arXiv subjects

Andreas Fischer

Publications and source records attributed to Andreas Fischer.

At least 73 records · Page 4Linked to original sources

arXiv2018

Offline Signature Verification by Combining Graph Edit Distance and Triplet Networks

Biometric authentication by means of handwritten signatures is a challenging pattern recognition task, which aims to infer a writer model from only a handful of genuine signatures. In order to make it more difficult for a forger to attack the verification system, a promising strategy is to combine different writer models. In this work, we propose to complement a recent structural approach to offline signature verification based on graph edit distance with a statistical approach based on metric learning with deep neural networks. On the MCYT and GPDS benchmark datasets, we demonstrate that combining the structural and statistical models leads to significant improvements in performance, profiting from their complementary properties.

cs.CV↗

arXiv2018

Linear rheology of reversibly cross-linked biopolymer networks

We suggest a simple model for reversible cross-links, binding and unbinding to/from a network of semiflexible polymers. The resulting frequency dependent response of the network to an applied shear is calculated via Brownian dynamics simulations. It is shown to be rather complex with the timescale of the linkers competing with the excitations of the network. If the lifetime of the linkers is the longest timescale, as is indeed the case in most biological networks, then a distinct low frequency peak of the loss modulus develops. The storage modulus shows a corresponding decay from its plateau value, which for irreversible cross-linkers extends all the way to the static limit. This additional relaxation mechanism can be controlled by the relative weight of reversible and irreversible linkers.

cond-mat.soft↗

arXiv2018

Signatures of long-range spin-spin interactions in an (In,Ga)As quantum dot ensemble

We present an investigation of the electron spin dynamics in an ensemble of singly-charged quantum dots subject to an external magnetic field and laser pumping with circularly polarized light. The spectral laser width is tailored such that different groups of quantum dots are coherently pumped. Surprisingly, the dephasing time $T^*$ of the electron spin polarization depends only weakly on the laser spectral width. These findings can be consistently explained by a cluster theory of coupled quantum dots with a long range electronic spin-spin interaction. We present a numerical simulation of the spin dynamics based on the central spin model that includes a quantum mechanical description of the laser pulses as well as a time-independent Heisenberg interaction between each pair of electron spins. We discuss the individual dephasing contributions stemming from the Overhauser field, the distribution of the electron $g$-factors and the electronic spin-spin interaction as well as the spectral width of the laser pulse. This analysis reveals the counterbalancing effect of the total dephasing time when increasing the spectral laser width. On one hand, the deviations of the electron $g$-factors increase. On the other hand, an increasing number of coherently pumped electron spins synchronize due to the spin-spin interaction. We find an excellent agreement between the experimental data and the dephasing time in the simulation using an exponential distribution of Heisenberg couplings with a mean value $\overline{J}\approx 0.26\,\mathrm{μeV}$.

cond-mat.mes-hall↗

arXiv2018

DLL: A Blazing Fast Deep Neural Network Library

Deep Learning Library (DLL) is a new library for machine learning with deep neural networks that focuses on speed. It supports feed-forward neural networks such as fully-connected Artificial Neural Networks (ANNs) and Convolutional Neural Networks (CNNs). It also has very comprehensive support for Restricted Boltzmann Machines (RBMs) and Convolutional RBMs. Our main motivation for this work was to propose and evaluate novel software engineering strategies with potential to accelerate runtime for training and inference. Such strategies are mostly independent of the underlying deep learning algorithms. On three different datasets and for four different neural network models, we compared DLL to five popular deep learning frameworks. Experimentally, it is shown that the proposed framework is systematically and significantly faster on CPU and GPU. In terms of classification performance, similar accuracies as the other frameworks are reported.

