Showing 1–6 of 6 results for author: Hoffmann, T

Are large language models superhuman chemists?

Authors: Adrian Mirza, Nawaf Alampara, Sreekanth Kunchapu, Benedict Emoekabu, Aswanth Krishnan, Mara Wilhelmi, Macjonathan Okereke, Juliane Eberhardt, Amir Mohammad Elahi, Maximilian Greiner, Caroline T. Holick, Tanya Gupta, Mehrdad Asgari, Christina Glaubitz, Lea C. Klepsch, Yannik Köster, Jakob Meyer, Santiago Miret, Tim Hoffmann, Fabian Alexander Kreth, Michael Ringleb, Nicole Roesner, Ulrich S. Schubert, Leanne M. Stafast, Dinga Wonanke , et al. (3 additional authors not shown)

Abstract: Large language models (LLMs) have gained widespread interest due to their ability to process human language and perform tasks on which they have not been explicitly trained. This is relevant for the chemical sciences, which face the problem of small and diverse datasets that are frequently in the form of text. LLMs have shown promise in addressing these issues and are increasingly being harnessed… ▽ More Large language models (LLMs) have gained widespread interest due to their ability to process human language and perform tasks on which they have not been explicitly trained. This is relevant for the chemical sciences, which face the problem of small and diverse datasets that are frequently in the form of text. LLMs have shown promise in addressing these issues and are increasingly being harnessed to predict chemical properties, optimize reactions, and even design and conduct experiments autonomously. However, we still have only a very limited systematic understanding of the chemical reasoning capabilities of LLMs, which would be required to improve models and mitigate potential harms. Here, we introduce "ChemBench," an automated framework designed to rigorously evaluate the chemical knowledge and reasoning abilities of state-of-the-art LLMs against the expertise of human chemists. We curated more than 7,000 question-answer pairs for a wide array of subfields of the chemical sciences, evaluated leading open and closed-source LLMs, and found that the best models outperformed the best human chemists in our study on average. The models, however, struggle with some chemical reasoning tasks that are easy for human experts and provide overconfident, misleading predictions, such as about chemicals' safety profiles. These findings underscore the dual reality that, although LLMs demonstrate remarkable proficiency in chemical tasks, further research is critical to enhancing their safety and utility in chemical sciences. Our findings also indicate a need for adaptations to chemistry curricula and highlight the importance of continuing to develop evaluation frameworks to improve safe and useful LLMs. △ Less

Submitted 1 April, 2024; originally announced April 2024.

Simulation of alnico coercivity

Authors: Liqin Ke, Ralph Skomski, Todd D. Hoffmann, Lin Zhou, Wei Tang, Duane D. Johnson, Matthew J. Kramer, Iver E. Anderson, C. -Z. Wang

Abstract: Micromagnetic simulations of alnico show substantial deviations from Stoner-Wohlfarth behavior due to the unique size and spatial distribution of the rod-like Fe-Co phase formed during spinodal decomposition in an external magnetic field. The maximum coercivity is limited by single-rod effects, especially deviations from ellipsoidal shape, and by interactions between the rods. Both the exchange in… ▽ More Micromagnetic simulations of alnico show substantial deviations from Stoner-Wohlfarth behavior due to the unique size and spatial distribution of the rod-like Fe-Co phase formed during spinodal decomposition in an external magnetic field. The maximum coercivity is limited by single-rod effects, especially deviations from ellipsoidal shape, and by interactions between the rods. Both the exchange interaction between connected rods and magnetostatic interaction between rods are considered, and the results of our calculations show good agreement with recent experiments. Unlike systems dominated by magnetocrystalline anisotropy, coercivity in alnico is highly dependent on size, shape, and geometric distribution of the Fe-Co phase, all factors that can be tuned with appropriate chemistry and thermal-magnetic annealing. △ Less

Submitted 13 July, 2017; originally announced July 2017.

Journal ref: Applied Physics Letters, 111(2):022403 (2017)

Decentralized Routing on Spatial Networks with Stochastic Edge Weights

Authors: Till Hoffmann, Renaud Lambiotte, Mason A. Porter

Abstract: We investigate algorithms to find short paths in spatial networks with stochastic edge weights. Our formulation of the problem of finding short paths differs from traditional formulations because we specifically do not make two of the usual simplifying assumptions: (1) we allow edge weights to be stochastic rather than deterministic; and (2) we do not assume that global knowledge of a network is a… ▽ More We investigate algorithms to find short paths in spatial networks with stochastic edge weights. Our formulation of the problem of finding short paths differs from traditional formulations because we specifically do not make two of the usual simplifying assumptions: (1) we allow edge weights to be stochastic rather than deterministic; and (2) we do not assume that global knowledge of a network is available. We develop a decentralized routing algorithm that provides en route guidance for travelers on a spatial network with stochastic edge weights without the need to rely on global knowledge about the network. To guide a traveler, our algorithm uses an estimation function that evaluates cumulative arrival probability distributions based on distances between pairs of nodes. The estimation function carries a notion of proximity between nodes and thereby enables routing without global knowledge. In testing our decentralized algorithm, we define a criterion that allows one to discriminate among arrival probability distributions, and we test our algorithm and this criterion using both synthetic and real networks. △ Less

Submitted 5 September, 2013; v1 submitted 29 September, 2012; originally announced October 2012.

