@article{13375, abstract = {Dissipative self-assembly leads to structures and materials that exist away from equilibrium by continuously exchanging energy and materials with the external environment. Although this mode of self-assembly is ubiquitous in nature, where it gives rise to functions such as signal processing, motility, self-healing, self-replication, and ultimately life, examples of dissipative self-assembly processes in man-made systems are few and far between. Herein, recent progress in developing diverse synthetic dissipative self-assembly systems is discussed. The systems reported thus far can be categorized into three classes, in which: i) the fuel chemically modifies the building blocks, thus triggering their self-assembly, ii) the fuel acts as a template interacting with the building blocks noncovalently, and iii) transient states are induced by the addition of two mutually exclusive stimuli. These early studies give rise to materials that would be difficult to obtain otherwise, including hydrogels with programmable lifetimes, vesicular nanoreactors, and membranes exhibiting transient conductivity.}, author = {De, Soumen and Klajn, Rafal}, issn = {1521-4095}, journal = {Advanced Materials}, keywords = {Mechanical Engineering, Mechanics of Materials, General Materials Science}, number = {41}, publisher = {Wiley}, title = {{Dissipative self-assembly driven by the consumption of chemical fuels}}, doi = {10.1002/adma.201706750}, volume = {30}, year = {2018}, } @article{13376, abstract = {Efficient molecular switching in confined spaces is critical for the successful development of artificial molecular machines. However, molecular switching events often entail large structural changes and therefore require conformational freedom, which is typically limited under confinement conditions. Here, we investigated the behavior of azobenzene—the key building block of light-controlled molecular machines—in a confined environment that is flexible and can adapt its shape to that of the bound guest. To this end, we encapsulated several structurally diverse azobenzenes within the cavity of a flexible, water-soluble coordination cage, and investigated their light-responsive behavior. Using UV/Vis absorption spectroscopy and a combination of NMR methods, we showed that each of the encapsulated azobenzenes exhibited distinct switching properties. An azobenzene forming a 1:1 host–guest inclusion complex could be efficiently photoisomerized in a reversible fashion. In contrast, successful switching in inclusion complexes incorporating two azobenzene guests was dependent on the availability of free cages in the system, and it involved reversible trafficking of azobenzene between the cages. In the absence of extra cages, photoswitching was either suppressed or it involved expulsion of azobenzene from the cage and consequently its precipitation from the solution. This finding was utilized to develop an information storage medium in which messages could be written and erased in a reversible fashion using light.}, author = {Samanta, Dipak and Gemen, Julius and Chu, Zonglin and Diskin-Posner, Yael and Shimon, Linda J. W. and Klajn, Rafal}, issn = {1091-6490}, journal = {Proceedings of the National Academy of Sciences}, keywords = {Multidisciplinary}, number = {38}, pages = {9379--9384}, publisher = {Proceedings of the National Academy of Sciences}, title = {{Reversible photoswitching of encapsulated azobenzenes in water}}, doi = {10.1073/pnas.1712787115}, volume = {115}, year = {2018}, } @article{13377, abstract = {Confining organic molecules to the surfaces of inorganic nanoparticles can induce intermolecular interactions between them, which can affect the composition of the mixed self-assembled monolayers obtained by co-adsorption from solution of two different molecules. Two thiolated ligands (a dialkylviologen and a zwitterionic sulfobetaine) that can interact with each other electrostatically were coadsorbed onto gold nanoparticles. The nanoparticles favor a narrow range of ratios of these two molecules that is largely independent of the molar ratio in solution. Changing the solution molar ratio of the two ligands by a factor of 5 000 affects the on-nanoparticle ratio of these ligands by only threefold. This behavior is reminiscent of the formation of insoluble inorganic salts (such as AgCl), which similarly compensate positive and negative charges upon crystallizing. Our results pave the way towards developing well-defined hybrid organic–inorganic nanostructures.