cs.LG↗

arXiv2017

Computation on Encrypted Data using Data Flow Authentication

Encrypting data before sending it to the cloud protects it against hackers and malicious insiders, but requires the cloud to compute on encrypted data. Trusted (hardware) modules, e.g., secure enclaves like Intel's SGX, can very efficiently run entire programs in encrypted memory. However, it already has been demonstrated that software vulnerabilities give an attacker ample opportunity to insert arbitrary code into the program. This code can then modify the data flow of the program and leak any secret in the program to an observer in the cloud via SGX side-channels. Since any larger program is rife with software vulnerabilities, it is not a good idea to outsource entire programs to an SGX enclave. A secure alternative with a small trusted code base would be fully homomorphic encryption (FHE) -- the holy grail of encrypted computation. However, due to its high computational complexity it is unlikely to be adopted in the near future. As a result researchers have made several proposals for transforming programs to perform encrypted computations on less powerful encryption schemes. Yet, current approaches fail on programs that make control-flow decisions based on encrypted data. In this paper, we introduce the concept of data flow authentication (DFAuth). DFAuth prevents an adversary from arbitrarily deviating from the data flow of a program. Hence, an attacker cannot perform an attack as outlined before on SGX. This enables that all programs, even those including operations on control-flow decision variables, can be computed on encrypted data. We implemented DFAuth using a novel authenticated homomorphic encryption scheme, a Java bytecode-to-bytecode compiler producing fully executable programs, and SGX enclaves. A transformed neural network that performs machine learning on sensitive medical data can be evaluated on encrypted inputs and encrypted weights in 0.86 seconds.

cs.CR↗

arXiv2017

The Heterogeneous Multiscale Finite Element Method for the Homogenization of Linear Elastic Solids and a Comparison with the FE$^2$ Method

The Heterogeneous Multiscale Finite Element Method (FE-HMM) is a two-scale FEM based on asymptotic homogenization for solving multiscale partial differential equations. It was introduced in [W. E and B. Engquist, \emph{Commun. Math. Sci.}, 1 (2003), 87--132]. The objective of the present work is an FE-HMM formulation for the homogenization of linear elastic solids in a geometrical linear frame, and doing so, for the first time, of a vector-valued field problem. A key ingredient of FE-HMM is that macrostiffness is estimated by stiffness sampling on heterogeneous microdomains in terms a of modified quadrature formula, which implies an equivalence of energy densities of the microscale with the macroscale. Beyond this coincidence with the Hill-Mandel macrohomogeneity condition, which is the cornerstone of the FE$^2$ method, we elaborate a conceptual comparison with the latter method. After developing an algorithmic framework we (i) assess the existing a priori convergence estimates for the micro- and macro-errors in various norms, (ii) verify optimal strategies in uniform micro-macro mesh refinements based on the estimates, (iii) analyze superconvergence properties of FE-HMM, and (iv) compare the numerical results of FE-HMM with those of FE$^2$.

math.NA↗

arXiv2017

Non-equilibrium nuclear spin distribution function in quantum dots subject to periodic pulses

Electron spin dephasing in a singly charged semiconductor quantum dot can partially be suppressed by periodic laser pulsing. We propose a semi-classical approach describing the decoherence of the electron spin polarization governed by the hyperfine interaction with the nuclear spins as well as the probabilistic nature of the photon absorption. We use the steady-state Floquet condition to analytically derive two subclasses of resonance conditions excellently predicting the peak locations in the part of the Overhauser field distribution which is projected in the direction of the external magnetic field. As a consequence of the periodic pulsing, a non-equilibrium distribution develops as a function of time. The numerical simulation of the coupled dynamics reveals the influence of the hyperfine coupling constant distribution onto the evolution of the electron spin polarisation before the next laser pulse. Experimental indications are provided for both subclasses of resonance conditions.

