Rapid construction of phase diagrams is a central tenet of combinatorial materials science with accelerated materials discovery efforts often hampered by challenges in interpreting combinatorial X-ray diffraction data sets, which we address by developing AgileFD, an artificial intelligence algorithm that enables rapid phase mapping from a combinatorial library of X-ray diffraction patterns. AgileFD models alloying-based peak shifting through a novel expansion of convolutional nonnegative matrix factorization, which not only improves the identification of constituent phases but also maps their concentration and lattice parameter as a function of composition. By incorporating Gibbs' phase rule into the algorithm, physically meaningful phase maps are obtained with unsupervised operation, and more refined solutions are attained by injecting expert knowledge of the system. The algorithm is demonstrated through investigation of the V-Mn-Nb oxide system where decomposition of eight oxide phases, including two with substantial alloying, provides the first phase map for this pseudoternary system. This phase map enables interpretation of high-throughput band gap data, leading to the discovery of new solar light absorbers and the alloying-based tuning of the direct-allowed band gap energy of MnVO. The open-source family of AgileFD algorithms can be implemented into a broad range of high throughput workflows to accelerate materials discovery.
The conversion efficiency of electric microwave signals into surface acoustic waves in different types of superconducting transducers is studied with the aim of quantum applications. We compare delay lines containing either conventional symmetric transducers (IDTs) or unidirectional transducers (UDTs) at 2.3 GHz and 10 mK. The UDT delay lines improve the insertion loss with 4.7 dB and a directivity of 22 dB is found for each UDT, indicating that 99.4 % of the acoustic power goes in the desired direction. The power lost in the undesired direction accounts for more than 90 % of the total loss in IDT delay lines, but only ∼3 % percent of the total loss in the FEUDT delay lines.Surface acoustic waves (SAWs) are Rayleigh waves propagating on the surface of a solid [1]. It has recently been suggested [2] and shown [3] that SAWs can interact with artificial atoms at the quantum level. This is fundamentally interesting because the artificial atoms can be made much larger than the wavelength of the SAW, which is not possible in other systems [4]. There are extensive new possibilities for quantum devices utilizing SAW; such as resonators [5,6], absorption in double quantum dots [7], transport of quantum information [8][9][10] and phonon assisted tunneling [11].When SAWs are used to carry quantum information, it is important to have low losses. The purpose here is to lower the conversion loss between electric signals (photons) and SAWs (phonons). In all studies about quantum SAW applications, SAWs are converted to and from electric microwave signals using conventional symmetric interdigital transducers (IDTs). The IDT can be described by a three port scattering matrix, where port 1 and 2 are acoustic and port 3 is electric [12]. It has the same electric to SAW conversion in both ports, i.e. S 13 = S 23 , and hence 50 % of the power is converted in the wrong direction. This means that IDTs are limited by a theoretical minimum insertion loss of -3 dB and because of reciprocity delay lines with two IDTs are theoretically limited to -6 dB.Unlike the symmetric IDT, a unidirectional transducer (UDT) [13,14] can be optimized to release most of its SAW energy in one preferred direction, by maximizing the scattering element S 13 while minimizing S 23 . In this way UDTs can exceed the -3 dB loss, and therefore UDTs are interesting to study for quantum SAW applications.
Weight decay (WD) is a traditional regularization technique in deep learning, but despite its ubiquity, its behavior is still an area of active research. Golatkar et al. have recently shown that WD only matters at the start of the training in computer vision, upending traditional wisdom. Loshchilov et al. show that for adaptive optimizers, manually decaying weights can outperform adding an l2 penalty to the loss. This technique has become increasingly popular and is referred to as decoupled WD. The goal of this paper is to investigate these two recent empirical observations. We demonstrate that by applying WD only at the start, the network norm stays small throughout training. This has a regularizing effect as the effective gradient updates become larger. However, traditional generalizations metrics fail to capture this effect of WD, and we show how a simple scale-invariant metric can. We also show how the growth of network weights is heavily influenced by the dataset and its generalization properties. For decoupled WD, we perform experiments in NLP and RL where adaptive optimizers are the norm. We demonstrate that the primary issue that decoupled WD alleviates is the mixing of gradients from the objective function and the l2 penalty in the buffers of Adam (which stores the estimates of the first-order moment). Adaptivity itself is not problematic and decoupled WD ensures that the gradients from the l2 term cannot "drown out" the true objective, facilitating easier hyperparameter tuning.
No abstract
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.