@inproceedings{8704, abstract = {Traditional robotic control suits require profound task-specific knowledge for designing, building and testing control software. The rise of Deep Learning has enabled end-to-end solutions to be learned entirely from data, requiring minimal knowledge about the application area. We design a learning scheme to train end-to-end linear dynamical systems (LDS)s by gradient descent in imitation learning robotic domains. We introduce a new regularization loss component together with a learning algorithm that improves the stability of the learned autonomous system, by forcing the eigenvalues of the internal state updates of an LDS to be negative reals. We evaluate our approach on a series of real-life and simulated robotic experiments, in comparison to linear and nonlinear Recurrent Neural Network (RNN) architectures. Our results show that our stabilizing method significantly improves test performance of LDS, enabling such linear models to match the performance of contemporary nonlinear RNN architectures. A video of the obstacle avoidance performance of our method on a mobile robot, in unseen environments, compared to other methods can be viewed at https://youtu.be/mhEsCoNao5E.}, author = {Lechner, Mathias and Hasani, Ramin and Rus, Daniela and Grosu, Radu}, booktitle = {Proceedings - IEEE International Conference on Robotics and Automation}, isbn = {9781728173955}, issn = {10504729}, location = {Paris, France}, pages = {5446--5452}, publisher = {IEEE}, title = {{Gershgorin loss stabilizes the recurrent neural network compartment of an end-to-end robot learning scheme}}, doi = {10.1109/ICRA40945.2020.9196608}, year = {2020}, } @unpublished{14096, abstract = {A binary neutron star merger has been observed in a multi-messenger detection of gravitational wave (GW) and electromagnetic (EM) radiation. Binary neutron stars that merge within a Hubble time, as well as many other compact binaries, are expected to form via common envelope evolution. Yet five decades of research on common envelope evolution have not yet resulted in a satisfactory understanding of the multi-spatial multi-timescale evolution for the systems that lead to compact binaries. In this paper, we report on the first successful simulations of common envelope ejection leading to binary neutron star formation in 3D hydrodynamics. We simulate the dynamical inspiral phase of the interaction between a 12M⊙ red supergiant and a 1.4M⊙ neutron star for different initial separations and initial conditions. For all of our simulations, we find complete envelope ejection and final orbital separations of af≈1.3-5.1R⊙ depending on the simulation and criterion, leading to binary neutron stars that can merge within a Hubble time. We find αCE-equivalent efficiencies of ≈0.1-2.7 depending on the simulation and criterion, but this may be specific for these extended progenitors. We fully resolve the core of the star to ≲0.005R⊙ and our 3D hydrodynamics simulations are informed by an adjusted 1D analytic energy formalism and a 2D kinematics study in order to overcome the prohibitive computational cost of simulating these systems. The framework we develop in this paper can be used to simulate a wide variety of interactions between stars, from stellar mergers to common envelope episodes leading to GW sources.}, author = {Jamie A. P. Law-Smith, Jamie A. P. Law-Smith and Everson, Rosa Wallace and Enrico Ramirez-Ruiz, Enrico Ramirez-Ruiz and Mink, Selma E. de and Son, Lieke A. C. van and Götberg, Ylva Louise Linsdotter and Zellmann, Stefan and Alejandro Vigna-Gómez, Alejandro Vigna-Gómez and Renzo, Mathieu and Wu, Samantha and Schrøder, Sophie L. and Foley, Ryan J. and Tenley Hutchinson-Smith, Tenley Hutchinson-Smith}, booktitle = {arXiv}, title = {{Successful common envelope ejection and binary neutron star formation in 3D hydrodynamics}}, doi = {10.48550/arXiv.2011.06630}, year = {2020}, } @article{8699, abstract = {In the high spin–orbit-coupled Sr2IrO4, the high sensitivity of the ground state to the details of the local lattice structure shows a large potential for the manipulation