Search

The ability to understand and engineer molecular structures relies on having accurate descriptions of the energy as a function of atomic coordinates. Here we outline a new paradigm for deriving energy functions of hyperdimensional molecular systems, which involves generating data for low-dimensional systems in virtual reality (VR) to then efficiently train atomic neural networks (ANNs). This generates high quality data for specific areas of interest within the hyperdimensional space that characterizes a molecule's potential energy surface (PES). We demonstrate the utility of this approach by gathering data within VR to train ANNs on chemical reactions involving fewer than 8 heavy atoms. This strategy enables us to predict the energies of much higher-dimensional systems, e.g. containing nearly 100 atoms. Training on datasets containing only 15K geometries, this approach generates mean absolute errors around 2 kcal/mol. This represents one of the first times that an ANN-PES for a large reactive radical has been generated using such a small dataset. Our results suggest VR enables the intelligent curation of high-quality data, which accelerates the learning process.

As molecular scientists have made progress in their ability to engineer nano-scale molecular structure, we are facing new challenges in our ability to engineer molecular dynamics (MD) and flexibility. Dynamics at the molecular scale differs from the familiar mechanics of everyday objects, because it involves a complicated, highly correlated, and three-dimensional many-body dynamical choreography which is often non-intuitive even for highly trained researchers. We recently described how interactive molecular dynamics in virtual reality (iMD-VR) can help to meet this challenge, enabling researchers to manipulate real-time MD simulations of flexible structures in 3D. In this article, we outline efforts to extend immersive technologies to the molecular sciences, and we introduce 'Narupa', a flexible, open-source, multi-person iMD-VR software framework which enables groups of researchers to simultaneously cohabit real-time simulation environments to interactively visualize and manipulate the dynamics of molecular structures with atomic-level precision. We outline several application domains where iMD-VR is facilitating research, communication, and creative approaches within the molecular sciences, including training machines to learn reactive potential energy surfaces (PESs), biomolecular conformational sampling, protein-ligand binding, reaction discovery using 'on-the-fly' quantum chemistry, and transport dynamics in materials. We touch on iMD-VR's various cognitive and perceptual affordances, and how these provide research insight for molecular systems. By synergistically combining human spatial reasoning and design insight with computational automation, technologies like iMD-VR have the potential to improve our ability to understand, engineer, and communicate microscopic dynamical behavior, offering the potential to usher in a new paradigm for engineering molecules and nano-architectures.

Whilst the primary bottleneck to a number of computational workflows was not so long ago limited by processing power, the rise of machine learning technologies has resulted in a paradigm shift which places increasing value on issues related to data curation - i.e., data size, quality, bias, format, and coverage. Increasingly, data-related issues are equally as important as the algorithmic methods used to process and learn from the data. Here we introduce an open source GPU-accelerated neural network (NN) framework for learning reactive potential energy surfaces (PESs), and investigate the use of real-time interactive ab initio molecular dynamics in virtual reality (iMD-VR) as a new strategy for rapidly sampling geometries along reaction pathways which can be used to train NNs to learn reactive PESs. Focussing on hydrogen abstraction reactions of CN radical with isopentane, we compare the performance of NNs trained using iMD-VR data versus NNs trained using a more traditional method, namely molecular dynamics (MD) constrained to sample a predefined grid of points along hydrogen abstraction reaction coordinates. Both the NN trained using iMD-VR data and the NN trained using the constrained MD data reproduce important qualitative features of the reactive PESs, such as a low and early barrier to abstraction. Quantitatively, learning is sensitive to the training dataset. Our results show that user-sampled structures obtained with the quantum chemical iMD-VR machinery enable better sampling in the vicinity of the minimum energy path (MEP). As a result, the NN trained on the iMD-VR data does very well predicting energies in the vicinity of the MEP, but less well predicting energies for 'off-path' structures. The NN trained on the constrained MD data does better in predicting energies for 'off-path' structures, given that it included a number of such structures in its training set.

We have used a mesh experiment in order to measure the sub-pixel structure of the EPIC MOS CCDs on-board the XMM/NEWTON satellite. The EPIC MOS CCDs have 40 $\mu$m-square pixels which have an open electrode structure in order to improve the detection efficiency for low-energy X-rays. We obtained restored pixel images for various X-ray event grades (e.g. split-pixel events, single pixel events, etc.) at various X-ray energies. We confirmed that the open electrode structure results in a distorted horizontal pixel boundary. The open electrode region generates both single pixel events and vertically split events, but no horizontally split events. Because the single pixel events usually show the best energy resolution, we discuss a method of increasing the fraction of single pixel events from the open electrode region. Furthermore, we have directly measured the thickness of the electrodes and dead-layers by comparing spectra from the open electrode region with those from the other regions: electrodes, electrode finger and channel stop. We can say that EPIC MOS CCDs are more radiation hard than front-illumination chips of ACIS on-board Chandra X-ray Observatory because of their extra absorption thickness above the charge transfer channel. We calcurated the mean pixel response and found that our estimation has a good agreement with that of the ground calibration of EPIC MOS CCD., Comment: 20pages including 2 tables, 10 figures,Accepted for publication in : Nuclear Instruments and Methods in Physics Research A

