Search

Convolutional Neural Networks (CNN) are the state-of-the-art in the field of visual computing. However, a major problem with CNNs is the large number of floating point operations (FLOPs) required to perform convolutions for large inputs. When considering the application of CNNs to video data, convolutional filters become even more complex due to the extra temporal dimension. This leads to problems when respective applications are to be deployed on mobile devices, such as smart phones, tablets, micro-controllers or similar, indicating less computational power. Kim et al. (2016) proposed using a Tucker-decomposition to compress the convolutional kernel of a pre-trained network for images in order to reduce the complexity of the network, i.e. the number of FLOPs. In this paper, we generalize the aforementioned method for application to videos (and other 3D signals) and evaluate the proposed method on a modified version of the THETIS data set, which contains videos of individuals performing tennis shots. We show that the compressed network reaches comparable accuracy, while indicating a memory compression by a factor of 51. However, the actual computational speed-up (factor 1.4) does not meet our theoretically derived expectation (factor 6)., Comment: Presented at the Northern Lights Deep Learning Conference 2022 in Troms{\o}, Norway

Quantum critical fluctuations may prove to play an instrumental role in the formation of unconventional superconductivity. Here, we show that the characteristic scaling of a marginal Fermi liquid is present in inelastic light scattering data of an Fe-based superconductor tuned through a quantum critical point (QCP) by chemical substitution or doping. From the doping dependence of the imaginary time dynamics we are able to distinguish regions dominated by quantum critical behavior from those having classical critical responses. This dichotomy reveals a connection between the marginal Fermi liquid behavior and quantum criticality. In particular, the overlap between regions of high superconducting transition temperatures and quantum critical scaling suggests a contribution from quantum fluctuations to the formation of superconductivity.

Connections between crystal chemistry and critical temperature $T_c$ have been in the focus of superconductivity, one of the most widely studied phenomena in physics, chemistry and materials science alike. In most Fe-based superconductors, materials chemistry and physics conspire so that $T_c$ correlates with the average anion height above the Fe plane, i. e. with the geometry of the FeAs4 or FeCh4 (Ch = Te, Se, or S) tetrahedron. By synthesizing Fe$_{1-y}$Se$_{1-x}$S$_x$ (0$\leq$$x$$\leq$1, $y$$\leq$0.1), we find that in alloyed crystals $T_c$ is not correlated with the anion height like it is for most other Fe superconductors. Instead, changes in $T_c$($x$) and tetragonal-to-orthorombic (nematic) transition $T_s$($x$) upon cooling are correlated with disorder in Fe vibrations in the direction orthogonal to Fe planes, along the crystallographic c-axis. The disorder stems from the random nature of S substitution, causing deformed Fe(Se,S)4 tetrahedra with different Fe-Se and Fe-S bond distances. Our results provide evidence of $T_c$ and $T_s$ suppression by disorder in anion height. The connection to local crystal chemistry may be exploited in computational prediction of new superconducting materials with Fe/S building blocks., Comment: 10 pages, 5 figures

We investigate properties of a spin-1 Heisenberg model with extended and biquadratic interactions, which captures crucial aspects of the low energy physics in FeSe. While we show that the model exhibits a rich phase diagram with four different magnetic ordering tendencies, we identify a parameter regime with strong competition between N\`eel, staggered dimer, and stripe-like magnetic fluctuations, accounting for the physical properties of FeSe. Through the comparison of numerically evaluated spin and Raman response with experiments, we find evidence for enhanced magnetic frustration between N\`eel and co-linear stripe ordering tendencies, which increases with increasing temperature. The explanation of these spectral behaviors with this frustrated spin model supports the idea of local spin interactions in FeSe., Comment: 8 pages, 11 figures

The charge and spin dynamics of the structurally simplest iron-based superconductor, FeSe, may hold the key to understanding the physics of high temperature superconductors in general. Unlike the iron pnictides, FeSe lacks long range magnetic order in spite of a similar structural transition around 90\,K. Here, we report results of Raman scattering experiments as a function of temperature and polarization and simulations based on exact diagonalization of a frustrated spin model. Both experiment and theory find a persistent low energy peak close to 500cm$^{-1}$ in $B_{1g}$ symmetry, which softens slightly around 100\,K, that we assign to spin excitations. By comparing with results from neutron scattering, this study provides evidence for nearly frustrated stripe order in FeSe., Comment: 12 pages, 12 figures

