Search

Despite nearly a decade of literature on style transfer, there is no undisputed definition of artistic style. State-of-the-art models produce impressive results but are difficult to interpret since, without a coherent definition of style, the problem of style transfer is inherently ill-posed. Early work framed style-transfer as an optimization problem but treated style as a measure only of texture. This led to artifacts in the outputs of early models where content features from the style image sometimes bled into the output image. Conversely, more recent work with diffusion models offers compelling empirical results but provides little theoretical grounding. To address these issues, we propose an alternative definition of artistic style. We suggest that style should be thought of as a set of global symmetries that dictate the arrangement of local textures. We validate this perspective empirically by learning the symmetries of a large dataset of paintings and showing that symmetries are predictive of the artistic movement to which each painting belongs. Finally, we show that by considering both local and global features, using both Lie generators and traditional measures of texture, we can quantitatively capture the stylistic similarity between artists better than with either set of features alone. This approach not only aligns well with art historians' consensus but also offers a robust framework for distinguishing nuanced stylistic differences, allowing for a more interpretable, theoretically grounded approach to style transfer.

There is now substantial evidence for traveling waves and other structured spatiotemporal recurrent neural dynamics in cortical structures; but these observations have typically been difficult to reconcile with notions of topographically organized selectivity and feedforward receptive fields. We introduce a new 'spacetime' perspective on neural computation in which structured selectivity and dynamics are not contradictory but instead are complimentary. We show that spatiotemporal dynamics may be a mechanism by which natural neural systems encode approximate visual, temporal, and abstract symmetries of the world as conserved quantities, thereby enabling improved generalization and long-term working memory.

A prominent theory of affective response to music revolves around the concepts of surprisal and expectation. In prior work, this idea has been operationalized in the form of probabilistic models of music which allow for precise computation of song (or note-by-note) probabilities, conditioned on a 'training set' of prior musical or cultural experiences. To date, however, these models have been limited to compute exact probabilities through hand-crafted features or restricted to linear models which are likely not sufficient to represent the complex conditional distributions present in music. In this work, we propose to use modern deep probabilistic generative models in the form of a Diffusion Model to compute an approximate likelihood of a musical input sequence. Unlike prior work, such a generative model parameterized by deep neural networks is able to learn complex non-linear features directly from a training set itself. In doing so, we expect to find that such models are able to more accurately represent the 'surprisal' of music for human listeners. From the literature, it is known that there is an inverted U-shaped relationship between surprisal and the amount human subjects 'like' a given song. In this work we show that pre-trained diffusion models indeed yield musical surprisal values which exhibit a negative quadratic relationship with measured subject 'liking' ratings, and that the quality of this relationship is competitive with state of the art methods such as IDyOM. We therefore present this model a preliminary step in developing modern deep generative models of music expectation and subjective likability.

A prominent goal of representation learning research is to achieve representations which are factorized in a useful manner with respect to the ground truth factors of variation. The fields of disentangled and equivariant representation learning have approached this ideal from a range of complimentary perspectives; however, to date, most approaches have proven to either be ill-specified or insufficiently flexible to effectively separate all realistic factors of interest in a learned latent space. In this work, we propose an alternative viewpoint on such structured representation learning which we call Flow Factorized Representation Learning, and demonstrate it to learn both more efficient and more usefully structured representations than existing frameworks. Specifically, we introduce a generative model which specifies a distinct set of latent probability paths that define different input transformations. Each latent flow is generated by the gradient field of a learned potential following dynamic optimal transport. Our novel setup brings new understandings to both \textit{disentanglement} and \textit{equivariance}. We show that our model achieves higher likelihoods on standard representation learning benchmarks while simultaneously being closer to approximately equivariant models. Furthermore, we demonstrate that the transformations learned by our model are flexibly composable and can also extrapolate to new data, implying a degree of robustness and generalizability approaching the ultimate goal of usefully factorized representation learning., Comment: NeurIPS23

Traveling waves of neural activity have been observed throughout the brain at a diversity of regions and scales; however, their precise computational role is still debated. One physically inspired hypothesis suggests that the cortical sheet may act like a wave-propagating system capable of invertibly storing a short-term memory of sequential stimuli through induced waves traveling across the cortical surface, and indeed many experimental results from neuroscience correlate wave activity with memory tasks. To date, however, the computational implications of this idea have remained hypothetical due to the lack of a simple recurrent neural network architecture capable of exhibiting such waves. In this work, we introduce a model to fill this gap, which we denote the Wave-RNN (wRNN), and demonstrate how such an architecture indeed efficiently encodes the recent past through a suite of synthetic memory tasks where wRNNs learn faster and reach significantly lower error than wave-free counterparts. We further explore the implications of this memory storage system on more complex sequence modeling tasks such as sequential image classification and find that wave-based models not only again outperform comparable wave-free RNNs while using significantly fewer parameters, but additionally perform comparably to more complex gated architectures such as LSTMs and GRUs.

