@article{KramerKreisbeckRihaetal., author = {Kramer, Tobias and Kreisbeck, Christoph and Riha, Christian and Chiatti, Olivio and Buchholz, Sven and Wieck, Andreas and Reuter, Dirk and Fischer, Saskia}, title = {Thermal energy and charge currents in multi-terminal nanorings}, series = {AIP Advances}, volume = {6}, journal = {AIP Advances}, doi = {10.1063/1.4953812}, pages = {065306}, abstract = {We study in experiment and theory thermal energy and charge transfer close to the quantum limit in a ballistic nanodevice, consisting of multiply connected one-dimensional electron waveguides. The fabricated device is based on an AlGaAs/GaAs heterostructure and is covered by a global top-gate to steer the thermal energy and charge transfer in the presence of a temperature gradient, which is established by a heating current. The estimate of the heat transfer by means of thermal noise measurements shows the device acting as a switch for charge and thermal energy transfer. The wave-packet simulations are based on the multi-terminal Landauer-B{\"u}ttiker approach and confirm the experimental finding of a mode-dependent redistribution of the thermal energy current, if a scatterer breaks the device symmetry.}, language = {en} } @article{CharronMusilGuljasetal., author = {Charron, Nicholas and Musil, F{\´e}lix and Guljas, Andrea and Chen, Yaoyi and Bonneau, Klara and Pasos-Trejo, Aldo and Jacopo, Venturin and Daria, Gusew and Zaporozhets, Iryna and Kr{\"a}mer, Andreas and Templeton, Clark and Atharva, Kelkar and Durumeric, Aleksander and Olsson, Simon and P{\´e}rez, Adri{\`a} and Majewski, Maciej and Husic, Brooke and Patel, Ankit and De Fabritiis, Gianni and No{\´e}, Frank and Clementi, Cecilia}, title = {Navigating protein landscapes with a machine-learned transferable coarse-grained model}, series = {Arxiv}, journal = {Arxiv}, doi = {https://doi.org/10.48550/arXiv.2310.18278}, abstract = {The most popular and universally predictive protein simulation models employ all-atom molecular dynamics (MD), but they come at extreme computational cost. The development of a universal, computationally efficient coarse-grained (CG) model with similar prediction performance has been a long-standing challenge. By combining recent deep learning methods with a large and diverse training set of all-atom protein simulations, we here develop a bottom-up CG force field with chemical transferability, which can be used for extrapolative molecular dynamics on new sequences not used during model parametrization. We demonstrate that the model successfully predicts folded structures, intermediates, metastable folded and unfolded basins, and the fluctuations of intrinsically disordered proteins while it is several orders of magnitude faster than an all-atom model. This showcases the feasibility of a universal and computationally efficient machine-learned CG model for proteins.}, language = {en} } @article{KraemerDurumericCharronetal., author = {Kr{\"a}mer, Andreas and Durumeric, Aleksander and Charron, Nicholas and Chen, Yaoyi and Clementi, Cecilia and No{\´e}, Frank}, title = {Statistically optimal force aggregation for coarse-graining molecular dynamics}, series = {The Journal of Physical Chemistry Letters}, volume = {14}, journal = {The Journal of Physical Chemistry Letters}, number = {17}, doi = {10.1021/acs.jpclett.3c00444}, pages = {3970 -- 3979}, abstract = {Machine-learned coarse-grained (CG) models have the potential for simulating large molecular complexes beyond what is possible with atomistic molecular dynamics. However, training accurate CG models remains a challenge. A widely used methodology for learning bottom-up CG force fields maps forces from all-atom molecular dynamics to the CG representation and matches them with a CG force field on average. We show that there is flexibility in how to map all-atom forces to the CG representation and that the most commonly used mapping methods are statistically inefficient and potentially even incorrect in the presence of constraints in the all-atom simulation. We define an optimization statement for force mappings and demonstrate that substantially improved CG force fields can be learned from the same simulation data when using optimized force maps. The method is demonstrated on the miniproteins chignolin and tryptophan cage and published as open-source code.}, language = {en} }