cond-mat.mes-hall↗

arXiv2016

Fowler-Nordheim Plot Analysis: a Progress Report

The commonest method of characterizing a cold field electron emitter is to measure its current-voltage characteristics, and the commonest method of analysing these characteristics is by means of a Fowler-Nordheim (FN) plot. This tutorial/review-type paper outlines a more systematic method of setting out the Fowler-Nordheim-type theory of cold field electron emission, and brings together and summarises the current state of work by the authors on developing the theory and methodology of FN plot analysis. This has turned out to be far more complicated than originally expected. Emphasis is placed in this paper on: (a) the interpretation of FN-plot slopes, which is currently both easier and of more experimental interest than the analysis of FN-plot intercepts; and (b) preliminary explorations into developing methodology for interpreting current-voltage characteristics when there is series resistance in the conduction path from the high-voltage generator to the emitter's emitting regions. This work reinforces our view that FN-plot analysis is best carried out on the raw measured current-voltage data, without pre-conversion into another data format, particularly if series resistance is present in the measuring circuit. Relevant formulae are given for extracting field-enhancement-factor values from such an analysis.

cond-mat.mes-hall↗

arXiv2016

End-to-end simulations of the Visible Tunable Filter for the Daniel K. Inouye Solar Telescope

The Visible Tunable Filter (VTF) is a narrowband tunable filter system for imaging spectroscopy and spectropolarimetry based. The instrument will be one of the first-light instruments of the Daniel K. Inouye Solar Telescope that is currently under construction on Maui (Hawaii). The VTF is being developed by the Kiepenheuer Institut fuer Sonnenphysik in Freiburg as a German contribution to the DKIST. We perform end-to-end simulations of spectropolarimetric observations with the VTF to verify the science requirements of the instrument. The instrument is simulated with two Etalons, and with a single Etalon. The clear aperture of the Etalons is 250 mm, corresponding to a field of view with a diameter of 60 arcsec in the sky (42,000 km on the Sun). To model the large-scale figure errors we employ low-order Zernike polynomials (power and spherical aberration) with amplitudes of 2.5 nm RMS. We use an ideal polarization modulator with equal modulation coefficients of 3-1/2 for the polarization modulation We synthesize Stokes profiles of two iron lines (630.15 nm and 630.25 nm) and for the 854.2 nm line of calcium, for a range of magnetic field values and for several inclination angles. We estimated the photon noise on the basis of the DKIST and VTF transmission values, the atmospheric transmission and the spectral flux from the Sun. For the Fe 630.25 nm line, we obtain a sensitivity of 20 G for the longitudinal component and for 150 G for the transverse component, in agreement with the science requirements for the VTF.

astro-ph.IM↗

arXiv2016

Viscoelasticity of reversibly crosslinked networks of semiflexible polymers

We present a theoretical framework for the linear and nonlinear visco-elastic properties of reversibly crosslinked networks of semiflexible polymers. In contrast to affine models where network strain couples to the polymer end-to-end distance, in our model strain rather serves to locally distort the network structure. This induces bending modes in the polymer filaments, the properties of wich are slaved to the surrounding network structure. Specifically, we investigate the frequency-dependent linear rheology, in particular in combination with crosslink binding/unbinding processes. We also develop schematic extensions to describe the nonlinear response during creep measurements as well as during constant-strainrate ramps.

cond-mat.soft↗

arXiv2015

Resistive Graphene Humidity Sensors with Rapid and Direct Electrical Readout

We demonstrate humidity sensing using a change of electrical resistance of a single- layer chemical vapor deposited (CVD) graphene that is placed on top of a SiO2 layer on a Si wafer. To investigate the selectivity of the sensor towards the most common constituents in air, its signal response was characterized individually for water vapor (H2O), nitrogen (N2), oxygen (O2), and argon (Ar). In order to assess the humidity sensing effect for a range from 1% relative humidity (RH) to 96% RH, devices were characterized both in a vacuum chamber and in a humidity chamber at atmospheric pressure. The measured response and recovery times of the graphene humidity sensors are on the order of several hundred milliseconds. Density functional theory simulations are employed to further investigate the sensitivity of the graphene devices towards water vapor. Results from the interaction between the electrostatic dipole moment of the water and the impurity bands in the SiO2 substrate, which in turn leads to electrostatic doping of the graphene layer. The proposed graphene sensor provides rapid response direct electrical read out and is compatible with back end of the line (BEOL) integration on top of CMOS-based integrated circuits.