Comments: 10 pages, 9 figures (some with multiple parts)

Journal ref: Phys. Rev. E 88, 022815 (2013)

Time-resolved imaging of magnetoelectric domain switching in multiferroic MnWO4

Authors: T. Hoffmann, P. Thielen, P. Becker, L. Bohaty, M. Fiebig

Abstract: By an electrical-pump--optical-probe technique we show that the electric-field-induced reversal of the magnetic order parameter in multiferroic MnWO$_4$ occurs on the time scale of milliseconds and maintains a rigid coupling of the magnetization to the magnetically induced electric polarization. The temporal progression of the spatially resolved domain structure was imaged with nanosecond resoluti… ▽ More By an electrical-pump--optical-probe technique we show that the electric-field-induced reversal of the magnetic order parameter in multiferroic MnWO$_4$ occurs on the time scale of milliseconds and maintains a rigid coupling of the magnetization to the magnetically induced electric polarization. The temporal progression of the spatially resolved domain structure was imaged with nanosecond resolution by optical second harmonic generation and compared to the quasi-static domain reversal. A qualitative model gives an estimate of why the magnetoelectric order-parameter reversal in the magnetically induced ferroelectrics is not inherently ultrafast. △ Less

Submitted 29 September, 2011; v1 submitted 10 March, 2011; originally announced March 2011.

Unconventional Multiferroicity in Cupric Oxide

Authors: Pierre Toledano, Naemi Leo, Dmitry D. Khalyavin, Laurent C. Chapon, Tim Hoffmann, Dennis Meier, Manfred Fiebig

Abstract: The magnetic phase transitions reported below 230 K in cupric oxide are analyzed theoretically at the macroscopic and microscopic levels. The incommensurate multiferroic and lock-in commensurate phases are shown to realize an inverted sequence of symmetry-breaking mechanisms with respect to the usual sequence occurring in low temperature multiferroic compounds. The higher temperature spin-spiral p… ▽ More The magnetic phase transitions reported below 230 K in cupric oxide are analyzed theoretically at the macroscopic and microscopic levels. The incommensurate multiferroic and lock-in commensurate phases are shown to realize an inverted sequence of symmetry-breaking mechanisms with respect to the usual sequence occurring in low temperature multiferroic compounds. The higher temperature spin-spiral phase results from coupled order-parameters which decouple at the lock-in transition to the commensurate ground state phase. Expressing the order-parameters in function of the magnetic spins allows determining the symmetries and magnetic structures of the equilibrium phases and the microscopic interactions which give rise to the polarization. △ Less

Submitted 22 January, 2011; originally announced January 2011.

Self-diffusion and Cooperative Diffusion in Semidilute Polymer Solutions as measured by Fluorescence Correlation Spectroscopy

Authors: Ute Zettl, Sebastian T. Hoffmann, Felix Koberling, Georg Krausch, Joerg Enderlein, Ludger Harnau, Matthias Ballauff

Abstract: We present a comprehensive investigation of polymer diffusion in the semidilute regime by fluorescence correlation spectroscopy (FCS) and dynamic light scattering (DLS). Using single-labeled polystyrene chains, FCS leads to the self-diffusion coefficient while DLS gives the cooperative diffusion coefficient for exactly the same molecular weights and concentrations. Using FCS we observe a new fas… ▽ More We present a comprehensive investigation of polymer diffusion in the semidilute regime by fluorescence correlation spectroscopy (FCS) and dynamic light scattering (DLS). Using single-labeled polystyrene chains, FCS leads to the self-diffusion coefficient while DLS gives the cooperative diffusion coefficient for exactly the same molecular weights and concentrations. Using FCS we observe a new fast mode in the semidilute entangled concentration regime beyond the slower mode which is due to self-diffusion. Comparison of FCS data with data obtained by DLS on the same polymers shows that the second mode observed in FCS is identical to the cooperative diffusion coefficient measured with DLS. An in-depth analysis and a comparison with current theoretical models demonstrates that the new cooperative mode observed in FCS is due to the effective long-range interaction of the chains through the transient entanglement network. △ Less

Submitted 4 February, 2010; originally announced February 2010.

Journal ref: Macromolecules 42, 9537 (2009)