We present nebular-phase spectra of the Type Ia supernova (SN Ia) 2016brx, a member of the 1991bg-like subclass that lies at the faint end of the SN Ia luminosity function. Nebular spectra are available for only three other 1991bg-like SNe, and their Co line centers are all within <~ 500 km/s of each other. In contrast, the nebular Co line center of SN 2016brx is blue-shifted by >1500 km/s compared to them and by ~1200 km/s compared to the rest frame. This is a significant shift relative to the narrow nebular line velocity dispersion of <~ 2000 km/s of these SNe. The large range of nebular line shifts implies that the Ni56 in the ejecta of SN 1991bg-like events is off-center by ~1000 km/s rather than universally centrally confined as previously suggested. With the addition of SN 2016brx, the Co nebular line shapes of 1991bg-like objects appear to connect with the brighter SNe Ia that show double-peak profiles, hinting at a continuous distribution of line profiles among SNe Ia. One class of models to produce both off-center and bi-modal Ni56 distributions is collisions of white dwarfs with unequal and equal masses.

Many Natural Language Processing and Computational Linguistics applications involves the generation of new texts based on some existing texts, such as summarization, text simplification and machine translation. However, there has been a serious problem haunting these applications for decades, that is, how to automatically and accurately assess quality of these applications. In this paper, we will present some preliminary results on one especially useful and challenging problem in NLP system evaluation: how to pinpoint content differences of two text passages (especially for large pas-sages such as articles and books). Our idea is intuitive and very different from existing approaches. We treat one text passage as a small knowledge base, and ask it a large number of questions to exhaustively identify all content points in it. By comparing the correctly answered questions from two text passages, we will be able to compare their content precisely. The experiment using 2007 DUC summarization corpus clearly shows promising results.

We present optical observations of an exceptional Type II supernova, ASASSN-15nx. Its peak luminosity at M_V = -20 mag is between those of typical core-collapse supernovae (CCSNe) and super-luminous supernovae (SLSNe). The post-peak multi-band optical light curves show a remarkably long, linear decline with a steep slope of 2.5 mag / 100 d (i.e., an exponential decline in flux), through to the end of our observations at phase ~ 260d. In contrast, the light curves of the two major subclasses of SNe II, Type II-P ("plateau") and Type II-L ("linear"), always show breaks in slopes at phase ~100 d, before settling onto Co56 radioactive decay tails with a decline rate typically of about 1 mag / 100 d. The spectra of ASASSN-15nx do not exhibit the narrow emission-line features characteristic of Type IIn SNe, which can have a wide variety of light-curve shapes usually attributed to strong interactions with circumstellar medium (CSM). ASASSN-15nx also exhibits a number of peculiar spectroscopic properties, including a relatively weak H-alpha emission line that has an unusually triangularly-shaped profile with no absorption component. The physical origin giving rise to these peculiarities is unclear, while the long and linear post-peak light curve without a break likely suggests a single dominant powering mechanism. The light curve is consistent with being predominantly powered by the decay of a large amount of Ni56 (M_Ni = 1.6 +/- 0.2 M_sun), and substantial gamma-ray escape from the ejecta is needed to explain the steep light-curve slope, which is possibly due to a low-mass hydrogen envelope of the progenitor. Another possibility is strong CSM interactions mostly powering the light curve, and the CSM needs to be sculpted to produce the unique light-curve shape as well as avoid producing SN IIn-like narrow emission lines.

Topology has been revealed to play a fundamental role in physics in the past decades. Topological insulators have unconventional gapless edge states where disorder-induced back-scattering is suppressed. In photonics, such edge states lead to unidirectional waveguides which are useful for integrated photonic chips. Cavity modes, another type of fundamental components in photonic chips, however, are not protected by band topology because of their lower dimensions. Here we demonstrate that concurrent wavevector-space and real-space topology, dubbed as the "dual-topology", can lead to light-trapping in lower-dimensions. The resultant photonic bound state emerges as a Jackiw-Rebbi soliton mode localized on a dislocation in a two-dimensional (2D) photonic crystal, as predicted theoretically and discovered experimentally. Such a strongly-confined 0D localized mode, which is solely due to the topological mechanism, is found to be robust against perturbations. Our study unveils a new mechanism for topological light-trapping in lower-dimensions, which is valuable for fundamental physics and a variety of applications in photonics.