}, author = {Chu, Zonglin and Han, Yanxiao and Král, Petr and Klajn, Rafal}, issn = {1521-3773}, journal = {Angewandte Chemie International Edition}, keywords = {General Chemistry, Catalysis}, number = {24}, pages = {7023--7027}, publisher = {Wiley}, title = {{“Precipitation on nanoparticles”: Attractive intermolecular interactions stabilize specific ligand ratios on the surfaces of nanoparticles}}, doi = {10.1002/anie.201800673}, volume = {57}, year = {2018}, } @article{13379, author = {Bléger, David and Klajn, Rafal}, issn = {1521-3927}, journal = {Macromolecular Rapid Communications}, keywords = {Materials Chemistry, Polymers and Plastics, Organic Chemistry}, number = {1}, publisher = {Wiley}, title = {{Integrating macromolecules with molecular switches}}, doi = {10.1002/marc.201700827}, volume = {39}, year = {2018}, } @article{134, abstract = {The current state of the art in real-time two-dimensional water wave simulation requires developers to choose between efficient Fourier-based methods, which lack interactions with moving obstacles, and finite-difference or finite element methods, which handle environmental interactions but are significantly more expensive. This paper attempts to bridge this long-standing gap between complexity and performance, by proposing a new wave simulation method that can faithfully simulate wave interactions with moving obstacles in real time while simultaneously preserving minute details and accommodating very large simulation domains. Previous methods for simulating 2D water waves directly compute the change in height of the water surface, a strategy which imposes limitations based on the CFL condition (fast moving waves require small time steps) and Nyquist's limit (small wave details require closely-spaced simulation variables). This paper proposes a novel wavelet transformation that discretizes the liquid motion in terms of amplitude-like functions that vary over space, frequency, and direction, effectively generalizing Fourier-based methods to handle local interactions. Because these new variables change much more slowly over space than the original water height function, our change of variables drastically reduces the limitations of the CFL condition and Nyquist limit, allowing us to simulate highly detailed water waves at very large visual resolutions. Our discretization is amenable to fast summation and easy to parallelize. We also present basic extensions like pre-computed wave paths and two-way solid fluid coupling. Finally, we argue that our discretization provides a convenient set of variables for artistic manipulation, which we illustrate with a novel wave-painting interface.}, author = {Jeschke, Stefan and Skrivan, Tomas and Mueller Fischer, Matthias and Chentanez, Nuttapong and Macklin, Miles and Wojtan, Christopher J}, journal = {ACM Transactions on Graphics}, number = {4}, publisher = {ACM}, title = {{Water surface wavelets}}, doi = {10.1145/3197517.3201336}, volume = {37}, year = {2018}, } @article{13473, abstract = {Stripped-envelope stars form in binary systems after losing mass through Roche-lobe overflow. They bear astrophysical significance as sources of UV and ionizing radiation in older stellar populations and, if sufficiently massive, as stripped supernova progenitors. Binary evolutionary models predict that they are common, but only a handful of subdwarfs with B-type companions are known. The question is whether a large population of such systems has evaded detection as a result of biases, or whether the model predictions are wrong. We reanalyze the well-studied post-interaction binary φ Persei. Recently, new data have improved the orbital solution of the system, which contains an ~1.2M⊙ stripped-envelope star and a rapidly rotating ~9.6M⊙ Be star. We compare with an extensive grid of evolutionary models using a Bayesian approach and constrain the initial masses of the progenitor to 7.2 ± 0.4M⊙ for the stripped star and 3.8 ± 0.4M⊙ for the Be star. The system must have evolved through near-conservative mass transfer. These findings are consistent with earlier studies. The age we obtain, 57 ± 9 Myr, is in excellent agreement with the age of the α Persei cluster. We note that neither star was initially massive enough to produce a core-collapse supernova, but mass exchange pushed the Be star above the mass threshold. We find that the subdwarf is overluminous for its mass by almost an order of magnitude, compared to the expectations for a helium core burning star. We can only reconcile this if the subdwarf resides in a late phase of helium shell burning, which lasts only 2–3% of the total lifetime as a subdwarf. Assuming continuous star formation implies that up to ~50 less evolved, dimmer subdwarfs exist for each system similar to φ Persei, but have evaded detection so far. Our findings can be interpreted as a strong indication that a substantial population of stripped-envelope stars indeed exists, but has so far evaded detection because of observational biases and lack of large-scale systematic searches.