Multiview Self-Supervised Learning (MSSL) is based on learning invariances with respect to a set of input transformations. However, invariance partially or totally removes transformation-related information from the representations, which might harm performance for specific downstream tasks that require such information. We propose 2D strUctured and EquivarianT representations (coined DUET), which are 2d representations organized in a matrix structure, and equivariant with respect to transformations acting on the input data. DUET representations maintain information about an input transformation, while remaining semantically expressive. Compared to SimCLR (Chen et al., 2020) (unstructured and invariant) and ESSL (Dangovski et al., 2022) (unstructured and equivariant), the structured and equivariant nature of DUET representations enables controlled generation with lower reconstruction error, while controllability is not possible with SimCLR or ESSL. DUET also achieves higher accuracy for several discriminative tasks, and improves transfer learning., Comment: Accepted at ICML 2023

Despite the significant recent progress in deep generative models, the underlying structure of their latent spaces is still poorly understood, thereby making the task of performing semantically meaningful latent traversals an open research challenge. Most prior work has aimed to solve this challenge by modeling latent structures linearly, and finding corresponding linear directions which result in `disentangled' generations. In this work, we instead propose to model latent structures with a learned dynamic potential landscape, thereby performing latent traversals as the flow of samples down the landscape's gradient. Inspired by physics, optimal transport, and neuroscience, these potential landscapes are learned as physically realistic partial differential equations, thereby allowing them to flexibly vary over both space and time. To achieve disentanglement, multiple potentials are learned simultaneously, and are constrained by a classifier to be distinct and semantically self-consistent. Experimentally, we demonstrate that our method achieves both more qualitatively and quantitatively disentangled trajectories than state-of-the-art baselines. Further, we demonstrate that our method can be integrated as a regularization term during training, thereby acting as an inductive bias towards the learning of structured representations, ultimately improving model likelihood on similarly structured data., Comment: ICML 2023

In this work, we observe that many existing self-supervised learning algorithms can be both unified and generalized when seen through the lens of equivariant representations. Specifically, we introduce a general framework we call Homomorphic Self-Supervised Learning, and theoretically show how it may subsume the use of input-augmentations provided an augmentation-homomorphic feature extractor. We validate this theory experimentally for simple augmentations, demonstrate how the framework fails when representational structure is removed, and further empirically explore how the parameters of this framework relate to those of traditional augmentation-based self-supervised learning. We conclude with a discussion of the potential benefits afforded by this new perspective on self-supervised learning.

Materials realizing the XY model in two dimensions (2D) are sparse. Here we use neutron triple-axis spectroscopy to investigate the critical static and dynamical magnetic fluctuations in the square-lattice antiferromagnets Ca$_{2}$RuO$_{4}$ and Ca$_{3}$Ru$_2$O$_{7}$. We probe the temperature-dependence of the antiferromagnetic (AFM) Bragg-intensity, the $Q$-width, the amplitude, and the energy-width of the magnetic diffuse scattering in vicinity to the N\'eel temperature $T_N$ to determine the critical behavior of the magnetic order parameter $M$, correlation length $\xi$, susceptibility $\chi$, and the characteristic energy $\Gamma$ with the corresponding critical exponents $\beta$, $\nu$, $\gamma$, and $z$, respectively. We find that the critical behaviors of the single-layer compound Ca$_{2}$RuO$_{4}$ follow universal scaling laws that are compatible with predictions of the 2D-XY model. The bilayer compound Ca$_{3}$Ru$_2$O$_{7}$ is only partly consistent with the 2D-XY theory and best described by the three-dimensional Ising (3D-I) model, which is likely a consequence of the intra-bilayer exchange interactions in combination with an orthorhombic single-ion anisotropy. Hence, our results suggest that layered ruthenates are promising solid-state platforms for research on the 2D-XY model and the effects of 3D interactions and additional spin-space anisotropies on the magnetic fluctuations., Comment: 19 pages, 17 figures

Category-selectivity in the brain describes the observation that certain spatially localized areas of the cerebral cortex tend to respond robustly and selectively to stimuli from specific limited categories. One of the most well known examples of category-selectivity is the Fusiform Face Area (FFA), an area of the inferior temporal cortex in primates which responds preferentially to images of faces when compared with objects or other generic stimuli. In this work, we leverage the newly introduced Topographic Variational Autoencoder to model the emergence of such localized category-selectivity in an unsupervised manner. Experimentally, we demonstrate our model yields spatially dense neural clusters selective to faces, bodies, and places through visualized maps of Cohen's d metric. We compare our model with related supervised approaches, namely the Topographic Deep Artificial Neural Network (TDANN) of Lee et al., and discuss both theoretical and empirical similarities. Finally, we show preliminary results suggesting that our model yields a nested spatial hierarchy of increasingly abstract categories, analogous to observations from the human ventral temporal cortex.

