A reinforcement learning with switching controllers for a continuous action space

Reinforcement learning (RL) attracts much attention as a technique for realizing computational intelligence such as adaptive and autonomous decentralized systems. In general, however, it is not easy to put RL to practical use. This difficulty includes the problem of designing a suitable action space for an agent, i.e., satisfying two requirements in trade-off: (i) to keep the characteristics (or structure) of an original search space as much as possible in order to seek strategies that lie close to the optimal, and (ii) to reduce the search space as much as possible in order to expedite the learning process. In order to design a suitable action space adaptively, in this article, we propose a RL model with switching controllers based on Q-learning and an actor-critic to mimic the process of an infant’s motor development in which gross motor skills develop before fine motor skills. Then a method for switching controllers is constructed by introducing and referring to the “entropy.” Further, through computational experiments by using a path-planning problem with continuous action space, the validity and potential of the proposed method have been confirmed.     

In Situ Observation of Deformation in Semi-solid Fe-C Alloys at High Shear Rate

Synchrotron X-ray radiography at 125 frames per second was used to study deformation mechanisms in semi-solid Fe-C alloys at high solid fraction and shear strain rates of 10^?1/s. An image correlation approach was also used to quantify the shear strain fields and study shear-induced dilation and the origin of shear cracking. It was shown that, at high solid fraction (90 to 93 pct solid), rearrangement including rotation and translation of solid particles became restricted and shear strain localized into narrow liquid-filled channels/fissures. Shear cracking was shown to originate from inadequate liquid flow into the expanding spaces between solid particles caused by shear-induced dilation. At lower solid fraction (~85 pct solid), solid particles rearranged with a significantly higher component of rotation leading to more shear-induced dilation and a wider shear band.     

Sintering Kinetics at Isothermal Shrinkage: Effect of Specific Surface Area on the Initial Sintering Stage of Fine Zirconia Powder

The isothermal shrinkage behaviors of fine zirconia powders (containing 2.8–2.9 mol% Y₂O₃) with specific surface areas of about 6 and 16 m²/g were investigated to clarify the effect of specific surface area on the initial sintering stage. The shrinkage of powder compact was measured under constant temperatures in the range of 1000°–1100°C. The increase in specific surface area enhanced the densification rate with increasing temperature. The values of activation energy ( Q ) and frequency-factor term (β₀) of diffusion at initial sintering were estimated by applying the sintering-rate equation to the isothermal shrinkage data. The Q of diffusion changes little but the β₀ increases with the increase in specific surface area. It is therefore concluded that the increase in the specific surface area of fine zirconia powder enhances the shrinkage rate because of an increase in the β₀ at the initial stage of sintering.     

Damage Analysis of Hanshin Expressway Viaducts during 1995 Kobe Earthquake. III: Three-Span Continuous Girder Bridges

Almost all the single reinforced concrete (RC) piers from P35 to P350 received consistently severe damage, considering the large residual inclination of piers included in earthquake-induced severe damage. However, some of the piers in the section from P35 to P350 remained lightly damaged, and this phenomenon is observed especially in many piers under fixed bearings in continuous girder bridges. In this study, using experimentally based models for metal bearings and installing them to an existing FEM code, a nonlinear dynamic response analysis of a continuous girder bridge system is conducted. It is shown that the results depend on the ground motion, but the fuse effect of the breaking of the bearings could have been a reason for the phenomenon.     

Oxygen Diffusion in Single- and Poly-Crystalline Zinc Oxides

¹⁸O diffusion coefficients were measured in zinc oxide ceramics using a secondary ion mass spectrometer. The results are interpreted as indicating extrinsic behavior. The values of the lattice diffusion coefficients with higher valence dopants compared with zinc ions are greater than lower valence dopant such as lithium ions. Using the data at deeper depth, the grain boundary diffusivity of oxide ions was also evaluated. Although the lattice diffusion coefficients varied by two orders of magnitude, the products of grain boundary width and grain boundary diffusion coefficient were less sensitive to the type of dopants.     

Partition of Mn during the growth of proeutectoid ferrite allotriomorphs in an Fe-1.6 at. pct C-2.8 at. pct Mn alloy

A STEM analysis is made of the Mn distribution around grain boundary allotriomorphs of proeutectoid ferrite in an Fe-1.6 at. Pct C-2.8 at. Pct Mn alloy. Whereas the Mn enriched region is readily observed to extend along the austenite grain boundary, no substantial build-up or depletion of Mn near the ferrite : austenite interface is detected, consistent with the electron probe microanalysis previously reported. In the temperature range where the partition-local equilibrium (P-LE) mode has been proposed to prevail, measured parabolic growth rate constantsfall 1 to 2 orders of magnitude above that predicted from this model, but also below that calculated from the paraequilibrium (PE) model by roughly the same amount. A modification of the theory of grain boundary diffusion-aided growth of precipitates,i.e., the collector/rejector plate mechanism, on the other hand, accounts fairly well for the observed growth kinetics of ferrite allotriomorphs. However, only a slightly better accounting than the P-LE model is provided by this mechanism for the temperature dependence of Mn partition. Data on Ni partition, obtained in an Fe-0.5 at. Pct C-3.1 at. Pct Ni alloy, are also analyzed with the rejector plate model.     

Growth and led evaluation ofN-type conductive lec gaas co-doped with in and si

Dislocation-free and low dislocation densityn-type conductive GaAs crystals, 50 mm in diameter, were grown by the In and Si co-doping LEC technique. Two-dimensional LED arrays were fabricated on substrates obtained from these crystals by the MOCVD technique and the influence of the In doping on the LED characteristics was examined. The light output power of LEDs fabricated on co-doped substrate with an In concentration of 1 ×10²⁰atoms/cm³ are low and are non-uniformly distributed, as compared with the boat-grown substrate, even though the co-doped substrate is dislocation-free. However, the LED properties of a substrate with an In concentration of 2 × 10¹⁸atoms/cm³ are the same as those of a boat-grown substrate. The light output power of the LEDs becomes higher as the In concentration in the substrate decreased.