• Traditional approaches to semantic parsing (SP) work by training individual models for each available parallel dataset of text-meaning pairs. In this paper, we explore the idea of polyglot semantic translation, or learning semantic parsing models that are trained on multiple datasets and natural languages. In particular, we focus on translating text to code signature representations using the software component datasets of Richardson and Kuhn (2017a,b). The advantage of such models is that they can be used for parsing a wide variety of input natural languages and output programming languages, or mixed input languages, using a single unified model. To facilitate modeling of this type, we develop a novel graph-based decoding framework that achieves state-of-the-art performance on the above datasets, and apply this method to two other benchmark SP tasks.
  • We present early laboratory simulations and extensive on-sky tests validating of the performance of a shaped pupil coronagraph (SPC) behind an extreme-AO corrected beam of the Subaru Coronagraphic Extreme Adaptive Optics (SCExAO) system. In tests with the SCExAO internal source/wavefront error simulator, the normalized intensity profile for the SPC degrades more slowly than for the Lyot coronagraph as low-order aberrations reduce the Strehl ratio from extremely high values (S.R. $\sim$ 0.93--0.99) to those characteristic of current ground-based extreme AO systems (S.R. $\sim$ 0.74--0.93) and then slightly lower values down to S.R. $\sim$ 0.57. On-sky SCExAO data taken with the SPC and other coronagraphs for brown dwarf/planet-hosting stars HD 1160 and HR 8799 provide further evidence for the SPC's robustness to low-order aberrations. From H-band Strehl ratios of 80% to 70%, the Lyot coronagraph's performance vs. that of the SPC may degrade even faster on sky than is seen in our internal source simulations. The 5-$\sigma$ contrast also degrades faster (by a factor of two) for the Lyot than the SPC. The SPC we use was designed as a technology demonstrator only, with a contrast floor, throughput, and outer working angle poorly matched for SCExAO's current AO performance and poorly tuned for imaging the HR 8799 planets. Nevertheless, we detect HR 8799 cde with SCExAO/CHARIS using the SPC in broadband mode, where the S/N for planet e is within 30% of that obtained using the vortex coronagraph. The shaped-pupil coronagraph is a promising design demonstrated to be robust in the presence of low-order aberrations and may be well-suited for future ground and space-based direct imaging observations, especially those focused on follow-up exoplanet characterization and technology demonstration of deep contrast within well-defined regions of the image plane.
  • The vector vortex is a coronagraphic imaging mode of the recently commissioned Subaru Coronagraphic Extreme-Adaptive Optics (SCExAO) platform on the 8-m Subaru Telescope. This multi-purpose high-contrast visible and near-infrared (R- to K-band) instrument is not only intended to serve as a VLT-class "planet-imager" instrument in the Northern hemisphere, but also to operate as a technology demonstration testbed ahead of the ELTs-era, with a particular emphasis on small inner-working angle (IWA) coronagraphic capabilities. The given priority to small-IWA imaging led to the early design choice to incorporate focal-plane phase-mask coronagraphs. In this context, a test H-band vector vortex liquid crystal polymer waveplate was provided to SCExAO, to allow a one-to-one comparison of different small-IWA techniques on the same telescope instrument, before considering further steps. Here we present a detailed overview of the vector vortex coronagraph, from its installation and performances on the SCExAO optical bench, to the on-sky results in the extreme AO regime, as of late 2016/early 2017. To this purpose, we also provide a few recent on-sky imaging examples, notably high-contrast ADI detection of the planetary-mass companion \kappa Andromedae b, with a signal-to-noise ratio above 100 reached in less than 10 mn exposure time.
  • In this paper, we describe Function Assistant, a lightweight Python-based toolkit for querying and exploring source code repositories using natural language. The toolkit is designed to help end-users of a target API quickly find information about functions through high-level natural language queries and descriptions. For a given text query and background API, the tool finds candidate functions by performing a translation from the text to known representations in the API using the semantic parsing approach of Richardson and Kuhn (2017). Translations are automatically learned from example text-code pairs in example APIs. The toolkit includes features for building translation pipelines and query engines for arbitrary source code projects. To explore this last feature, we perform new experiments on 27 well-known Python projects hosted on Github.