}, author = {Schootemeijer, A. and Götberg, Ylva Louise Linsdotter and de Mink, S. E. and Gies, D. and Zapartas, E.}, issn = {1432-0746}, journal = {Astronomy & Astrophysics}, keywords = {Space and Planetary Science, Astronomy and Astrophysics}, publisher = {EDP Sciences}, title = {{Clues about the scarcity of stripped-envelope stars from the evolutionary state of the sdO+Be binary system φ Persei}}, doi = {10.1051/0004-6361/201731194}, volume = {615}, year = {2018}, } @article{13474, abstract = {Recent surveys of the Magellanic Clouds have revealed a subtype of Wolf–Rayet (WR) star with peculiar properties. WN3/O3 spectra exhibit both WR-like emission and O3 V-like absorption – but at lower luminosity than O3 V or WN stars. We examine the projected spatial distribution of WN3/O3 stars in the Large Magellanic Cloud as compared to O-type stars. Surprisingly, WN3/O3 stars are among the most isolated of all classes of massive stars; they have a distribution similar to red supergiants dominated by initial masses of 10–15 M⊙, and are far more dispersed than classical WR stars or luminous blue variables. Their lack of association with clusters of O-type stars suggests strongly that WN3/O3 stars are not the descendants of single massive stars (30 M⊙ or above). Instead, they are likely products of interacting binaries at lower initial mass (10–18 M⊙). Comparison with binary models suggests a probable origin with primaries in this mass range that were stripped of their H envelopes through non-conservative mass transfer by a low-mass secondary. We show that model spectra and positions on the Hertzsprung–Russell diagram for binary-stripped stars are consistent with WN3/O3 stars. Monitoring radial velocities with high-resolution spectra can test for low-mass companions or runaway velocities. With lower initial mass and environments that avoid very massive stars, the WN3/O3 stars fit expectations for progenitors of Type Ib and possibly Type Ibn supernovae.}, author = {Smith, Nathan and Götberg, Ylva Louise Linsdotter and de Mink, Selma E}, issn = {1365-2966}, journal = {Monthly Notices of the Royal Astronomical Society}, keywords = {Space and Planetary Science, Astronomy and Astrophysics}, number = {1}, pages = {772--782}, publisher = {Oxford University Press}, title = {{Extreme isolation of WN3/O3 stars and implications for their evolutionary origin as the elusive stripped binaries}}, doi = {10.1093/mnras/stx3181}, volume = {475}, year = {2018}, } @article{13475, abstract = {Stars stripped of their hydrogen-rich envelope through interaction with a binary companion are generally not considered when accounting for ionizing radiation from stellar populations, despite the expectation that stripped stars emit hard ionizing radiation, form frequently, and live 10–100 times longer than single massive stars. We compute the first grid of evolutionary and spectral models specially made for stars stripped in binaries for a range of progenitor masses (2–20 M⊙) and metallicities ranging from solar to values representative for pop II stars. For stripped stars with masses in the range 0.3–7 M⊙, we find consistently high effective temperatures (20 000–100 000 K, increasing with mass), small radii (0.2–1 R⊙), and high bolometric luminosities, comparable to that of their progenitor before stripping. The spectra show a continuous sequence that naturally bridges subdwarf-type stars at the low-mass end and Wolf-Rayet-like spectra at the high-mass end. For intermediate masses we find hybrid spectral classes showing a mixture of absorption and emission lines. These appear for stars with mass-loss rates of 10−8−10−6 M⊙ yr−1, which have semi-transparent atmospheres. At low metallicity, substantial hydrogen-rich layers are left at the surface and we predict spectra that resemble O-type stars instead. We obtain spectra undistinguishable from subdwarfs for stripped stars with masses up to 1.7 M⊙, which questions whether the widely adopted canonical value of 0.47 M⊙ is uniformly valid. Only a handful of stripped stars of intermediate mass have currently been identified observationally. Increasing this sample will provide necessary tests for the physics of interaction, internal mixing, and stellar winds. We use our model spectra to investigate the feasibility to detect stripped stars next to an optically bright companion and recommend systematic searches for their UV excess and possible emission lines, most notably HeII λ4686 in the optical and HeII λ1640 in the UV. Our models are publicly available for further investigations or inclusion in spectral synthesis simulations.