Hydrogen-poor superluminous supernovae (SLSNe-I) have been predominantly found in low-metallicity, star-forming dwarf galaxies. Here we identify Gaia17biu/SN 2017egm as an SLSN-I occurring in a "normal" spiral galaxy (NGC 3191) in terms of stellar mass (several times 10^10 M_sun) and metallicity (roughly Solar). At redshift z=0.031, Gaia17biu is also the lowest redshift SLSN-I to date, and the absence of a larger population of SLSNe-I in dwarf galaxies of similar redshift suggests that metallicity is likely less important to the production of SLSNe-I than previously believed. With the smallest distance and highest apparent brightness for an SLSN-I, we are able to study Gaia17biu in unprecedented detail. Its pre-peak near-ultraviolet to optical color is similar to that of Gaia16apd and among the bluest observed for an SLSN-I while its peak luminosity (M_g = -21 mag) is substantially lower than Gaia16apd. Thanks to the high signal-to-noise ratios of our spectra, we identify several new spectroscopic features that may help to probe the properties of these enigmatic explosions. We detect polarization at the ~0.5% level that is not strongly dependent on wavelength, suggesting a modest, global departure from spherical symmetry. In addition, we put the tightest upper limit yet on the radio luminosity of an SLSN-I with <5.4x10^26 erg/s/Hz (at 10 GHz), which is almost a factor of 40 better than previous upper limits and one of the few measured at an early stage in the evolution of an SLSN-I. This limit largely rules out an association of this SLSNe-I with known populations of gamma-ray burst (GRB) like central engines.

We present photometric and spectroscopic follow-up observations of the highly luminous Type Ibn supernova ASASSN-14ms, which was discovered on UT 2014-12-26.61 at $m_V \sim 16.5$. With a peak absolute $V$-band magnitude brighter than $-20.5$, a peak bolometric luminosity of $1.7 \times 10^{44}$ ergs s$^{-1}$, and a total radiated energy of $2.1 \times 10^{50}$ ergs, ASASSN-14ms is one of the most luminous Type Ibn supernovae yet discovered. In simple models, the most likely power source for this event is a combination of the radioactive decay of $^{56}$Ni and $^{56}$Co at late times and the interaction of supernova ejecta with the progenitor's circumstellar medium at early times, although we cannot rule out the possibility of a magnetar-powered light curve. The presence of a dense circumstellar medium is indicated by the intermediate-width He I features in the spectra. The faint ($m_g \sim 21.6$) host galaxy SDSS J130408.52+521846.4 has an oxygen abundance below $12+\log(O/H) \lesssim 8.3$, a stellar mass of $M_* \sim 2.6 \times 10^8 M_{\odot}$, and a star formation rate of $\textrm{SFR} \sim 0.02$ $M_{\odot}$ yr$^{-1}$.

An important and difficult challenge in building computational models for narratives is the automatic evaluation of narrative quality. Quality evaluation connects narrative understanding and generation as generation systems need to evaluate their own products. To circumvent difficulties in acquiring annotations, we employ upvotes in social media as an approximate measure for story quality. We collected 54,484 answers from a crowd-powered question-and-answer website, Quora, and then used active learning to build a classifier that labeled 28,320 answers as stories. To predict the number of upvotes without the use of social network features, we create neural networks that model textual regions and the interdependence among regions, which serve as strong benchmarks for future research. To our best knowledge, this is the first large-scale study for automatic evaluation of narrative quality.

Accelerometer measurements are the prime type of sensor information most think of when seeking to measure physical activity. On the market, there are many fitness measuring devices which aim to track calories burned and steps counted through the use of accelerometers. These measurements, though good enough for the average consumer, are noisy and unreliable in terms of the precision of measurement needed in a scientific setting. The contribution of this paper is an innovative and highly accurate regression method which uses an intermediary two-stage classification step to better direct the regression of energy expenditure values from accelerometer counts. We show that through an additional unsupervised layer of intermediate feature construction, we can leverage latent patterns within accelerometer counts to provide better grounds for activity classification than expert-constructed timeseries features. For this, our approach utilizes a mathematical model originating in natural language processing, the bag-of-words model, that has in the past years been appearing in diverse disciplines outside of the natural language processing field such as image processing. Further emphasizing the natural language connection to stochastics, we use a gaussian mixture model to learn the dictionary upon which the bag-of-words model is built. Moreover, we show that with the addition of these features, we're able to improve regression root mean-squared error of energy expenditure by approximately 1.4 units over existing state-of-the-art methods.