In this work we seek to bridge the concepts of topographic organization and equivariance in neural networks. To accomplish this, we introduce the Topographic VAE: a novel method for efficiently training deep generative models with topographically organized latent variables. We show that such a model indeed learns to organize its activations according to salient characteristics such as digit class, width, and style on MNIST. Furthermore, through topographic organization over time (i.e. temporal coherence), we demonstrate how predefined latent space transformation operators can be encouraged for observed transformed input sequences -- a primitive form of unsupervised learned equivariance. We demonstrate that this model successfully learns sets of approximately equivariant features (i.e. "capsules") directly from sequences and achieves higher likelihood on correspondingly transforming test sequences. Equivariance is verified quantitatively by measuring the approximate commutativity of the inference network and the sequence transformations. Finally, we demonstrate approximate equivariance to complex transformations, expanding upon the capabilities of existing group equivariant neural networks.

Magnetic state of exchanged biased CoO(20nm)/Co($d_F$) bilayer ($d_F$=5-20nm) was studied by means of polarized neutron reflectometry. By spacing of CoO/Co bilayer and Al$_2$O$_3$ substrate with Nb(20nm) layer we created waveguide structure which allowed us to significantly enhance intensity of spin-flip (SF) scattering in the position of optical resonances. For the trained sample with thinnest Co(5nm) we detected strong SF scattering at the resonance position (up to 30\% of incoming intensity) speaking about high non-collinearity of the system. As $d_F$ increases, the intensity of SF scattering linearly decreases. At the same time we observed asymmetry of up-down and down-up scattering channels at the resonance positions. We attribute this asymmetry to the Zeeman splitting of neutrons energies with different initial polarization taking place in high external field. Analysis, however, shows that the applied in the PNR experiment external field is not enough to quantitatively explain the observed asymmetry for the samples with $d_F > $ 5nm and we have to postulate presence of additional magnetic field produced by sample. We attribute this additional field to the stray field produced by chiral Bloch domain walls. The chirality of the domain walls can be explained by Dzyaloshinskii-Moriya interaction arising at the CoO/Co interface. Our results can be useful for designing of spintronic devices using exchange bias effect.

High levels of missing data and strong class imbalance are ubiquitous challenges that are often presented simultaneously in real-world time series data. Existing methods approach these problems separately, frequently making significant assumptions about the underlying data generation process in order to lessen the impact of missing information. In this work, we instead demonstrate how a general self-supervised training method, namely Autoregressive Predictive Coding (APC), can be leveraged to overcome both missing data and class imbalance simultaneously without strong assumptions. Specifically, on a synthetic dataset, we show that standard baselines are substantially improved upon through the use of APC, yielding the greatest gains in the combined setting of high missingness and severe class imbalance. We further apply APC on two real-world medical time-series datasets, and show that APC improves the classification performance in all settings, ultimately achieving state-of-the-art AUPRC results on the Physionet benchmark., Comment: Accepted to the NeurIPS 2021 Workshop on Self-Supervised Learning: Theory and Practice

Spin-flop transition (SFT) consists in a jump-like reversal of antiferromagnetic magnetic moments into a non-collinear state when the magnetic field increases above the critical value. Potentially the SFT can be utilized in many applications of a rapidly developing antiferromagnetic spintronics. However, the difficulty of using them in conventional antiferromagnets lies in (a) too large switching magnetic fields (b) the need for presence of a magnetic anisotropy, and (c) requirement to apply magnetic field along the correspondent anisotropy axis. In this work we propose to use artificial ferrimagnets in which the spin-flop transition occurs without anisotropy and the transition field can be lowered by adjusting exchange coupling in the structure. This is proved by experiment on artificial Fe-Gd ferrimagnets where usage of Pd spacers allowed us to suppress the transition field by two orders of magnitude., Comment: article, 5 pages, 3 figures

Efficient gradient computation of the Jacobian determinant term is a core problem in many machine learning settings, and especially so in the normalizing flow framework. Most proposed flow models therefore either restrict to a function class with easy evaluation of the Jacobian determinant, or an efficient estimator thereof. However, these restrictions limit the performance of such density models, frequently requiring significant depth to reach desired performance levels. In this work, we propose Self Normalizing Flows, a flexible framework for training normalizing flows by replacing expensive terms in the gradient by learned approximate inverses at each layer. This reduces the computational complexity of each layer's exact update from $\mathcal{O}(D^3)$ to $\mathcal{O}(D^2)$, allowing for the training of flow architectures which were otherwise computationally infeasible, while also providing efficient sampling. We show experimentally that such models are remarkably stable and optimize to similar data likelihood values as their exact gradient counterparts, while training more quickly and surpassing the performance of functionally constrained counterparts.