of the functional properties by inducing local lattice distortions. We use epitaxial strain to modify the Ir–O bond geometry in Sr2IrO4 and perform momentum-dependent resonant inelastic X-ray scattering (RIXS) at the metal and at the ligand sites to unveil the response of the low-energy elementary excitations. We observe that the pseudospin-wave dispersion for tensile-strained Sr2IrO4 films displays large softening along the [h,0] direction, while along the [h,h] direction it shows hardening. This evolution reveals a renormalization of the magnetic interactions caused by a strain-driven cross-over from anisotropic to isotropic interactions between the magnetic moments. Moreover, we detect dispersive electron–hole pair excitations which shift to lower (higher) energies upon compressive (tensile) strain, manifesting a reduction (increase) in the size of the charge gap. This behavior shows an intimate coupling between charge excitations and lattice distortions in Sr2IrO4, originating from the modified hopping elements between the t2g orbitals. Our work highlights the central role played by the lattice degrees of freedom in determining both the pseudospin and charge excitations of Sr2IrO4 and provides valuable information toward the control of the ground state of complex oxides in the presence of high spin–orbit coupling.}, author = {Paris, Eugenio and Tseng, Yi and Paerschke, Ekaterina and Zhang, Wenliang and Upton, Mary H and Efimenko, Anna and Rolfs, Katharina and McNally, Daniel E and Maurel, Laura and Naamneh, Muntaser and Caputo, Marco and Strocov, Vladimir N and Wang, Zhiming and Casa, Diego and Schneider, Christof W and Pomjakushina, Ekaterina and Wohlfeld, Krzysztof and Radovic, Milan and Schmitt, Thorsten}, issn = {10916490}, journal = {Proceedings of the National Academy of Sciences of the United States of America}, number = {40}, pages = {24764--24770}, publisher = {National Academy of Sciences}, title = {{Strain engineering of the charge and spin-orbital interactions in Sr2IrO4}}, doi = {10.1073/pnas.2012043117}, volume = {117}, year = {2020}, } @article{8737, abstract = {Mitochondrial complex I couples NADH:ubiquinone oxidoreduction to proton pumping by an unknown mechanism. Here, we present cryo-electron microscopy structures of ovine complex I in five different conditions, including turnover, at resolutions up to 2.3 to 2.5 angstroms. Resolved water molecules allowed us to experimentally define the proton translocation pathways. Quinone binds at three positions along the quinone cavity, as does the inhibitor rotenone that also binds within subunit ND4. Dramatic conformational changes around the quinone cavity couple the redox reaction to proton translocation during open-to-closed state transitions of the enzyme. In the induced deactive state, the open conformation is arrested by the ND6 subunit. We propose a detailed molecular coupling mechanism of complex I, which is an unexpected combination of conformational changes and electrostatic interactions.}, author = {Kampjut, Domen and Sazanov, Leonid A}, issn = {10959203}, journal = {Science}, number = {6516}, publisher = {American Association for the Advancement of Science}, title = {{The coupling mechanism of mammalian respiratory complex I}}, doi = {10.1126/science.abc4209}, volume = {370}, year = {2020}, } @inproceedings{8722, abstract = {Load imbalance pervasively exists in distributed deep learning training systems, either caused by the inherent imbalance in learned tasks or by the system itself. Traditional synchronous Stochastic Gradient Descent (SGD) achieves good accuracy for a wide variety of tasks, but relies on global synchronization to accumulate the gradients at every training step. In this paper, we propose eager-SGD, which relaxes the global synchronization for decentralized accumulation. To implement eager-SGD, we propose to use two partial collectives: solo and majority. With solo allreduce, the faster processes contribute their gradients eagerly without waiting for the slower processes, whereas with majority allreduce, at least half of the participants must contribute gradients before continuing, all without using a central parameter server. We theoretically prove the convergence of the algorithms and describe the partial collectives in detail. Experimental results on load-imbalanced environments (CIFAR-10, ImageNet, and UCF101 datasets) show that eager-SGD achieves 1.27x speedup over the state-of-the-art synchronous SGD, without losing accuracy.