The bright eclipsing and bursting low-mass X-ray binary EXO 0748-676 has been observed at several occasions by XMM-Newton during the initial calibration and performance verification (CAL/PV) phase. We present here the results obtained from observations with the EPIC cameras. Apart from several type-I X-ray bursts, the source shows a high degree of variability with the presence of soft flares. The wide energy coverage and high sensitivity of XMM-Newton allows for the first time a detailed description of the spectral variability. The source is found to be the superposition of a central (~2 10^8 cm) Comptonized emission, most probably a corona surrounding the inner edge of an accretion disk, associated with a more extended (~3 10^10 cm) thermal halo at a typical temperature of ~0.6 keV with an indication of non-solar abundances. Most of the variations of the source can be accounted for by a variable absorption affecting only the central comptonized component and reaching up to NH ~1.3 10^23 cm^{-2}. The characteristics of the surrounding halo are found compatible with an irradiated atmosphere of an accretion disc which intercepts the central emission due to the system high inclination., Comment: 6 pages, 4 figures, accepted for publication in A&A Letters, XMM special issue

Results from observations of the young oxygen-rich supernova remnant SNR 0102-72.3 in the Small Magellanic Cloud during the calibration phase of the XMM-Newton Observatory are presented. Both EPIC-PN and MOS observations show a ringlike structure with a radius of ~15'' already known from Einstein, ROSAT and Chandra observations. Spectra of the entire SNR as well as parts in the eastern half were analyzed confirming shocked hot plasma in non-uniform ionization stages as the origin of the X-ray emission. The spectra differ in the northeastern and the southeastern part of the X-ray ring, showing emission line features of different strength. The temperature in the northeastern part is significantly higher than in the southeast, reflected by the lines of higher ionization stages and the harder continuum. Comparison to radio data shows the forward shock of the blast wave dominating in the northern part of the SNR, while the southern emission is most likely produced by the recently formed reverse shock in the ejecta. In the case of the overall spectrum of SNR 0102-72.3, the two-temperature non-equilibrium ionization model is more consistent with the data in comparison to the single plane-parallel shock model. The structure of SNR 0102-72.3 is complex due to variations in shock propagation leading to spatially differing X-ray spectra., Comment: 5 pages, accepted for publication in the special A&A Letters issue for XMM-Newton

This paper presents the XMM-Newton first-light observations of the Hickson-16 compact group of galaxies. Groups are possibly the oldest large-scale structures in the Universe, pre-dating clusters of galaxies, and are highly evolved. This group of small galaxies, at a redshift of 0.0132 (or 80 Mpc) is exceptional in the having the highest concentration of starburst or AGN activity in the nearby Universe. So it is a veritable laboratory for the study of the relationship between galaxy interactions and nuclear activity. Previous optical emission line studies indicated a strong ionising continuum in the galaxies, but its origin, whether from starbursts, or AGN, was unclear. Combined imaging and spectroscopy with the EPIC X-ray CCDs unequivocally reveals a heavily obscured AGN and a separately identified thermal (starburst) plasma, in NGC 835, NGC 833 and NGC 839. NGC 838 shows only starburst thermal emission. Starbursts and AGN can evidently coexist in members of this highly evolved system of merged and merging galaxies, implying a high probability for the formation of AGN as well as starbursts in post-merger galaxies., Comment: 6 pages, accepted for publication in A&A Special Letters Edition (First results from XMM-Newton)

We present XMM-Newton observations of the high redshift z=3.104, radio-loud quasar PKS 0537-286. The EPIC CCD cameras provide the highest signal-to-noise spectrum of a high-z quasar to date. The EPIC observations show that PKS 0537-286 is extremely X-ray luminous (Lx=2x10^47 erg/s), with an unusually hard X-ray spectrum (Gamma=1.27+/-0.02). The flat power-law emission extends over the whole observed energy range; there is no evidence of intrinsic absorption, which has been claimed in PKS 0537-286 and other high z quasars. However, there is evidence for weak Compton reflection. A redshifted iron K line, observed at 1.5 keV - corresponding to 6.15 keV in the quasar rest frame - is detected at 95% confidence. If confirmed, this is the most distant iron K line known. The line equivalent width is small (33eV), consistent with the `X-ray Baldwin effect' observed in other luminous quasars. The reflected continuum is also weak (R=0.25). We find the overall spectral energy distribution of PKS 0537-286 is dominated by the X-ray emission, which, together with the flat power-law and weak reflection features, suggests that the X-radiation from PKS 0537-286 is dominated by inverse Compton emission associated with a face-on relativistic jet., Comment: 6 pages, accepted for publication in Astronomy and Astrophysics Letters

We report here the results of a mesh experiment to measure the subpixel structure of the EPIC MOS CCDs on board the XMM X-ray observatory. The pixel size is 40$\mu$m square while the mesh hole spacing is 48$\mu$m, a combination quite different from our standard mesh experiment. We have verified that this combination functions properly and have analyzed the CCD structure with sub-pixel resolution. The EPIC MOS CCD has an open electrode structure to improve detection efficiency at low energies. We obtained the distribution of various grades of X-ray events inside the pixel. A horizontally split two-pixel event is generated near the channel stop which forms a straight vertical pixel boundary whereas a vertically split two-pixel event is generated where the potential due to the thinned gate structure forms a wavy horizontal pixel boundary. Therefore, the effective pixel shape is not a square but is distorted. The distribution of X-ray events clearly shows that the two etched regions in each pixel, separated by the bridging finger of the enlarged (open) electrode. We measured the difference in X-ray transmission between the conventional and open regions of the pixel using O-K and Cu-L X-ray emission lines, and found it to be consistent with an electrode thickness comprising $0.2\pm0.1\mu$m of Si and $0.6\pm0.2\mu$m of SiO2., Comment: 9 pages, 5 figures, accepted for publication in Nucl. Instr. and Methods A