We have investigated the structural, magnetic and superconducting properties of [Nb(1.5nm)/Fe(x)]$_{10}$ superlattices deposited on a thick Nb(50nm) layer. Our investigation showed that the Nb(50nm) layer grows epitaxially at 800$^\circ$C on Al$_2$O$_3$(1$\bar{1}$02) substrate. Samples grown at this condition posses a high residual resistivity ratio of 15-20. By using neutron reflectometry we show that Fe/Nb superlattices with $x<$ 4 nm form a depth-modulated FeNb alloy with concentration of iron varying between 60% and 90%. This alloy has properties of a weak ferromagnet. Proximity of this weak ferromagnetic layer to a thick superconductor leads to an intermediate phase that is characterized by co-existing superconducting and normal-state domains. By increasing the thickness of the Fe layer to $x$ = 4 nm the intermediate phase disappears. We attribute the intermediate state to proximity induced non-homogeneous superconductivity in the periodic Fe/Nb structure.

A detailed study of the near-surface structure and composition of Nb, the material of choice for Superconducting Radio Frequency accelerator (SRF) cavities, is of great importance in order to understand the effects of different treatments applied during cavity production. By means of surface-sensitive techniques such as grazing incidence diffuse X-ray scattering, X-ray reflectivity and X-ray photoelectron spectroscopy, single-crystalline Nb(100) samples were investigated in and ex-situ during annealing in UHV as well as in nitrogen atmospheres with temperatures and pressures similar to the ones employed in real Nb cavity treatments. Annealing of Nb specimens up to 800{\deg}C in vacuum promotes partial reduction of the natural surface oxides (Nb2O5, NbO2, NbO) into NbO. Upon cooling to 120{\deg}C, no evidence of nitrogen-rich layers was detected after nitrogen exposure times of up to 48 hours. Oxygen enrichment below the Nb/oxide interface and posterior diffusion of oxygen species towards the Nb matrix, along with a partial reduction of the natural surface oxides was observed upon a stepwise annealing up to 250{\deg}C. Nitrogen introduction to the system at 250{\deg}C neither promotes N diffusion into the Nb matrix nor the formation of new surface layers. Upon further heating to 500{\deg}C in a nitrogen atmosphere, the growth of a new subsurface Nb$_x$N$_y$ layer was detected. These results shed light on the composition of the near-surface region of Nb after low-temperature nitrogen treatments, which are reported to lead to a performance enhancement of SRF cavities., Comment: 15 pages, 8 figures, 1 table, to be submitted to PRB

We have used spin-polarized neutron reflectometry to investigate the magnetization profile of superlattices composed of ferromagnetic Gd and superconducting Nb layers. We have observed a partial suppression of ferromagnetic (F) order of Gd layers in [Gd($d_F$)/Nb(25nm)]$_{12}$ superlattices below the superconducting (S) transition of the Nb layers. The amplitude of the suppression decreases with increasing $d_F$. By analyzing the neutron spin asymmetry we conclude that the observed effect has an electromagnetic origin - the proximity-coupled S layers screen out the external magnetic field and thus suppress the F response of the Gd layers inside the structure. Our investigation demonstrates the considerable influence of electromagnetic effects on the magnetic properties of S/F systems.

We describe a new sensitive method for the investigation of weakly magnetic films placed inside a tri-layer planar waveguide. Polarized neutrons tunnel into the waveguide through the surface, channel along the layers and are emitted from the end face as a narrow and slightly divergent microbeam. Polarization analysis permits to detect very small magnetization in the order of a few 10 Gauss. The magnetic film containing the rare-earth element Tb was investigated using both fixed wavelength and time-of-flight polarized neutron reflectometers. The experimental results are presented and discussed., Comment: Submitted to NIM A

Neutron planar waveguides are focusing devices generating a narrow neutron beam of submicron width. Such a neutron microbeam can be used for the investigation of local microstructures with high spatial resolution. The essential parameter of the microbeam is its angular width. The main contribution to the microbeam angular divergence is Fraunhofer diffraction on a narrow slit. We review and discuss various ways to characterize the angular divergence of the neutron microbeam using time-of-flight and fixed wavelength reflectometers., Comment: 17 pages, 16 figures. Submitted to NIM A

We present a new study of magnetic structures with controllable effective exchange energy for Josephson switches and memory. As a basis for a weak link we propose to use a periodic structure comprised of ferromagnetic (F) layers spaced by thin superconductors (s). Our calculations based on Usadel equations show that switching from parallel (P) to antiparallel (AP) alignment of neighboring F layers can lead to a significant enhancement of the critical current through the junction. To control magnetic alignment we propose to use periodic system where unit cell is a pseudo spin-valve $F_1$/s/$F_2$/s with $F_1$ and $F_2$ two magnetic layers having different coercive fields. In order to check feasibility of controllable switching between AP and P states through the \emph{whole} periodic structure we prepared a superlattice [Co(1.5nm)/Nb(8nm)/Co(2.5nm)/Nb(8nm)]$_6$ between two superconducting layers of Nb(25nm). Neutron scattering showed that parallel and antiparallel alignment can be organized by using of magnetic fields of only several tens of Oersted.