}, author = {Götberg, Ylva Louise Linsdotter and de Mink, S. E. and Groh, J. H. and Kupfer, T. and Crowther, P. A. and Zapartas, E. and Renzo, M.}, issn = {1432-0746}, journal = {Astronomy & Astrophysics}, keywords = {Space and Planetary Science, Astronomy and Astrophysics}, publisher = {EDP Sciences}, title = {{Spectral models for binary products: Unifying subdwarfs and Wolf-Rayet stars as a sequence of stripped-envelope stars}}, doi = {10.1051/0004-6361/201732274}, volume = {615}, year = {2018}, } @article{135, abstract = {The Fluid Implicit Particle method (FLIP) reduces numerical dissipation by combining particles with grids. To improve performance, the subsequent narrow band FLIP method (NB‐FLIP) uses a FLIP‐based fluid simulation only near the liquid surface and a traditional grid‐based fluid simulation away from the surface. This spatially‐limited FLIP simulation significantly reduces the number of particles and alleviates a computational bottleneck. In this paper, we extend the NB‐FLIP idea even further, by allowing a simulation to transition between a FLIP‐like fluid simulation and a grid‐based simulation in arbitrary locations, not just near the surface. This approach leads to even more savings in memory and computation, because we can concentrate the particles only in areas where they are needed. More importantly, this new method allows us to seamlessly transition to smooth implicit surface geometry wherever the particle‐based simulation is unnecessary. Consequently, our method leads to a practical algorithm for avoiding the noisy surface artifacts associated with particle‐based liquid simulations, while simultaneously maintaining the benefits of a FLIP simulation in regions of dynamic motion.}, author = {Sato, Takahiro and Wojtan, Christopher J and Thuerey, Nils and Igarashi, Takeo and Ando, Ryoichi}, issn = {0167-7055}, journal = {Computer Graphics Forum}, number = {2}, pages = {169 -- 177}, publisher = {Wiley}, title = {{Extended narrow band FLIP for liquid simulations}}, doi = {10.1111/cgf.13351}, volume = {37}, year = {2018}, } @article{136, abstract = {Recent studies suggest that unstable, nonchaotic solutions of the Navier-Stokes equation may provide deep insights into fluid turbulence. In this article, we present a combined experimental and numerical study exploring the dynamical role of unstable equilibrium solutions and their invariant manifolds in a weakly turbulent, electromagnetically driven, shallow fluid layer. Identifying instants when turbulent evolution slows down, we compute 31 unstable equilibria of a realistic two-dimensional model of the flow. We establish the dynamical relevance of these unstable equilibria by showing that they are closely visited by the turbulent flow. We also establish the dynamical relevance of unstable manifolds by verifying that they are shadowed by turbulent trajectories departing from the neighborhoods of unstable equilibria over large distances in state space.}, author = {Suri, Balachandra and Tithof, Jeffrey and Grigoriev, Roman and Schatz, Michael}, journal = {Physical Review E}, number = {2}, publisher = {American Physical Society}, title = {{Unstable equilibria and invariant manifolds in quasi-two-dimensional Kolmogorov-like flow}}, doi = {10.1103/PhysRevE.98.023105}, volume = {98}, year = {2018}, } @article{137, abstract = {Fluorescent sensors are an essential part of the experimental toolbox of the life sciences, where they are used ubiquitously to visualize intra- and extracellular signaling. In the brain, optical neurotransmitter sensors can shed light on temporal and spatial aspects of signal transmission by directly observing, for instance, neurotransmitter release and spread. Here we report the development and application of the first optical sensor for the amino acid glycine, which is both an inhibitory neurotransmitter and a co-agonist of the N-methyl-d-aspartate receptors (NMDARs) involved in synaptic plasticity. Computational design of a glycine-specific binding protein allowed us to produce the optical glycine FRET sensor (GlyFS), which can be used with single and two-photon excitation fluorescence microscopy. We took advantage of this newly developed sensor to test predictions about the uneven spatial distribution of glycine in extracellular space and to demonstrate that extracellular glycine levels are controlled by plasticity-inducing stimuli.