This catalog summarizes information for all supernovae discovered by the All-Sky Automated Survey for SuperNovae (ASAS-SN) and all other bright ($m_{peak}\leq17$), spectroscopically confirmed supernovae discovered in 2016. We then gather the near-IR through UV magnitudes of all host galaxies and the offsets of the supernovae from the centers of their hosts from public databases. We illustrate the results using a sample that now totals 668 supernovae discovered since 2014 May 1, including the supernovae from our previous catalogs, with type distributions closely matching those of the ideal magnitude limited sample from Li et al. (2011). This is the third of a series of yearly papers on bright supernovae and their hosts from the ASAS-SN team.

While we once thought of cancer as single monolithic diseases affecting a specific organ site, we now understand that there are many subtypes of cancer defined by unique patterns of gene mutations. These gene mutational data, which can be more reliably obtained than gene expression data, help to determine how the subtypes develop, evolve, and respond to therapies. Different from dense continuous-value gene expression data, which most existing cancer subtype discovery algorithms use, somatic mutational data are extremely sparse and heterogeneous, because there are less than 0.5\% mutated genes in discrete value 1/0 out of 20,000 human protein-coding genes, and identical mutated genes are rarely shared by cancer patients. Our focus is to search for cancer subtypes from extremely sparse and high dimensional gene mutational data in discrete 1 and 0 values using unsupervised learning. We propose a new network-based distance metric. We project cancer patients' mutational profile into their gene network structure and measure the distance between two patients using the similarity between genes and between the gene vertexes of the patients in the network. Experimental results in synthetic data and real-world data show that our approach outperforms the top competitors in cancer subtype discovery. Furthermore, our approach can identify cancer subtypes that cannot be detected by other clustering algorithms in real cancer data.

Given its peak luminosity and early-time spectra, ASASSN-15lh was classified as the most luminous supernova (SN) ever discovered (Dong et al. 2016). Here we report a UV rebrightening of ASASSN-15lh observed with Swift during our follow-up campaign. The rebrightening began at $t \simeq$ 90 days (observer frame) after the primary peak and was followed by a $\sim 120$-day long plateau in the bolometric luminosity, before starting to fade again at $t\simeq 210$ days. ASASSN-15lh rebrightened in the Swift UV bands by $\Delta m_{UVW2} \simeq -1.75$ mag, $\Delta m_{UVM2} \simeq -1.25$ mag, and $\Delta m_{UVW1} \simeq -0.8$ mag, but did not rebrighten in the optical bands. Throughout its initial decline, subsequent rebrightening, and renewed decline, the spectra did not show evidence of interactions between the ejecta and circumstellar medium (CSM) such as narrow emission lines. There are hints of weak H$\alpha$ emission at late-times, but Margutti et al. (2016) have shown that it is narrow line emission consistent with star formation in the host nucleus. By fitting a blackbody we find that during the rebrightening the effective photospheric temperature increased from $T_{BB} \simeq 11000$ K to $T_{BB} \simeq 18000$ K. Over the $\sim$ 550 days since its detection, ASASSN-15lh has radiated $\sim 1.7 - 1.9 \times 10^{52}$ ergs. Although its physical nature remains uncertain, the evolution of ASASSN-15lh's photospheric radius, its radiated energy, and the implied event rate, are all more similar to those of H-poor superluminous supernovae (SLSNe-I) than to tidal disruption events (TDEs).

This manuscript presents information for all supernovae discovered by the All-Sky Automated Survey for SuperNovae (ASAS-SN) during 2015, its second full year of operations. The same information is presented for bright ($m_V\leq17$), spectroscopically confirmed supernovae discovered by other sources in 2015. As with the first ASAS-SN bright supernova catalog, we also present redshifts and near-UV through IR magnitudes for all supernova host galaxies in both samples. Combined with our previous catalog, this work comprises a complete catalog of 455 supernovae from multiple professional and amateur sources, allowing for population studies that were previously impossible. This is the second of a series of yearly papers on bright supernovae and their hosts from the ASAS-SN team.