cond-mat.mes-hall↗

arXiv2015

Thermal and Vibrational Properties of Thermoelectric ZnSb - Exploring the Origin of Low Thermal Conductivity

The intermetallic compound ZnSb is an interesting thermoelectric material, largely due to its low lattice thermal conductivity. The origin of the low thermal conductivity has so far been speculative. Using multi-temperature single crystal X-ray diffraction (9 - 400 K) and powder X-ray diffraction (300 - 725 K) measurements we characterized the volume expansion and the evolution of structural properties with temperature and identify an increasingly anharmonic behavior of the Zn atoms. From a combination of Raman spectroscopy and first principles calculations of phonons we consolidate the presence of low-energy optic modes with wavenumbers below 60 cm-1. Heat capacity measurements between 2 and 400 K can be well described by a Debye-Einstein model containing one Debye and two Einstein contributions with temperatures ΘD = 195K, ΘE1 = 78 K and ΘE2 = 277 K as well as a significant contribution due to anharmonicity above 150 K. The presence of a multitude of weakly dispersed low-energy optical modes (which couple with the acoustic, heat carrying phonons) combined with anharmonic thermal behavior provides an effective mechanism for low lattice thermal conductivity. The peculiar vibrational properties of ZnSb are attributed to its chemical bonding properties which are characterized by multicenter bonded structural entities. We argue that the proposed mechanism to explain the low lattice thermal conductivity of ZnSb might also control the thermoelectric properties of electron poor semiconductors, such as Zn4Sb3, CdSb, Cd4Sb3, Cd13-xInyZn10, and Zn5Sb4In2-x.

cond-mat.mtrl-sci↗

arXiv2014

Molecular wave-packet dynamics on laser-controlled transition states

Understanding and controlling the electronic as well as ro-vibrational motion and, thus, the entire chemical dynamics in molecules is the ultimate goal of ultrafast laser and imaging science. In photochemistry, laser-induced dissociation has become a valuable tool for modification and control of reaction pathways and kinetics. Here, we present a pump-probe study of the dissociation dynamics of H$_2^+$ using ultrashort extreme-ultraviolet (XUV) and near-infrared (IR) laser pulses. The reaction kinematics can be controlled by varying the pump-probe delay. We demonstrate that the nuclear motion through the transition state can be reduced to isolated pairs of initial vibrational states. The dynamics is well reproduced by intuitive semi-classical trajectories on a time-dependent potential curve. From this most fundamental scenario we gain insight in the underlying mechanisms which can be applied as design principles for molecular quantum control, particularly for ultrafast reactions involving protons.

physics.atom-ph↗

arXiv2014

Order-disorder transition in swirled granular disks

We study the order-disorder transition of horizontally swirled dry and wet granular disks by means of computer simulations. Our systematic investigation of the local order formation as a function of amplitude and period of the external driving force shows that a large cluster of hexagonally ordered particles forms for both dry and wet granular particles at intermediate driving energies. Disordered states are found at small and large driving energies. Wet granular particles reach a higher degree of local hexagonal order, with respect to the dry case. For both cases we report a qualitative phase diagram showing the amount of local order at different state points. Furthermore we find that the transition from hexagonal order to a disordered state is characterised by the appearance of particles with square local order.