}, author = {Zhang, William and Herde, Michel and Mitchell, Joshua and Whitfield, Jason and Wulff, Andreas and Vongsouthi, Vanessa and Sanchez Romero, Inmaculada and Gulakova, Polina and Minge, Daniel and Breithausen, Björn and Schoch, Susanne and Janovjak, Harald L and Jackson, Colin and Henneberger, Christian}, journal = {Nature Chemical Biology}, number = {9}, pages = {861 -- 869}, publisher = {Nature Publishing Group}, title = {{Monitoring hippocampal glycine with the computationally designed optical sensor GlyFS}}, doi = {10.1038/s41589-018-0108-2}, volume = {14}, year = {2018}, } @article{139, abstract = {Genome-scale diversity data are increasingly available in a variety of biological systems, and can be used to reconstruct the past evolutionary history of species divergence. However, extracting the full demographic information from these data is not trivial, and requires inferential methods that account for the diversity of coalescent histories throughout the genome. Here, we evaluate the potential and limitations of one such approach. We reexamine a well-known system of mussel sister species, using the joint site frequency spectrum (jSFS) of synonymousmutations computed either fromexome capture or RNA-seq, in an Approximate Bayesian Computation (ABC) framework. We first assess the best sampling strategy (number of: individuals, loci, and bins in the jSFS), and show that model selection is robust to variation in the number of individuals and loci. In contrast, different binning choices when summarizing the jSFS, strongly affect the results: including classes of low and high frequency shared polymorphisms can more effectively reveal recent migration events. We then take advantage of the flexibility of ABC to compare more realistic models of speciation, including variation in migration rates through time (i.e., periodic connectivity) and across genes (i.e., genome-wide heterogeneity in migration rates). We show that these models were consistently selected as the most probable, suggesting that mussels have experienced a complex history of gene flow during divergence and that the species boundary is semi-permeable. Our work provides a comprehensive evaluation of ABC demographic inference in mussels based on the coding jSFS, and supplies guidelines for employing different sequencing techniques and sampling strategies. We emphasize, perhaps surprisingly, that inferences are less limited by the volume of data, than by the way in which they are analyzed.}, author = {Fraisse, Christelle and Roux, Camille and Gagnaire, Pierre and Romiguier, Jonathan and Faivre, Nicolas and Welch, John and Bierne, Nicolas}, journal = {PeerJ}, number = {7}, publisher = {PeerJ}, title = {{The divergence history of European blue mussel species reconstructed from Approximate Bayesian Computation: The effects of sequencing techniques and sampling strategies}}, doi = {10.7717/peerj.5198}, volume = {2018}, year = {2018}, } @article{14, abstract = {The intercellular transport of auxin is driven by PIN-formed (PIN) auxin efflux carriers. PINs are localized at the plasma membrane (PM) and on constitutively recycling endomembrane vesicles. Therefore, PINs can mediate auxin transport either by direct translocation across the PM or by pumping auxin into secretory vesicles (SVs), leading to its secretory release upon fusion with the PM. Which of these two mechanisms dominates is a matter of debate. Here, we addressed the issue with a mathematical modeling approach. We demonstrate that the efficiency of secretory transport depends on SV size, half-life of PINs on the PM, pH, exocytosis frequency and PIN density. 3D structured illumination microscopy (SIM) was used to determine PIN density on the PM. Combining this data with published values of the other parameters, we show that the transport activity of PINs in SVs would have to be at least 1000× greater than on the PM in order to produce a comparable macroscopic auxin transport. If both transport mechanisms operated simultaneously and PINs were equally active on SVs and PM, the contribution of secretion to the total auxin flux would be negligible. In conclusion, while secretory vesicle-mediated transport of auxin is an intriguing and theoretically possible model, it is unlikely to be a major mechanism of auxin transport inplanta.