This paper investigates a class of multi-player discrete games where each player aims to maximize its own utility function. Two particular challenges are considered. Firstly, each player is unaware of the structure of its utility function and the actions of other players, but is able to access the corresponding utility value given an action profile. Second, utility values are subject to delays and errors. We propose a robust adaptive learning algorithm which converges in probability to the set of action profiles which have maximal stochastic potential. Furthermore, the convergence rate of the proposed algorithm is quantified. When the interactions of the players consist of a weakly acyclic game, the convergence to the set of pure Nash equilibria is guaranteed. A set of numerical simulations are conducted to validate the algorithm performance.

We present ground-based and Swift photometric and spectroscopic observations of the tidal disruption event (TDE) ASASSN-15oi, discovered at the center of 2MASX J20390918-3045201 ($d\simeq216$ Mpc) by the All-Sky Automated Survey for SuperNovae (ASAS-SN). The source peaked at a bolometric luminosity of $L\simeq1.3\times10^{44}$ ergs s$^{-1}$ and radiated a total energy of $E\simeq6.6\times10^{50}$ ergs over the first $\sim3.5$ months of observations. The early optical/UV emission of the source can be fit by a blackbody with temperature increasing from $T\sim2\times10^4$ K to $T\sim4\times10^4$ K while the luminosity declines from $L\simeq1.3\times10^{44}$ ergs s$^{-1}$ to $L\simeq2.3\times10^{43}$ ergs s$^{-1}$, requiring the photosphere to be shrinking rapidly. The optical/UV luminosity decline during this period is most consistent with an exponential decline, $L\propto e^{-(t-t_0)/\tau}$, with $\tau \simeq46.5$ days for $t_0\simeq57241.6$ (MJD), while a power-law decline of $L\propto (t-t_0)^{-\alpha}$ with $t_0\simeq57212.3$ and $\alpha=1.62$ provides a moderately worse fit. ASASSN-15oi also exhibits roughly constant soft X-ray emission that is significantly weaker than the optical/UV emission. Spectra of the source show broad helium emission lines and strong blue continuum emission in early epochs, although these features fade rapidly and are not present $\sim3$ months after discovery. The early spectroscopic features and color evolution of ASASSN-15oi are consistent with a TDE, but the rapid spectral evolution is unique among optically-selected TDEs.

This paper proposes a chaotic map-based multicast scheme for multiuser speech wireless communication and implements it in an ARM platform. The scheme compresses the digital audio signal decoded by a sound card and then encrypts it with a three-level chaotic encryption scheme. First, the position of every bit of the compressed data is permuted randomly with a pseudo-random number sequence (PRNS) generated by a 6-D chaotic map. Then, the obtained data are further permuted in the level of byte with a PRNS generated by a 7-D chaotic map. Finally, it is operated with a multiround chaotic stream cipher. The whole system owns the following merits: the redundancy in the original audio file is reduced effectively and the corresponding unicity distance is increased; the balancing point between a high security level of the system and real-time conduction speed is achieved well. In the ARM implementation, the framework of communication of multicast-multiuser in a subnet and the Internet Group Manage Protocol is adopted to obtain the function of communication between one client and other ones. Comprehensive test results were provided to show the feasibility and security performance of the whole system.

Inferring topics from the overwhelming amount of short texts becomes a critical but challenging task for many content analysis tasks, such as content charactering, user interest profiling, and emerging topic detecting. Existing methods such as probabilistic latent semantic analysis (PLSA) and latent Dirichlet allocation (LDA) cannot solve this prob- lem very well since only very limited word co-occurrence information is available in short texts. This paper studies how to incorporate the external word correlation knowledge into short texts to improve the coherence of topic modeling. Based on recent results in word embeddings that learn se- mantically representations for words from a large corpus, we introduce a novel method, Embedding-based Topic Model (ETM), to learn latent topics from short texts. ETM not only solves the problem of very limited word co-occurrence information by aggregating short texts into long pseudo- texts, but also utilizes a Markov Random Field regularized model that gives correlated words a better chance to be put into the same topic. The experiments on real-world datasets validate the effectiveness of our model comparing with the state-of-the-art models.

Text simplification (TS) aims to reduce the lexical and structural complexity of a text, while still retaining the semantic meaning. Current automatic TS techniques are limited to either lexical-level applications or manually defining a large amount of rules. Since deep neural networks are powerful models that have achieved excellent performance over many difficult tasks, in this paper, we propose to use the Long Short-Term Memory (LSTM) Encoder-Decoder model for sentence level TS, which makes minimal assumptions about word sequence. We conduct preliminary experiments to find that the model is able to learn operation rules such as reversing, sorting and replacing from sequence pairs, which shows that the model may potentially discover and apply rules such as modifying sentence structure, substituting words, and removing words for TS.