We investigate neutron propagation in a middle layer of a planar waveguide which is a tri-layer thin film. A narrow divergent microbeam emitted from the end face of the film is registered. The neutron channeling length is experimentally measured as a function of the guiding channel width. Experimental results are compared with calculations., Comment: 11 pages, 12 figures, submitted to JETP

Associative memory using fast weights is a short-term memory mechanism that substantially improves the memory capacity and time scale of recurrent neural networks (RNNs). As recent studies introduced fast weights only to regular RNNs, it is unknown whether fast weight memory is beneficial to gated RNNs. In this work, we report a significant synergy between long short-term memory (LSTM) networks and fast weight associative memories. We show that this combination, in learning associative retrieval tasks, results in much faster training and lower test error, a performance boost most prominent at high memory task difficulties.

Studying the prototypical ferromagnetic superconductor UGe$_2$ we demonstrate the potential of the Modulated IntEnsity by Zero Effort (MIEZE) technique---a novel neutron spectroscopy method with ultra-high energy resolution of at least 1~$\mu$eV---for the study of quantum matter. We reveal purely longitudinal spin fluctuations in UGe$_2$ with a dual nature arising from $5f$ electrons that are hybridized with the conduction electrons. Local spin fluctuations are perfectly described by the Ising universality class in three dimensions, whereas itinerant spin fluctuations occur over length scales comparable to the superconducting coherence length, showing that MIEZE is able to spectroscopically disentangle the complex low-energy behavior characteristic of quantum materials., Comment: 7 pages, 9 figures, supplemental materials

We report on a study of the structural, magnetic and superconducting properties of Nb(25nm)/Gd($d_f$)/Nb(25nm) hybrid structures of a superconductor/ ferromagnet (S/F) type. The structural characterization of the samples, including careful determination of the layer thickness, was performed using neutron and X-ray scattering with the aid of depth sensitive mass-spectrometry. The magnetization of the samples was determined by SQUID magnetometry and polarized neutron reflectometry and the presence of magnetic ordering for all samples down to the thinnest Gd(0.8nm) layer was shown. The analysis of the neutron spin asymmetry allowed us to prove the absence of magnetically dead layers in junctions with Gd interlayer thickness larger than one monolayer. The measured dependence of the superconducting transition temperature $T_c(d_f)$ has a damped oscillatory behavior with well defined positions of the minimum at $d_f$=3nm and the following maximum at $d_f$=4nm; the behavior, which is in qualitative agreement with the prior work (J.S. Jiang et al, PRB 54, 6119). The analysis of the $T_c(d_f)$ dependence based on Usadel equations showed that the observed minimum at $d_f$=3nm can be described by the so called "$0$" to "$\pi$" phase transition of highly transparent S/F interfaces with the superconducting correlation length $\xi_f \approx 4$nm in Gd. This penetration length is several times higher than for strong ferromagnets like Fe, Co or Ni, simplifying thus preparation of S/F structures with $d_f \sim \xi_f$ which are of topical interest in superconducting spintronics.

In this work we analyse the role of a thin Cr spacer between Fe and Gd layers on structure and magnetic properties of a [Fe(35A)/Cr(tCr)/Gd(50A)/Cr(tCr)]x12 superlattice. Samples without the Cr spacer (tCr=0) and with a thin tCr=4A are investigated using X-ray diffraction, polarized neutron and resonance X-ray magnetic reflectometry, SQUID magnetometery, magneto-optical Kerr effect and ferromagnetic resonance techniques. Magnetic properties are studied experimentally in a wide temperature range 4-300K and analysed theoretically using numerical simulation on the basis of the mean-field model. We show that a reasonable agreement with the experimental data can be obtained considering temperature dependence of the effective field parameter in gadolinium layers. The analysis of the experimental data shows that besides a strong reduction of the antiferromagnetic coupling between Fe and Gd, the introduction of Cr spacers into Fe/Gd superlattice leads to modification of both structural and magnetic characteristics of the ferromagnetic layers.

Magnetic spinels (with chemical formula $AX_{2}$O$_{4}$, with $X$ a 3$d$ transition metal ion) that also have an orbital degeneracy are Jahn-Teller active and hence possess a coupling between spin and lattice degrees of freedom. At high temperatures, MgV$_{2}$O$_{4}$ is a cubic spinel based on V$^{3+}$ ions with a spin $S$=1 and a triply degenerate orbital ground state. A structural transition occurs at T$_{OO}$=63 K to an orbitally ordered phase with a tetragonal unit cell followed by an antiferromagnetic transition of T$_{N}$=42 K on cooling. We apply neutron spectroscopy in single crystals of MgV$_{2}$O$_{4}$ to show an anomaly for intermediate wavevectors at T$_{OO}$ associated with the acoustic phonon sensitive to the shear elastic modulus $\left(C_{11}-C_{12}\right)/2$. On warming, the shear mode softens for momentum transfers near close to half the Brillouin zone boundary, but recovers near the zone centre. High resolution spin-echo measurements further illustrate a temporal broadening with increased temperature over this intermediate range of wavevectors, indicative of a reduction in phonon lifetime. A subtle shift in phonon frequencies over the same range of momentum transfers is observed with magnetic fields. We discuss this acoustic anomaly in context of coupling to orbital and charge fluctuations., Comment: 10 pages, 8 figures

We report the synthesis of Ca$_2$RuO$_4$ thin films on NdCaAlO$_{4}$ (110), LaAlO$_3$ (100) and LaSrAlO$_4$ (001) substrates and show that epitaxial strain induces a transition from the Mott-insulating phase of bulk Ca$_2$RuO$_4$ into a metallic phase. Magnetometry and spin-polarized neutron reflectometry reveal a low-temperature, small-moment ferromagnetic state in metallic Ca$_2$RuO$_4$ films.