}, author = {Hille, Sander and Akhmanova, Maria and Glanc, Matous and Johnson, Alexander J and Friml, Jirí}, issn = {1422-0067}, journal = {International Journal of Molecular Sciences}, number = {11}, publisher = {MDPI}, title = {{Relative contribution of PIN-containing secretory vesicles and plasma membrane PINs to the directed auxin transport: Theoretical estimation}}, doi = {10.3390/ijms19113566}, volume = {19}, year = {2018}, } @inproceedings{140, abstract = {Reachability analysis is difficult for hybrid automata with affine differential equations, because the reach set needs to be approximated. Promising abstraction techniques usually employ interval methods or template polyhedra. Interval methods account for dense time and guarantee soundness, and there are interval-based tools that overapproximate affine flowpipes. But interval methods impose bounded and rigid shapes, which make refinement expensive and fixpoint detection difficult. Template polyhedra, on the other hand, can be adapted flexibly and can be unbounded, but sound template refinement for unbounded reachability analysis has been implemented only for systems with piecewise constant dynamics. We capitalize on the advantages of both techniques, combining interval arithmetic and template polyhedra, using the former to abstract time and the latter to abstract space. During a CEGAR loop, whenever a spurious error trajectory is found, we compute additional space constraints and split time intervals, and use these space-time interpolants to eliminate the counterexample. Space-time interpolation offers a lazy, flexible framework for increasing precision while guaranteeing soundness, both for error avoidance and fixpoint detection. To the best of out knowledge, this is the first abstraction refinement scheme for the reachability analysis over unbounded and dense time of affine hybrid systems, which is both sound and automatic. We demonstrate the effectiveness of our algorithm with several benchmark examples, which cannot be handled by other tools.}, author = {Frehse, Goran and Giacobbe, Mirco and Henzinger, Thomas A}, issn = {03029743}, location = {Oxford, United Kingdom}, pages = {468 -- 486}, publisher = {Springer}, title = {{Space-time interpolants}}, doi = {10.1007/978-3-319-96145-3_25}, volume = {10981}, year = {2018}, } @article{14003, abstract = {Molecular chirality plays an essential role in most biochemical processes. The observation and quantification of chirality-sensitive signals, however, remains extremely challenging, especially on ultrafast timescales and in dilute media. Here, we describe the experimental realization of an all-optical and ultrafast scheme for detecting chiral dynamics in molecules. This technique is based on high-harmonic generation by a combination of two-color counterrotating femtosecond laser pulses with polarization states tunable from linear to circular. We demonstrate two different implementations of chiral-sensitive high-harmonic spectroscopy on an ensemble of randomly oriented methyloxirane molecules in the gas phase. Using two elliptically polarized fields, we observe that the ellipticities maximizing the harmonic signal reach up to 4.4 ± 0.2 % (at 17.6 eV). Using two circularly polarized fields, we observe circular dichroisms ranging up to 13 ± 6 % (28.3–33.1 eV). Our theoretical analysis confirms that the observed chiral response originates from subfemtosecond electron dynamics driven by the magnetic component of the driving laser field. This assignment is supported by the experimental observation of a strong intensity dependence of the chiral effects and its agreement with theory. We moreover report and explain a pronounced variation of the signal strength and dichroism with the driving-field ellipticities and harmonic orders. Finally, we demonstrate the sensitivity of the experimental observables to the shape of the electron hole. This technique for chiral discrimination will yield femtosecond temporal resolution when integrated in a pump-probe scheme and subfemtosecond resolution on chiral charge migration in a self-probing scheme.