Outbursts on young stars are usually interpreted as accretion bursts caused by instabilities in the disk or the star-disk connection. However, some protostellar outbursts may not fit into this framework. In this paper, we analyze optical and near-infrared spectra and photometry to characterize the 2015 outburst of the probable young star ASASSN-15qi. The $\sim 3.5$ mag brightening in the $V$ band was sudden, with an unresolved rise time of less than one day. The outburst decayed exponentially by 1 mag for 6 days and then gradually back to the pre-outburst level after 200 days. The outburst is dominated by emission from $\sim10,000$ K gas. An explosive release of energy accelerated matter from the star in all directions, seen in a spectacular cool, spherical wind with a maximum velocity of 1000 km/s. The wind and hot gas both disappeared as the outburst faded and the source returned to its quiescent F-star spectrum. Nebulosity near the star brightened with a delay of 10-20 days. Fluorescent excitation of H$_2$ is detected in emission from vibrational levels as high as $v=11$, also with a possible time delay in flux increase. The mid-infrared spectral energy distribution does not indicate the presence of warm dust emission, although the optical photospheric absorption and CO overtone emission could be related to a gaseous disk. Archival photometry reveals a prior outburst in 1976. Although we speculate about possible causes for this outburst, none of the explanations are compelling.

The current computer programmings encapsulate attributes and behaviours into objects, but miss the mechanism to support the connection among objects. A programming paradigm is presented to connect all objects. The connection supports communications. Protocols are defined to coordinate the behaviours between objects, which enable the interaction of objects across different platforms. The connection also provides an efficient mechanism to support the concurrency, parallelism, distribution, pipeline and adaptability, etc. They can be governed transparently, autonomously, even adaptively. In this paper, an implementation is also discussed to show the effectiveness of protocol programming.

We present the first survey of dense gas towards Planck Galactic Cold Clumps (PGCCs). Observations in the J=1-0 transitions of HCO+ and HCN towards 621 molecular cores associated with PGCCs were performed using the Purple Mountain Observatory 13.7-m telescope. Among them, 250 sources have detection, including 230 cores detected in HCO+ and 158 in HCN. Spectra of the J=1-0 transitions from CO, 13CO, and C18O at the centers of the 250 cores were extracted from previous mapping observations to construct a multi-line data set. The significantly low detection rate of asymmetric double-peaked profiles, together with the well consistence among central velocities of CO, HCO+, and HCN spectra, suggests that the CO-selected Planck cores are more quiescent compared to classical star-forming regions. The small difference between line widths of C18O and HCN indicates that the inner regions of CO-selected Planck cores are not more turbulent than the exterior. The velocity-integrated intensities and abundances of HCO+ are positively correlated with those of HCN, suggesting these two species are well coupled and chemically connected. The detected abundances of both HCO+ and HCN are significantly lower than values in other low- to high-mass star-forming regions. The low abundances may be due to beam dilution. On the basis of the inspection of the parameters given in the PGCC catalog, we suggest that there may be about 1 000 PGCC objects having sufficient reservoir of dense gas to form stars.

We report the discovery of ASASSN-15lh (SN 2015L), which we interpret as the most luminous supernova yet found. At redshift z = 0.2326, ASASSN-15lh reached an absolute magnitude of M_{u,AB} = -23.5+/-0.1 and bolometric luminosity L_bol = (2.2+/-0.2)x 10^45 ergs s^-1, which is more than twice as luminous as any previously known supernova. It has several major features characteristic of the hydrogen-poor super-luminous supernovae (SLSNe-I), whose energy sources and progenitors are currently poorly understood. In contrast to most previously known SLSNe-I that reside in star-forming dwarf galaxies, ASASSN-15lh appears to be hosted by a luminous galaxy (M_K ~ -25.5) with little star formation. In the 4 months since first detection, ASASSN-15lh radiated (1.1+/- 0.2)x10^52 ergs, challenging the magnetar model for its engine.