We present results from light scattering experiments on three insulating antiferromagnetic cuprates, YBa$_2$Cu$_3$O$_{6.05}$, Bi$_2$Sr$_2$YCu$_2$O$_{8+\delta}$, and La$_2$CuO$_4$ as a function of polarization and excitation energy {using samples of the latest generation. From the raw data we derive symmetry-resolved spectra.} The spectral shape in $B_{1g}$ symmetry is found to be nearly universal and independent of the excitation energy. The spectra agree quantitatively with predictions by field theory [\onlinecite{Weidinger:2015}] facilitating the precise extraction of the Heisenberg coupling $J$. {In addition, the asymmetric line shape on the high-energy side is found to be related to amplitude fluctuations of the magnetization. In La$_2$CuO$_4$ alone minor contributions from resonance effects may be identified.} The spectra in the other symmetries are not universal. The variations may be traced back to weak resonance effects and extrinsic contributions. For all three compounds we find support for the existence of chiral excitations appearing as a continuum in $A_{2g}$ symmetry having an onset slightly below $3J$. In La$_2$CuO$_4$ an additional isolated excitation appears on top of the $A_{2g}$ continuum., Comment: 8 pages, 7 figures

Resolving the microscopic pairing mechanism and its experimental identification in unconventional superconductors is among the most vexing problems of contemporary condensed matter physics. We show that Raman spectroscopy provides an avenue for this quest by probing the structure of the pairing interaction at play in an unconventional superconductor. As we study the spectra of the prototypical Fe-based superconductor ${\rm Ba_{1-x}K_xFe_2As_2}$ for $0.22\le x \le 0.70$ in all symmetry channels, Raman spectroscopy allows us to distill the leading $s$-wave state. In addition, the spectra collected in the $B_{1g}$ symmetry channel reveal the existence of two collective modes which are indicative of the presence of two competing, yet sub-dominant, pairing tendencies of $d_{x^2-y^2}$ symmetry type. A comprehensive functional Renormalization Group (fRG) and random-phase approximation (RPA) study on this compound confirms the presence of the two sub-leading channels, and consistently matches the experimental doping dependence of the related modes. The synopsis of experimental evidence and theoretical modelling supports a spin-fluctuation mediated superconducting pairing mechanism., Comment: 13 pages, 13 figures

We present a study of charge density wave (CDW) ordering in CeTe$_3$ at room temperature using a scanning tunneling microscope and Raman spectroscopy. Two characteristic CDW ordering wavevectors obtained from the Fourier analysis are assessed to be $|{\bf \textbf{c}}^\ast-{\bf q}|=4.19\,{\rm nm}^{-1}$ and $|{\bf q}|=10.26\,{\rm nm}^{-1}$ where $|{\bf c}^\ast| = 2\pi/c$ is the reciprocal lattice vector. The scanning tunneling spectroscopy measurements, along with inelastic light (Raman) scattering measurements, show a CDW gap $\Delta_{\rm max}$ of approximately 0.37 eV. In addition to the CDW modulation, we observe an organization of the Te sheet atoms in an array of alternating V- and N- groups along the CDW modulation, as predicted in the literature.

An inelastic light (Raman) scattering study of nematicity and critical fluctuations in ${\rm Ba(Fe_{1-x}Co_x)_2As_2}$ ($0\le x \le 0.051$) is presented. It is shown that the response from fluctuations appears only in $B_{1g}$ (${x^2-y^2}$) symmetry. The scattering amplitude increases towards the structural transition at $T_s$ but vanishes only below the magnetic ordering transition at $T_{\rm SDW} < T_s$, suggesting a magnetic origin of the fluctuations. The theoretical analysis explains the selection rules and the temperature dependence of the fluctuation response. These results make magnetism the favorite candidate for driving the series of transitions.%Below $T_{\rm SDW}$ the gap of the magnetically ordered phase opens up.

We show that electronic Raman scattering affords a window into the essential properties of the pairing potential $V_{\mathbf{k},\mathbf{k^{\prime}}}$ of iron-based superconductors. In Ba$_{0.6}$K$_{0.4}$Fe$_2$As$_2$ we observe band dependent energy gaps along with excitonic Bardasis-Schrieffer modes characterizing, respectively, the dominant and subdominant pairing channel. The $d_{x^2-y^2}$ symmetry of all excitons allows us to identify the subdominant channel to originate from the interaction between the electron bands. Consequently, the dominant channel driving superconductivity results from the interaction between the electron and hole bands and has the full lattice symmetry. The results in Rb$_{0.8}$Fe$_{1.6}$Se$_2$ along with earlier ones in Ba(Fe$_{0.939}$Co$_{0.061}$)$_2$As$_2$ highlight the influence of the Fermi surface topology on the pairing interactions., Comment: 5 pages, 4 figures