Search for: All records

How do people perform general-purpose physical reasoning across a variety of scenarios in everyday life? Across two stud ies with seven different physical scenarios, we asked participants to predict whether or where two objects will make contact. People achieved high accuracy and were highly consistent with each other in their predictions. We hypothesize that this robust generalization is a consequence of mental simulations of noisy physics. We designed an “intuitive physics engine” model to capture this generalizable simulation. We find that this model generalized in human-like ways to unseen stimuli and to a different query of predictions. We evaluated several state-of-the-art deep learning and scene feature models on the same task and found that they could not explain human predictions as well. This study provides evidence that human’s robust generalization in physics predictions are supported by a probabilistic simulation model, and suggests the need for structure in learned dynamics models. 

The development of high-throughput experimentation (HTE) methods to efficiently screen multiparameter spaces is key to accelerating the discovery of high-performance multicomponent materials (e.g., polymer blends, colloids, etc.) for sensors, separations, energy, coatings, and other thin-film applications relevant to society. Although the generation and characterization of gradient thin-film library samples is a common approach to enable materials HTE, the ability to study many systems is impeded by the need to overcome unfavorable solubilities and viscosities among other processing challenges at ambient conditions. In this protocol, a solution coating system capable of operating temperatures over 110 °C is designed and demonstrated for the deposition of composition gradient polymer libraries. The system is equipped with a custom, solvent-resistant passive mixer module suitable for high-temperature mixing of polymer solutions at ambient pressure. Residence time distribution modeling was employed to predict the coating conditions necessary to generate composition gradient films using a poly(3-hexylthiophene) and poly(styrene) model system. Poly(propylene) and poly(styrene) blends were selected as a first demonstration of high temperature gradient film coating: the blend represents a polymer system where gradient films are traditionally difficult to generate via existing coating approaches due to solubility constraints at ambient conditions. The methodology developed here is expected to widen the range of solution processed materials that can be explored via high-throughput laboratory sampling and provides an avenue for efficiently screening multiparameter materials spaces and/or populating the large datasets required to enable data-driven materials science. 

Curating and analyzing centralized data repositories is a valuable approach in resolving the issue of reproducibility, gaining new insights and guiding future experiments, especially in the field of nanomaterials research. In this work, a data set containing processing information and mobility values of 115 DPP-DTT-based organic field effect transistors (OFET) was constructed from 15 publications. A customized classification algorithm was applied to the data set to help identify a reduced design region for polymer solution concentration that would be more likely to result in improved hole mobility. Experiments performed to confirm the insights from the data curation exercise revealed a strong influence of solution concentration on the polymer chain excitonic interactions and electronic performance. Devices fabricated at the critical chain overlap concentration of 5 g/L in chlorobenzene resulted in improved hole mobility, and were in good agreement with the insights provided by the classification algorithm.