Fine-grained Address Space Randomization has been considered as an effective protection against code reuse attacks such as ROP/JOP. However, it only employs a one-time randomization, and such a limitation has been exploited by recent just-in-time ROP and side channel ROP, which collect gadgets on-the-fly and dynamically compile them for malicious purposes. To defeat these advanced code reuse attacks, we propose a new defense principle: instantly obsoleting the address-code associations. We have initialized this principle with a novel technique called virtual space page table remapping and implemented the technique in a system CHAMELEON. CHAMELEON periodically re-randomizes the locations of code pages on-the-fly. A set of techniques are proposed to achieve our goal, including iterative instrumentation that instruments a to-be-protected binary program to generate a re-randomization compatible binary, runtime virtual page shuffling, and function reordering and instruction rearranging optimizations. We have tested CHAMELEON with over a hundred binary programs. Our experiments show that CHAMELEON can defeat all of our tested exploits by both preventing the exploit from gathering sufficient gadgets, and blocking the gadgets execution. Regarding the interval of our re-randomization, it is a parameter and can be set as short as 100ms, 10ms or 1ms. The experiment results show that CHAMELEON introduces on average 11.1%, 12.1% and 12.9% performance overhead for these parameters, respectively.

A study on the NO2 gas sorption ability of amorphous Ge33Se67 coated quartz crystal microbalance (QCM) is presented. The thin films have been characterized before and after sorption/desorption processes of NO2 by energy-dispersive X-ray spectroscopy (EDS), grazing angle X-ray diffraction (GAXRD), Raman spectroscopy, X-ray photoelectron spectroscopy (XPS) and atom force microscopy (AFM) measurements. These studies indicated that physisorption occurs when NO2 gas molecules are introduced into the chalcogenide film and the thin film composition or structure do not change. The mass loading due to NO2 gas sorption was calculated by the resonator's frequency shift. At the conditions of our experiment, up to 6.8 ng of the gas was sorbed into 200nm thick Ge33Se67 film at 5000 ppm NO2 concentration. It has been established that the process of gas molecules sorption is reversible.

In this study we investigated the sorption ability of Ge20Se80 thin films applied as active layers of quartz crystal microbalance (QCM) for NO2 gas sensing. To identify the chalcogenide system appropriate for gas sensing, we provided data for the packing fraction of a number of chalcogenide systems and discussed their suitability. We performed Raman spectroscopy, X-ray photoelectron spectroscopy (XPS) and atom force microscopy (AFM) measurements on the thin films both before and after gas absorption, which showed that the introduced gas molecules interact electrostatically with the chalcogen atoms of the host material and initiate some degree of structural changes in it. The weight change due to NO2 gas absorption was measured by frequency change of the resonator. The absorbed mass increased monotonically with the thickness of chalcogenide films and the NO2 gas concentration. At the conditions of our experiment, up to 11.4 ng of the gas was absorbed into 200nm thick Ge20Se80 film at 5000 ppm NO2 concentration. The process of gas molecules absorption is found irreversible at the purging conditions.

Long term aging is studied on several families of chalcogenide glasses including the Ge-Se, As-Se, Ge-P-Se and Ge-As-Se systems. Special attention is given to the As-Se binary, a system that displays a rich variety of aging behavior intimately tied to sample synthesis conditions and the ambient environment in which samples are aged. Calorimetric (Modulated DSC) and Raman scattering experiments are undertaken. Our results show all samples display a sub-Tg endotherm below Tg in glassy networks possessing a mean coordination number r in the 2.25 < r < 2.45 range. Two sets of AsxSe1-x samples aged for 8 years were compared, set A consisted of slow cooled samples aged in the dark, and set B consisted of melt quenched samples aged at laboratory environment. Samples of set B in the As concentration range, 35% < x < 60%, display a pre-Tg exotherm, but the feature is not observed in samples of set A. The aging behavior of set A presumably represents intrinsic aging in these glasses, while that of set B is extrinsic due to presence of light. The reversibility window persists in both sets of samples but is less well defined in set B. These findings contrast with a recent study by Golovchak et al., which finds the onset of the reversibility window moved up to the stoichiometric composition (x = 40%). Here we show that the upshifted window is better understood as resulting due to demixing of As4Se4 and As4Se3 molecules from the backbone, i.e., Nanoscale phase separation (NSPS). We attribute sub-Tg endotherms to compaction of the flexible part of networks upon long term aging, while the pre-Tg exotherm to NSPS. Finally, the narrowing and sharpening of the reversibility window upon aging is interpreted as the slow 'self-organizing' stress relaxation of the phases just outside the Intermediate phase.