We report grazing incidence small angle neutron scattering (GISANS) and complementary off-specular neutron reflectometry (OSR) of the magnetic order in a single-crystalline epitaxial MnSi film on Si(111) in the thick film limit. Providing a means of direct reciprocal space mapping, GISANS and OSR reveal a magnetic modulation perpendicular to the films under magnetic fields parallel and perpendicular to the film, where additional polarized neutron reflectometry (PNR) and magnetization measurements are in excellent agreement with the literature. Regardless of field orientation, our data does not suggest the presence of more complex spin textures, notably the formation of skyrmions. This observation establishes a distinct difference with bulk samples of MnSi of similar thickness under perpendicular field, in which a skyrmion lattice dominates the phase diagram. Extended x-ray absorption fine structure measurements suggest that small shifts of the Si positions within the unstrained unit cell control the magnetic state, representing the main difference between the films and thin bulk samples.

Electron-phonon interaction in copper oxide superconductors is still enigmatic. Strong coupling for certain optic phonons is now well established experimentally, but theoretical understanding is challenging. Scattering of electrons near the Fermi surface by the longitudinal acoustic (LA) phonons is expected from basic theory because these phonons modulate electron density. We used inelastic neutron scattering on a large single crystal sample of optimally-doped Bi$_2$Sr$_2$CaCu$_2$O$_{8+\delta}$ to show that low-energy LA phonons could couple to electronic density fluctuations only at small phonon wavevectors, which naturally limits any interaction to forward scattering. Such scattering should not be pairbreaking in the case of the d-wave gap. We also found that previously the reported low energy phonon spectral weight half-way to the zone boundary is consistent with conventional lattice dynamics and does not reflect an incipient charge density wave., Comment: Updated Oct. 2019 to include new measurements and additional discussion. 7 pages, 4 figures, Supplementary included: 6 pages, 5 figures

Pi-conjugated segments - chromophores - constitute the electronically active units of polymer materials used in organic electronics. To elucidate the effect of bending of these linear moieties on elementary electronic properties such as luminescence colour and radiative rate we introduce a series of molecular polygons. The pi-system in these molecules becomes so distorted in bichromophores (digons) that these absorb and emit light of arbitrary polarisation: any part of the chain absorbs and emits radiation with equal probability. Bending leads to a cancellation of transition dipole moment (TDM), increasing excited-state lifetime. Simultaneously, fluorescence shifts to the red as radiative transitions require mixing of the excited state with vibrational modes. However, strain can become so large that excited-state localisation on shorter units of the chain occurs, compensating TDM cancellation. Since these effects counteract, underlying correlations between shape and photophysics can only be resolved in single molecules.

We report results of a magnetic characterization of [Cu$_{30}$Ni$_{70}$(6nm)]$_{20}$ (x=1-7nm) superlattices using Polarized Neutron Reflectometry (PNR) and SQUID magnetometry. The study has shown that the magnetic moment of the structures growths almost linearly from H = 0 to H$_{sat}$ = 1.3kOe which is an indirect evidence of antiferromagnetic (AF) coupling of the magnetic moments in neighbouring layers. PNR, however, did not detect any in-plane AF coupling. Taking into account the out-of-plane easy axis of the Cu$_{30}$Ni$_{70}$ layers, this may mean that only the out-of-plane component of the magnetic moments are AF coupled.

We report a neutron spin-echo study of the critical dynamics in the $S=5/2$ antiferromagnets MnF$_2$ and Rb$_2$MnF$_4$ with three-dimensional (3D) and two-dimensional (2D) spin systems, respectively, in zero external field. Both compounds are Heisenberg antiferromagnets with a small uniaxial anisotropy resulting from dipolar spin-spin interactions, which leads to a crossover in the critical dynamics close to the N\'eel temperature, $T_N$. By taking advantage of the $\mu\text{eV}$ energy resolution of the spin-echo spectrometer, we have determined the dynamical critical exponents $z$ for both longitudinal and transverse fluctuations. In MnF$_2$, both the characteristic temperature for crossover from 3D Heisenberg to 3D Ising behavior and the exponents $z$ in both regimes are consistent with predictions from the dynamical scaling theory. The amplitude ratio of longitudinal and transverse fluctuations also agrees with predictions. In Rb$_2$MnF$_4$, the critical dynamics crosses over from the expected 2D Heisenberg behavior for $T\gg T_N$ to a scaling regime with exponent $z = 1.387(4)$, which has not been predicted by theory and may indicate the influence of long-range dipolar interactions.