}, author = {Li, Shigang and Tal Ben-Nun, Tal Ben-Nun and Girolamo, Salvatore Di and Alistarh, Dan-Adrian and Hoefler, Torsten}, booktitle = {Proceedings of the 25th ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming}, location = {San Diego, CA, United States}, pages = {45--61}, publisher = {Association for Computing Machinery}, title = {{Taming unbalanced training workloads in deep learning with partial collective operations}}, doi = {10.1145/3332466.3374528}, year = {2020}, } @article{8744, abstract = {Understanding the conformational sampling of translation-arrested ribosome nascent chain complexes is key to understand co-translational folding. Up to now, coupling of cysteine oxidation, disulfide bond formation and structure formation in nascent chains has remained elusive. Here, we investigate the eye-lens protein γB-crystallin in the ribosomal exit tunnel. Using mass spectrometry, theoretical simulations, dynamic nuclear polarization-enhanced solid-state nuclear magnetic resonance and cryo-electron microscopy, we show that thiol groups of cysteine residues undergo S-glutathionylation and S-nitrosylation and form non-native disulfide bonds. Thus, covalent modification chemistry occurs already prior to nascent chain release as the ribosome exit tunnel provides sufficient space even for disulfide bond formation which can guide protein folding.}, author = {Schulte, Linda and Mao, Jiafei and Reitz, Julian and Sreeramulu, Sridhar and Kudlinzki, Denis and Hodirnau, Victor-Valentin and Meier-Credo, Jakob and Saxena, Krishna and Buhr, Florian and Langer, Julian D. and Blackledge, Martin and Frangakis, Achilleas S. and Glaubitz, Clemens and Schwalbe, Harald}, issn = {2041-1723}, journal = {Nature Communications}, keywords = {General Biochemistry, Genetics and Molecular Biology, General Physics and Astronomy, General Chemistry}, publisher = {Springer Nature}, title = {{Cysteine oxidation and disulfide formation in the ribosomal exit tunnel}}, doi = {10.1038/s41467-020-19372-x}, volume = {11}, year = {2020}, } @unpublished{14095, abstract = {The Habitable Exoplanet Observatory, or HabEx, has been designed to be the Great Observatory of the 2030s. For the first time in human history, technologies have matured sufficiently to enable an affordable space-based telescope mission capable of discovering and characterizing Earthlike planets orbiting nearby bright sunlike stars in order to search for signs of habitability and biosignatures. Such a mission can also be equipped with instrumentation that will enable broad and exciting general astrophysics and planetary science not possible from current or planned facilities. HabEx is a space telescope with unique imaging and multi-object spectroscopic capabilities at wavelengths ranging from ultraviolet (UV) to near-IR. These capabilities allow for a broad suite of compelling science that cuts across the entire NASA astrophysics portfolio. HabEx has three primary science goals: (1) Seek out nearby worlds and explore their habitability; (2) Map out nearby planetary systems and understand the diversity of the worlds they contain; (3) Enable new explorations of astrophysical systems from our own solar system to external galaxies by extending our reach in the UV through near-IR. This Great Observatory science will be selected through a competed GO program, and will account for about 50% of the HabEx primary mission. The preferred HabEx architecture is a 4m, monolithic, off-axis telescope that is diffraction-limited at 0.4 microns and is in an L2 orbit. HabEx employs two starlight suppression systems: a coronagraph and a starshade, each with their own dedicated instrument.