}, author = {Baykusheva, Denitsa Rangelova and Wörner, Hans Jakob}, issn = {2160-3308}, journal = {Physical Review X}, keywords = {General Physics and Astronomy}, number = {3}, publisher = {American Physical Society}, title = {{Chiral discrimination through bielliptical high-harmonic spectroscopy}}, doi = {10.1103/physrevx.8.031060}, volume = {8}, year = {2018}, } @inproceedings{141, abstract = {Given a model and a specification, the fundamental model-checking problem asks for algorithmic verification of whether the model satisfies the specification. We consider graphs and Markov decision processes (MDPs), which are fundamental models for reactive systems. One of the very basic specifications that arise in verification of reactive systems is the strong fairness (aka Streett) objective. Given different types of requests and corresponding grants, the objective requires that for each type, if the request event happens infinitely often, then the corresponding grant event must also happen infinitely often. All ω -regular objectives can be expressed as Streett objectives and hence they are canonical in verification. To handle the state-space explosion, symbolic algorithms are required that operate on a succinct implicit representation of the system rather than explicitly accessing the system. While explicit algorithms for graphs and MDPs with Streett objectives have been widely studied, there has been no improvement of the basic symbolic algorithms. The worst-case numbers of symbolic steps required for the basic symbolic algorithms are as follows: quadratic for graphs and cubic for MDPs. In this work we present the first sub-quadratic symbolic algorithm for graphs with Streett objectives, and our algorithm is sub-quadratic even for MDPs. Based on our algorithmic insights we present an implementation of the new symbolic approach and show that it improves the existing approach on several academic benchmark examples.}, author = {Chatterjee, Krishnendu and Henzinger, Monika H and Loitzenbauer, Veronika and Oraee, Simin and Toman, Viktor}, location = {Oxford, United Kingdom}, pages = {178--197}, publisher = {Springer}, title = {{Symbolic algorithms for graphs and Markov decision processes with fairness objectives}}, doi = {10.1007/978-3-319-96142-2_13}, volume = {10982}, year = {2018}, } @inproceedings{14198, abstract = {High-dimensional time series are common in many domains. Since human cognition is not optimized to work well in high-dimensional spaces, these areas could benefit from interpretable low-dimensional representations. However, most representation learning algorithms for time series data are difficult to interpret. This is due to non-intuitive mappings from data features to salient properties of the representation and non-smoothness over time. To address this problem, we propose a new representation learning framework building on ideas from interpretable discrete dimensionality reduction and deep generative modeling. This framework allows us to learn discrete representations of time series, which give rise to smooth and interpretable embeddings with superior clustering performance. We introduce a new way to overcome the non-differentiability in discrete representation learning and present a gradient-based version of the traditional self-organizing map algorithm that is more performant than the original. Furthermore, to allow for a probabilistic interpretation of our method, we integrate a Markov model in the representation space. This model uncovers the temporal transition structure, improves clustering performance even further and provides additional explanatory insights as well as a natural representation of uncertainty. We evaluate our model in terms of clustering performance and interpretability on static (Fashion-)MNIST data, a time series of linearly interpolated (Fashion-)MNIST images, a chaotic Lorenz attractor system with two macro states, as well as on a challenging real world medical time series application on the eICU data set. Our learned representations compare favorably with competitor methods and facilitate downstream tasks on the real world data.}, author = {Fortuin, Vincent and Hüser, Matthias and Locatello, Francesco and Strathmann, Heiko and Rätsch, Gunnar}, booktitle = {International Conference on Learning Representations}, location = {New Orleans, LA, United States}, title = {{SOM-VAE: Interpretable discrete representation learning on time series}}, year = {2018}, } @inproceedings{142, abstract = {We address the problem of analyzing the reachable set of a polynomial nonlinear continuous system by over-approximating the flowpipe of its dynamics. The common approach to tackle this problem is to perform a numerical integration over a given time horizon based on Taylor expansion and interval arithmetic. However, this method results to be very conservative when there is a large difference in speed between trajectories as