In this work we present a neutron Larmor diffraction study of the rare earth pyrochlores $R_2$Ti$_2$O$_7$, with $R$ = Tb, Dy, Ho. We measured the temperature dependence of the lattice parameter with precision $10^{-5}$, between 0.5 K and 300 K in each of the three compounds. The lattice parameter of the spin ices Dy$_2$Ti$_2$O$_7$ and Ho$_2$Ti$_2$O$_7$ enters into the derivation of the charge of the emergent magnetic monopole excitations suggested to exist in these materials. We found that throughout the range of applicability of the theory of emergent monopoles in the spin ices there will be no renormalization of the monopole charge due to lattice contraction. In Tb$_2$Ti$_2$O$_7$ strong magnetoelastic interactions have been reported. We found no sign of the previously reported negative thermal expansion, but did observe anomalies in the thermal expansion that can be correlated with previously observed interactions between phonon and crystal field excitations. Other features in the thermal expansion of all three compounds can be related to previously observed anomalies the elastic constants, and explained by the phonon band structure of the rare earth titanates. The temperature dependence of the lattice strain in all three compounds can be correlated with the thermal population of excited crystal field levels., Comment: 9 pages, 6 figures

An increase of the magnetic moment in superconductor/ferromagnet (S/F) bilayers V(40nm)/F [F$=$Fe(1,3nm), Co(3nm), Ni(3nm)] was observed using SQUID magnetometry upon cooling below the superconducting transition temperature Tc in magnetic fields of 10 Oe to 50 Oe applied parallel to the sample surface. A similar increase, often called the paramagnetic Meissner effect (PME), was observed before in various superconductors and superconductor/ferromagnet systems. To explain the PME effect in the presented S/F bilayers a model based on a row of vortices located at the S/F interface is proposed. According to the model the magnetic moment induced below Tc consists of the paramagnetic contribution of the vortex cores and the diamagnetic contribution of the vortex-free region of the S layer. Since the thickness of the S layer is found to be 3-4 times less than the magnetic field penetration depth, this latter diamagnetic contribution is negligible. The model correctly accounts for the sign, the approximate magnitude and the field dependence of the paramagnetic and the Meissner contributions of the induced magnetic moment upon passing the superconducting transition of a ferromagnet/superconductor bilayer.

In this article we investigate the time dependence of the gap mode of copper nitrate at various temperatures. We combine state-of-the-art theoretical calculations with high precision neutron resonance spin-echo measurements to understand the anomalous decoherence effects found previously in this material. It is shown that the time domain offers a complementary view on this phenomenon, which allows us to directly compare experimental data and theoretical predictions without the need of further intensive data analysis, such as (de)convolution., Comment: 6 pages, 5 figures

With x-ray absorption spectroscopy and polarized neutron reflectometry we studied how the magnetic proximity effect at the interface between the cuprate high-$T_C$ superconductor $\mathrm{YBa_2Cu_3O_7}$ (YBCO) and the ferromagnet $\mathrm{La_{2/3}Ca_{1/3}MnO_3}$ (LCMO) is related to the electronic and magnetic properties of the LCMO layers. In particular, we explored how the magnitude of the ferromagnetic Cu moment on the YBCO side depends on the strength of the antiferromagnetic (AF) exchange coupling with the Mn moment on the LCMO side. We found that the Cu moment remains sizeable if the AF coupling with the Mn moments is strongly reduced or even entirely suppressed. The ferromagnetic order of the Cu moments thus seems to be intrinsic to the interfacial CuO$_2$ planes and related to a weakly ferromagnetic intra-planar exchange interaction. The latter is discussed in terms of the partial occupation of the Cu $3d_{3z^2-r^2}$ orbitals, which occurs in the context of the so-called orbital reconstruction of the interfacial Cu ions.

We have used high-resolution neutron Larmor diffraction and capacitative dilatometry to investigate spontaneous and forced magnetostriction in undoped, antiferromagnetic YBa$_2$Cu$_3$O$_{6.0}$, the parent compound of a prominent family of high-temperature superconductors. Upon cooling below the N\'eel temperature, $T_N = 420$~K, Larmor diffraction reveals the formation of magneto-structural domains of characteristic size $\sim 240$~nm. In the antiferromagnetic state, dilatometry reveals a minute ($4 \times 10^{-6}$) orthorhombic distortion of the crystal lattice in external magnetic fields. We attribute these observations to exchange striction and spin-orbit coupling induced magnetostriction, respectively, and show that they have an important influence on the thermal and charge transport properties of undoped and lightly doped cuprates., Comment: to appear in PRL

We present and apply a new method to measure directly weak magnetization in thin films. The polarization of a neutron beam channeling through a thin film structure is measured after exiting the structure edge as a microbeam. We have applied the method to a tri-layer thin film structure acting as a planar waveguide for polarized neutrons. The middle guiding layer is a rare earth based ferrimagnetic material TbCo5 with a low magnetization of about 20 mT. We demonstrate that the channeling method is more sensitive than the specular neutron reflection method.