cond-mat.soft↗

arXiv2012

Improved approach to Fowler-Nordheim plot analysis

This article introduces an improved approach to Fowler-Nordheim (FN) plot analysis, based on a new type of intercept correction factor. This factor is more cleanly defined than the factor previously used. General enabling theory is given that applies to any type of FN plot of data that can be fitted using a FN-type equation. Practical use is limited to emission situations where slope correction factors can be reliably predicted. By making a series of well-defined assumptions and approximations, it is shown how the general formulas reduce to provide an improved theory of orthodox FN-plot data analysis. This applies to situations where the circuit current is fully controlled by the emitter characteristics, and tunneling can be treated as taking place through a Schottky-Nordheim (SN) barrier. For orthodox emission, good working formulas make numerical evaluation of the slope correction factor and the new intercept correction factor quick and straightforward. A numerical illustration, using simulated emission data, shows how to use this improved approach to derive values for parameters in the full FN-type equation for the SN barrier. Good self-consistency is demonstrated. The general enabling formulas also pave the way for research aimed at developing analogous data-analysis procedures for non-orthodox emission situations.

cond-mat.mes-hall↗

arXiv2012

Illustrating field emission theory by using Lauritsen plots of transmission probability and barrier strength

This technical note relates to the theory of cold field electron emission (CFE). It starts by suggesting that, to emphasize common properties in relation to CFE theory, the term 'Lauritsen plot' could be used to describe all graphical plots made with the reciprocal of barrier field (or the reciprocal of a quantity proportional to barrier field) on the horizontal axis. It then argues that Lauritsen plots related to barrier strength (G) and transmission probability (D) could play a useful role in discussion of CFE theory. Such plots would supplement conventional Fowler-Nordheim (FN) plots. All these plots would be regarded as particular types of Lauritsen plot. The Lauritsen plots of -G and lnD can be used to illustrate how basic aspects of FN tunnelling theory are influenced by the mathematical form of the tunnelling barrier. These, in turn, influence local emission current density and emission current. Illustrative applications used in this note relate to the well-known exact triangular and Schottky-Nordheim barriers, and to the Coulomb barrier (i.e., the electrostatic component of the electron potential energy barrier outside a model spherical emitter). For the Coulomb barrier, a good analytical series approximation has been found for the barrier-form correction factor; this can be used to predict the existence (and to some extent the properties) of related curvature in FN plots.

cond-mat.mes-hall↗

arXiv2012

Influence of barrier form on Fowler-Nordheim plot analysis

Recent research has described an improved method of Fowler-Nordheim plot analysis, based on the definition and evaluation of a slope correction factor and a new form of intercept correction factor. In this improved approach there exists a basic approximation that neglects certain terms in the general theory, and focuses on the influence of the form of the tunneling barrier on the values of basic slope (σB) and intercept (ρB) correction factors. Simple formulae exist that allow these to be evaluated numerically for a barrier of arbitrary well-behaved form. This paper makes an initial exploration of the effects of barrier form on FN plot analysis. For a planar emitter, two models for the correlation-and-exchange (C and E) potential energy (PE) are used. For the Schottky-Nordheim barrier, it is shown that numerical and analytical approaches generate equivalent results. This agreement supports the validity of the numerical methods used. Comparisons with results for the Cutler-Gibbons barrier show that small differences in the assumed C and E PE make little difference to values of σB and ρB. Schottky's planar image PE has then been used, in conjunction with the electrostatic PE variation associated with a spherical emitter model, to explore the influence of apex radius r_a on correction-factor values, for values of r_a greater than 20 nm. Both σB and ρB increase significantly as r_a decreases, especially ρB. At low values of barrier field F, σB depends approximately linearly on 1/F, with a slope that depends on r_a. Suggestions are made for how the exploratory work described in this paper might be extended.

cond-mat.mes-hall↗

arXiv2010

Definable versions of theorems by Kirszbraun and Helly

Kirszbraun's Theorem states that every Lipschitz map $S\to\mathbb R^n$, where $S\subseteq \mathbb R^m$, has an extension to a Lipschitz map $\mathbb R^m \to \mathbb R^n$ with the same Lipschitz constant. Its proof relies on Helly's Theorem: every family of compact subsets of $\mathbb R^n$, having the property that each of its subfamilies consisting of at most $n+1$ sets share a common point, has a non-empty intersection. We prove versions of these theorems valid for definable maps and sets in arbitrary definably complete expansions of ordered fields.

math.LO↗