time progresses. In this paper, we propose to use combinations of barrier functions, which we call piecewise barrier tube (PBT), to over-approximate flowpipe. The basic idea of PBT is that for each segment of a flowpipe, a coarse box which is big enough to contain the segment is constructed using sampled simulation and then in the box we compute by linear programming a set of barrier functions (called barrier tube or BT for short) which work together to form a tube surrounding the flowpipe. The benefit of using PBT is that (1) BT is independent of time and hence can avoid being stretched and deformed by time; and (2) a small number of BTs can form a tight over-approximation for the flowpipe, which means that the computation required to decide whether the BTs intersect the unsafe set can be reduced significantly. We implemented a prototype called PBTS in C++. Experiments on some benchmark systems show that our approach is effective.}, author = {Kong, Hui and Bartocci, Ezio and Henzinger, Thomas A}, location = {Oxford, United Kingdom}, pages = {449 -- 467}, publisher = {Springer}, title = {{Reachable set over-approximation for nonlinear systems using piecewise barrier tubes}}, doi = {10.1007/978-3-319-96145-3_24}, volume = {10981}, year = {2018}, } @inproceedings{14201, abstract = {Variational inference is a popular technique to approximate a possibly intractable Bayesian posterior with a more tractable one. Recently, boosting variational inference has been proposed as a new paradigm to approximate the posterior by a mixture of densities by greedily adding components to the mixture. However, as is the case with many other variational inference algorithms, its theoretical properties have not been studied. In the present work, we study the convergence properties of this approach from a modern optimization viewpoint by establishing connections to the classic Frank-Wolfe algorithm. Our analyses yields novel theoretical insights regarding the sufficient conditions for convergence, explicit rates, and algorithmic simplifications. Since a lot of focus in previous works for variational inference has been on tractability, our work is especially important as a much needed attempt to bridge the gap between probabilistic models and their corresponding theoretical properties.}, author = {Locatello, Francesco and Khanna, Rajiv and Ghosh, Joydeep and Rätsch, Gunnar}, booktitle = {Proceedings of the 21st International Conference on Artificial Intelligence and Statistics}, location = {Playa Blanca, Lanzarote}, pages = {464--472}, publisher = {ML Research Press}, title = {{Boosting variational inference: An optimization perspective}}, volume = {84}, year = {2018}, } @inproceedings{14202, abstract = {Approximating a probability density in a tractable manner is a central task in Bayesian statistics. Variational Inference (VI) is a popular technique that achieves tractability by choosing a relatively simple variational family. Borrowing ideas from the classic boosting framework, recent approaches attempt to \emph{boost} VI by replacing the selection of a single density with a greedily constructed mixture of densities. In order to guarantee convergence, previous works impose stringent assumptions that require significant effort for practitioners. Specifically, they require a custom implementation of the greedy step (called the LMO) for every probabilistic model with respect to an unnatural variational family of truncated distributions. Our work fixes these issues with novel theoretical and algorithmic insights. On the theoretical side, we show that boosting VI satisfies a relaxed smoothness assumption which is sufficient for the convergence of the functional Frank-Wolfe (FW) algorithm. Furthermore, we rephrase the LMO problem and propose to maximize the Residual ELBO (RELBO) which replaces the standard ELBO optimization in VI. These theoretical enhancements allow for black box implementation of the boosting subroutine. Finally, we present a stopping criterion drawn from the duality gap in the classic FW analyses and exhaustive experiments to illustrate the usefulness of our theoretical and algorithmic contributions.}, author = {Locatello, Francesco and Dresdner, Gideon and Khanna, Rajiv and Valera, Isabel and Rätsch, Gunnar}, booktitle = {Advances in Neural Information Processing Systems}, isbn = {9781510884472}, issn = {1049-5258}, location = {Montreal, Canada}, publisher = {Neural Information Processing Systems Foundation}, title = {{Boosting black box variational inference}}, volume = {31}, year = {2018}, }