Peculiarities of the formation of a neutron enhanced standing wave in the structure with a thin highly absorbing layer of gadolinium are considered in the article. An analogue of the poisoning effect well known in reactor physics was found. The effect is stronger for the Nb/Gd/Nb system. Despite of this effect, for a Nb/Gd bilayer and a Nb/Gd/Nb trilayer placed between Al2O3 substrate and Cu layer, it is shown theoretically and experimentally that one order of magnitude enhancement of neutron density is possible in the vicinity of the Gd layer. This enhancement makes it possible to study domain formation in the Gd layer under transition of the Nb layer(s) into the superconducting state (cryptoferromagnetic phase)., Comment: 5 pages, 2 figures

We report an investigation of the structural and electronic properties of hybrid superconductor/ferromagnet (S/F) bilayers of composition Nb/Cu$_{60}$Ni$_{40}$ prepared by magnetron sputtering. X-ray and neutron reflectometry show that both the overall interfacial roughness and vertical correlations of the roughness of different interfaces are lower for heterostructures deposited on Al$_2$O$_3$(1$\bar{1}$02) substrates than for those deposited on Si(111). Mutual inductance experiments were then used to study the influence of the interfacial roughness on the superconducting transition temperature, $T_C$. These measurements revealed a $\sim$ 4% higher $T_C$ in heterostructures deposited on Al$_2$O$_3$, compared to those on Si. We attribute this effect to a higher mean-free path of electrons in the S layer, caused by a suppression of diffusive scattering at the interfaces. However, the dependence of the $T_C$ on the thickness of the ferromagnetic layer is not significantly different in the two systems, indicating a weak influence of the interfacial roughness on the transparency for Cooper pairs., Comment: 8 pages, 5 figures

We use transport and neutron scattering to study electronic, structural, and magnetic properties of the electron-doped BaFe$_{2-x}$Ni$_x$As$_2$ iron pnictides in the external stress free detwinned state. Using a specially designed in-situ mechanical detwinning device, we demonstrate that the in-plane resistivity anisotropy observed in the uniaxial strained tetragonal state of BaFe$_{2-x}$Ni$_x$As$_2$ below a temperature $T^\ast$, previously identified as a signature of the electronic nematic phase, is also present in the stress free tetragonal phase below $T^{\ast\ast}$ ($

We use neutron resonance spin echo and Larmor diffraction to study the effect of uniaxial pressure on the tetragonal-to-orthorhombic structural ($T_s$) and antiferromagnetic (AF) phase transitions in iron pnictides BaFe$_{2-x}$Ni$_{x}$As$_{2}$ ($x=0,0.03,0.12$), SrFe$_{1.97}$Ni$_{0.03}$As$_2$, and BaFe$_2$(As$_{0.7}$P$_{0.3}$)$_2$. In antiferromagnetically ordered BaFe$_{2-x}$Ni$_{x}$As$_{2}$ and SrFe$_{1.97}$Ni$_{0.03}$As$_2$ with $T_N$ and $T_s$ ($T_N\leq T_s$), a uniaxial pressure necessary to detwin the sample also increases $T_N$, smears out the structural transition, and induces an orthorhombic lattice distortion at all temperatures. By comparing temperature and doping dependence of the pressure induced lattice parameter changes with the elastoresistance and nematic susceptibility obtained from transport and ultrasonic measurements, we conclude that the in-plane resistivity anisotropy found in the paramagnetic state of electron underdoped iron pnictides depends sensitively on the nature of the magnetic phase transition and a strong coupling between the uniaxial pressure induced lattice distortion and electronic nematic susceptibility., Comment: 18 pages, 15 figures

We use nuclear magnetic resonance (NMR), high-resolution x-ray and neutron scattering to study structural and magnetic phase transitions in phosphorus-doped BaFe$_2$(As$_{1-x}$P$_x$)$_2$. Previous transport, NMR, specific heat, and magnetic penetration depth measurements have provided compelling evidence for the presence of a quantum critical point (QCP) near optimal superconductivity at $x=0.3$. However, we show that the tetragonal-to-orthorhombic structural ($T_s$) and paramagnetic to antiferromagnetic (AF, $T_N$) transitions in BaFe$_2$(As$_{1-x}$P$_x$)$_2$ are always coupled and approach to $T_N\approx T_s \ge T_c$ ($\approx 29$ K) for $x=0.29$ before vanishing abruptly for $x\ge 0.3$. These results suggest that AF order in BaFe$_2$(As$_{1-x}$P$_x$)$_2$ disappears in a weakly first order fashion near optimal superconductivity, much like the electron-doped iron pnictides with an avoided QCP., Comment: 8 pages, 8 figures. including supplementary, Accepted by Physical Review Letters