}, author = {Gaudi, B. Scott and Seager, Sara and Mennesson, Bertrand and Kiessling, Alina and Warfield, Keith and Cahoy, Kerri and Clarke, John T. and Shawn Domagal-Goldman, Shawn Domagal-Goldman and Feinberg, Lee and Guyon, Olivier and Kasdin, Jeremy and Mawet, Dimitri and Plavchan, Peter and Robinson, Tyler and Rogers, Leslie and Scowen, Paul and Somerville, Rachel and Stapelfeldt, Karl and Stark, Christopher and Stern, Daniel and Turnbull, Margaret and Amini, Rashied and Kuan, Gary and Martin, Stefan and Morgan, Rhonda and Redding, David and Stahl, H. Philip and Webb, Ryan and Oscar Alvarez-Salazar, Oscar Alvarez-Salazar and Arnold, William L. and Arya, Manan and Balasubramanian, Bala and Baysinger, Mike and Bell, Ray and Below, Chris and Benson, Jonathan and Blais, Lindsey and Booth, Jeff and Bourgeois, Robert and Bradford, Case and Brewer, Alden and Brooks, Thomas and Cady, Eric and Caldwell, Mary and Calvet, Rob and Carr, Steven and Chan, Derek and Cormarkovic, Velibor and Coste, Keith and Cox, Charlie and Danner, Rolf and Davis, Jacqueline and Dewell, Larry and Dorsett, Lisa and Dunn, Daniel and East, Matthew and Effinger, Michael and Eng, Ron and Freebury, Greg and Garcia, Jay and Gaskin, Jonathan and Greene, Suzan and Hennessy, John and Hilgemann, Evan and Hood, Brad and Holota, Wolfgang and Howe, Scott and Huang, Pei and Hull, Tony and Hunt, Ron and Hurd, Kevin and Johnson, Sandra and Kissil, Andrew and Knight, Brent and Kolenz, Daniel and Kraus, Oliver and Krist, John and Li, Mary and Lisman, Doug and Mandic, Milan and Mann, John and Marchen, Luis and Colleen Marrese-Reading, Colleen Marrese-Reading and McCready, Jonathan and McGown, Jim and Missun, Jessica and Miyaguchi, Andrew and Moore, Bradley and Nemati, Bijan and Nikzad, Shouleh and Nissen, Joel and Novicki, Megan and Perrine, Todd and Pineda, Claudia and Polanco, Otto and Putnam, Dustin and Qureshi, Atif and Richards, Michael and Riggs, A. J. Eldorado and Rodgers, Michael and Rud, Mike and Saini, Navtej and Scalisi, Dan and Scharf, Dan and Schulz, Kevin and Serabyn, Gene and Sigrist, Norbert and Sikkia, Glory and Singleton, Andrew and Shaklan, Stuart and Smith, Scott and Southerd, Bart and Stahl, Mark and Steeves, John and Sturges, Brian and Sullivan, Chris and Tang, Hao and Taras, Neil and Tesch, Jonathan and Therrell, Melissa and Tseng, Howard and Valente, Marty and Buren, David Van and Villalvazo, Juan and Warwick, Steve and Webb, David and Westerhoff, Thomas and Wofford, Rush and Wu, Gordon and Woo, Jahning and Wood, Milana and Ziemer, John and Arney, Giada and Anderson, Jay and Jesús Maíz-Apellániz, Jesús Maíz-Apellániz and Bartlett, James and Belikov, Ruslan and Bendek, Eduardo and Cenko, Brad and Douglas, Ewan and Dulz, Shannon and Evans, Chris and Faramaz, Virginie and Feng, Y. Katherina and Ferguson, Harry and Follette, Kate and Ford, Saavik and García, Miriam and Geha, Marla and Gelino, Dawn and Götberg, Ylva Louise Linsdotter and Hildebrandt, Sergi and Hu, Renyu and Jahnke, Knud and Kennedy, Grant and Kreidberg, Laura and Isella, Andrea and Lopez, Eric and Marchis, Franck and Macri, Lucas and Marley, Mark and Matzko, William and Mazoyer, Johan and McCandliss, Stephan and Meshkat, Tiffany and Mordasini, Christoph and Morris, Patrick and Nielsen, Eric and Newman, Patrick and Petigura, Erik and Postman, Marc and Reines, Amy and Roberge, Aki and Roederer, Ian and Ruane, Garreth and Schwieterman, Edouard and Sirbu, Dan and Spalding, Christopher and Teplitz, Harry and Tumlinson, Jason and Turner, Neal and Werk, Jessica and Wofford, Aida and Wyatt, Mark and Young, Amber and Zellem, Rob}, booktitle = {arXiv}, title = {{The habitable exoplanet observatory (HabEx) mission concept study final report}}, doi = {10.48550/arXiv.2001.06683}, year = {2020}, } @article{8767, abstract = {Resources are rarely distributed uniformly within a population. Heterogeneity in the concentration of a drug, the quality of breeding sites, or wealth can all affect evolutionary dynamics. In this study, we represent a collection of properties affecting the fitness at a given location using a color. A green node is rich in resources while a red node is poorer. More colors