Drossel 2017 His. Phil. Sci. B 58 12-21) and thus leads to localization and stochastic dynamics of quantum particles that interact with it. We show furthermore that a theory that describes all the steps from the initial arrival of the quantum particle to the final pointer deflection must use elements from classical physics. This proposal also provides a contextual answer to the puzzle of the origin of the arrow of time when quantum measurements take place: it derives from the cosmological direction of time. Overall, our proposal is for contextual wavefunction collapse.
Journal Article
Classification with a disordered dopant-atom network in silicon
Classification is an important task at which both biological and artificial neural networks excel
1
,
2
. In machine learning, nonlinear projection into a high-dimensional feature space can make data linearly separable
3
,
4
, simplifying the classification of complex features. Such nonlinear projections are computationally expensive in conventional computers. A promising approach is to exploit physical materials systems that perform this nonlinear projection intrinsically, because of their high computational density
5
, inherent parallelism and energy efficiency
6
,
7
. However, existing approaches either rely on the systems’ time dynamics, which requires sequential data processing and therefore hinders parallel computation
5
,
6
,
8
, or employ large materials systems that are difficult to scale up
7
. Here we use a parallel, nanoscale approach inspired by filters in the brain
1
and artificial neural networks
2
to perform nonlinear classification and feature extraction. We exploit the nonlinearity of hopping conduction
9
,
10
–
11
through an electrically tunable network of boron dopant atoms in silicon, reconfiguring the network through artificial evolution to realize different computational functions. We first solve the canonical two-input binary classification problem, realizing all Boolean logic gates
12
up to room temperature, demonstrating nonlinear classification with the nanomaterial system. We then evolve our dopant network to realize feature filters
2
that can perform four-input binary classification on the Modified National Institute of Standards and Technology handwritten digit database. Implementation of our material-based filters substantially improves the classification accuracy over that of a linear classifier directly applied to the original data
13
. Our results establish a paradigm of silicon-based electronics for small-footprint and energy-efficient computation
14
.
The nonlinearity of hopping conduction in a disordered network of boron dopant atoms in silicon is used to perform nonlinear classification and feature extraction.
Journal Article