  • We propose a new shared task for tactical data-to-text generation in the domain of source code libraries. Specifically, we focus on text generation of function descriptions from example software projects. Data is drawn from existing resources used for studying the related problem of semantic parser induction (Richardson and Kuhn, 2017b; Richardson and Kuhn, 2017a), and spans a wide variety of both natural languages and programming languages. In this paper, we describe these existing resources, which will serve as training and development data for the task, and discuss plans for building new independent test sets.
  • Direct imaging of exoplanets or circumstellar disk material requires extreme contrast at the 10-6 to 10-12 levels at < 100 mas angular separation from the star. Focal-plane mask (FPM) coronagraphic imaging has played a key role in this field, taking advantage of progress in Adaptive Optics on ground-based 8+m class telescopes. However, large telescope entrance pupils usually consist of complex, sometimes segmented, non-ideal apertures, which include a central obstruction for the secondary mirror and its support structure. In practice, this negatively impacts wavefront quality and coronagraphic performance, in terms of achievable contrast and inner working angle. Recent theoretical works on structured darkness have shown that solutions for FPM phase profiles, optimized for non-ideal apertures, can be numerically derived. Here we present and discuss a first experimental validation of this concept, using reflective liquid crystal spatial light modulators as adaptive FPM coronagraphs.
  • We consider the problem of translating high-level textual descriptions to formal representations in technical documentation as part of an effort to model the meaning of such documentation. We focus specifically on the problem of learning translational correspondences between text descriptions and grounded representations in the target documentation, such as formal representation of functions or code templates. Our approach exploits the parallel nature of such documentation, or the tight coupling between high-level text and the low-level representations we aim to learn. Data is collected by mining technical documents for such parallel text-representation pairs, which we use to train a simple semantic parsing model. We report new baseline results on sixteen novel datasets, including the standard library documentation for nine popular programming languages across seven natural languages, and a small collection of Unix utility manuals.
  • We present $H$-band scattered light imaging of a bright debris disk around the A0 star HD 36546 obtained from the Subaru Coronagraphic Extreme Adaptive Optics (SCExAO) system with data recorded by the HiCIAO camera using the vector vortex coronagraph. SCExAO traces the disk from $r$ $\sim$ 0.3" to $r$ $\sim$ 1" (34--114 au). The disk is oriented in a near east-west direction (PA $\sim$ 75$^{o}$), is inclined by $i$ $\sim$ 70--75$^{o}$ and is strongly forward-scattering ($g$ $>$ 0.5). It is an extended disk rather than a sharp ring; a second, diffuse dust population extends from the disk's eastern side. While HD 36546 intrinsic properties are consistent with a wide age range ($t$ $\sim$ 1--250 $Myr$), its kinematics and analysis of coeval stars suggest a young age (3--10 $Myr$) and a possible connection to Taurus-Auriga's star formation history. SCExAO's planet-to-star contrast ratios are comparable to the first-light Gemini Planet Imager contrasts; for an age of 10 $Myr$, we rule out planets with masses comparable to HR 8799 b beyond a projected separation of 23 au. A massive icy planetesimal disk or an unseen superjovian planet at $r$ $>$ 20 au may explain the disk's visibility. The HD 36546 debris disk may be the youngest debris disk yet imaged, is the first newly-identified object from the now-operational SCExAO extreme AO system, is ideally suited for spectroscopic follow up with SCExAO/CHARIS in 2017, and may be a key probe of icy planet formation and planet-disk interactions.
  • We present a new instrument, the "Stellar Double Coronagraph" (SDC), a flexible coronagraphic platform. Designed for Palomar Observatory's 200" Hale telescope, its two focal and pupil planes allow for a number of different observing configurations, including multiple vortex coronagraphs in series for improved contrast at small angles. We describe the motivation, design, observing modes, wavefront control approaches, data reduction pipeline, and early science results. We also discuss future directions for the instrument.