can represent a broader spectrum of resource qualities. For a population evolving according to the birth-death Moran model, the first question we address is which structures, identified by graph connectivity and graph coloring, are evolutionarily equivalent. We prove that all properly two-colored, undirected, regular graphs are evolutionarily equivalent (where “properly colored” means that no two neighbors have the same color). We then compare the effects of background heterogeneity on properly two-colored graphs to those with alternative schemes in which the colors are permuted. Finally, we discuss dynamic coloring as a model for spatiotemporal resource fluctuations, and we illustrate that random dynamic colorings often diminish the effects of background heterogeneity relative to a proper two-coloring.}, author = {Kaveh, Kamran and McAvoy, Alex and Chatterjee, Krishnendu and Nowak, Martin A.}, issn = {1553-7358}, journal = {PLOS Computational Biology}, keywords = {Ecology, Modelling and Simulation, Computational Theory and Mathematics, Genetics, Ecology, Evolution, Behavior and Systematics, Molecular Biology, Cellular and Molecular Neuroscience}, number = {11}, publisher = {Public Library of Science}, title = {{The Moran process on 2-chromatic graphs}}, doi = {10.1371/journal.pcbi.1008402}, volume = {16}, year = {2020}, } @inproceedings{8750, abstract = {Efficiently handling time-triggered and possibly nondeterministic switches for hybrid systems reachability is a challenging task. In this paper we present an approach based on conservative set-based enclosure of the dynamics that can handle systems with uncertain parameters and inputs, where the uncertainties are bound to given intervals. The method is evaluated on the plant model of an experimental electro-mechanical braking system with periodic controller. In this model, the fast-switching controller dynamics requires simulation time scales of the order of nanoseconds. Accurate set-based computations for relatively large time horizons are known to be expensive. However, by appropriately decoupling the time variable with respect to the spatial variables, and enclosing the uncertain parameters using interval matrix maps acting on zonotopes, we show that the computation time can be lowered to 5000 times faster with respect to previous works. This is a step forward in formal verification of hybrid systems because reduced run-times allow engineers to introduce more expressiveness in their models with a relatively inexpensive computational cost.}, author = {Forets, Marcelo and Freire, Daniel and Schilling, Christian}, booktitle = {18th ACM-IEEE International Conference on Formal Methods and Models for System Design}, isbn = {9781728191485}, location = {Virtual Conference}, publisher = {IEEE}, title = {{Efficient reachability analysis of parametric linear hybrid systems with time-triggered transitions}}, doi = {10.1109/MEMOCODE51338.2020.9314994}, year = {2020}, } @article{8758, abstract = {We consider various modeling levels for spatially homogeneous chemical reaction systems, namely the chemical master equation, the chemical Langevin dynamics, and the reaction-rate equation. Throughout we restrict our study to the case where the microscopic system satisfies the detailed-balance condition. The latter allows us to enrich the systems with a gradient structure, i.e. the evolution is given by a gradient-flow equation. We present the arising links between the associated gradient structures that are driven by the relative entropy of the detailed-balance steady state. The limit of large volumes is studied in the sense of evolutionary Γ-convergence of gradient flows. Moreover, we use the gradient structures to derive hybrid models for coupling different modeling levels.}, author = {Maas, Jan and Mielke, Alexander}, issn = {15729613}, journal = {Journal of Statistical Physics}, number = {6}, pages = {2257--2303}, publisher = {Springer Nature}, title = {{Modeling of chemical reaction systems with detailed balance using gradient structures}}, doi = {10.1007/s10955-020-02663-4}, volume = {181}, year = {2020}, }