  • We report the discovery of a low-mass companion to HR3549, an A0V star surrounded by a debris disk with a warm excess detected by WISE at 22 $\mu$m ($10\sigma$ significance). We imaged HR3549 B in the L-band with NAOS-CONICA, the adaptive optics infrared camera of the Very Large Telescope, in January 2013 and confirmed its common proper motion in January 2015. The companion is at a projected separation of $\simeq 80$ AU and position angle of $\simeq 157^\circ$, so it is orbiting well beyond the warm disk inner edge of $r > 10$ AU. Our age estimate for this system corresponds to a companion mass in the range 15-80 $M_J$, spanning the brown dwarf regime, and so HR3549 B is another recent addition to the growing list of brown dwarf desert objects with extreme mass ratios. The simultaneous presence of a warm disk and a brown dwarf around HR3549 provides interesting empirical constraints on models of the formation of substellar companions.
  • We present a direct image of the innermost companion to the red giant delta Andromedae using the Stellar Double Coronagraph at the Palomar Observatory. We use a Markov chain Monte Carlo based algorithm to simultaneously reduce the data and perform astrometry and photometry of the companion. We determine that the companion is most likely a main-sequence K-type star and is certainly not the previously hypothesized white dwarf.
  • The ability to characterize exoplanets by spectroscopy of their atmospheres requires direct imaging techniques to isolate planet signal from the bright stellar glare. One of the limitations with the direct detection of exoplanets, either with ground- or space-based coronagraphs, is pointing errors and other low-order wavefront aberrations. The coronagraphic detection sensitivity at the diffraction limit therefore depends on how well low-order aberrations upstream of the focal plane mask are corrected. To prevent starlight leakage at the inner working angle of a phase mask coronagraph, we have introduced a Lyot-based low-order wavefront sensor (LLOWFS), which senses aberrations using the rejected starlight diffracted at the Lyot plane. In this paper, we present the implementation, testing and results of LLOWFS on the Subaru Coronagraphic Extreme Adaptive Optics system (SCExAO) at the Subaru Telescope. We have controlled thirty-five Zernike modes of a H-band vector vortex coronagraph in the laboratory and ten Zernike modes on sky with an integrator control law. We demonstrated a closed-loop pointing residual of 0.02 mas in the laboratory and 0.15 mas on sky for data sampled using the minimal 2-second exposure time of the science camera. We have also integrated the LLOWFS in the visible high-order control loop of SCExAO, which in closed-loop operation has validated the correction of the non-common path pointing errors between the infrared science channel and the visible wavefront sensing channel with pointing residual of 0.23 mas on sky.
  • We report on recent Ks-band interferometric observations of the young pre-main-sequence star AB Aurigae obtained with the Palomar Fiber Nuller (PFN). Reaching a contrast of a few 10-4 inside a field of view extending from 35 to 275 mas (5-40AU at AB Aur's distance), the PFN is able to explore angular scales that are intermediate between those accessed by coronagraphic imaging and long baseline interferometry. This intermediate region is of special interest given that many young stellar objects are believed to harbor extended halos at such angular scales. Using destructive interference (nulling) between two sub-apertures of the Palomar 200 inch telescope and rotating the telescope pupil, we measured a resolved circumstellar excess at all probed azimuth angles. The astrophysical null measured over the full rotation is fairly constant, with a mean value of 1.52%, and a slight additional azimuthal modulation of +/-0.2%. The isotropic astrophysical null is indicative of circumstellar emission dominated by an azimuthally extended source, possibly a halo, or one or more rings of dust, accounting for several percent of the total Ks-band flux. The modest azimuthal variation may be explained by some skewness or anisotropy of the spatially extended source, e.g., an elliptical or spiral geometry, or clumping, but it could also be due to the presence of a point source located at a separation of ~120 mas (17AU) with ~6*10-3 of the stellar flux. We combine our results with previous Infrared Optical Telescope Array observations of AB Aur at H band, and demonstrate that a dust ring located at ~30 mas (4.3AU) represents the best-fitting model to explain both sets of visibilities.We are also able to test a few previously hypothesized models of the incoherent component evident at longer interferometric baselines.
  • We present a new approach to stochastic modeling of constraint-based grammars that is based on log-linear models and uses EM for estimation from unannotated data. The techniques are applied to an LFG grammar for German. Evaluation on an exact match task yields 86% precision for an ambiguity rate of 5.4, and 90% precision on a subcat frame match for an ambiguity rate of 25. Experimental comparison to training from a parsebank shows a 10% gain from EM training. Also, a new class-based grammar lexicalization is presented, showing a 10% gain over unlexicalized models.