Tjong:A transformer‐based Mahjong AI via hierarchical decision‐making and fan backward

Mahjong,a complex game with hidden information and sparse rewards,poses significant challenges.Existing Mahjong AIs require substantial hardware resources and extensive datasets to enhance AI capabilities.The authors propose a transformer‐based Mahjong AI(Tjong)via hierarchical decision‐making.By utilising self‐attention mechanisms,Tjong effectively captures tile patterns and game dynamics,and it decouples the decision pro-cess into two distinct stages:action decision and tile decision.This design reduces de-cision complexity considerably.Additionally,a fan backward technique is proposed to address the sparse rewards by allocating reversed rewards for actions based on winning hands.Tjong consists of 15M parameters and is trained using approximately 0.5 M data over 7 days of supervised learning on a single server with 2 GPUs.The action decision achieved an accuracy of 94.63%,while the claim decision attained 98.55%and the discard decision reached 81.51%.In a tournament format,Tjong outperformed AIs(CNN,MLP,RNN,ResNet,VIT),achieving scores up to 230%higher than its opponents.Further-more,after 3 days of reinforcement learning training,it ranked within the top 1%on the leaderboard on the Botzone platform.

凝胶浴对钒电池用侧链型杂萘联苯聚醚酮多孔膜结构和性能的影响

以含苄基氯结构的杂萘联苯聚醚酮(CMPPEK-P)为原料、三甲胺为胺化试剂,通过先制膜后胺化的方式,利用水蒸气诱导相分离法(VIPS)成功制备了一种高性能钒电池用含有季铵基侧链的杂萘联苯聚醚酮(QAPPEK-P)多孔阴离子交换膜。研究了在VIPS过程中,不同浓度氯化钠水溶液(质量分数0%~26%,下同)作为凝胶浴对膜结构和性能的影响:提高氯化钠浓度,QAPPEK-P膜的孔隙率提高,面电阻显著降低,而其阻钒性能无明显变化。其中,在凝胶浴盐浓度为26%条件下制得的QAPPEK-P膜,装载其单电池的电池效率最佳:在电流密度为140 mA/cm2时,能量效率可达80%;且其电池性能在120 mA/cm2的电流密度下,充放电150个循环后无明显变化,显示出良好的化学稳定性,证明了QAPPEK-P在多孔阴离子交换膜在钒电池应用中的潜力。

Coordinated Restoration Method for Electric Buses and Network Reconfigurations in Distribution Systems Under Extreme Events

Distribution systems are facing challenges in serving lifeline loads after extreme events.Network reconfiguration is a traditional and practical method for power supply restoration,which has strong but inflexible power transfer capabilities influenced by network topology.Multiple failures of utility power under extreme events will further limit the efficiency of network reconfiguration.Electric buses(EBs)can be utilized to achieve power supply considering their discharging capabilities as mobile storage devices.However,the mobility of EBs and the influences of transport systems must be carefully considered to enhance the resilience of distribution systems.Reconfiguration and EBs are complementary in terms of recovery capabilities and location flexibility,and more important loads can be recovered by the coordination between EBs and network reconfiguration.This paper proposes a coordinated restoration method for EBs and reconfigurations considering the influences of transport systems.The post-disaster restoration problem is formulated as a bi-level model,in which the network topology is optimized in the upperlevel aiming at maximizing restoration loads through the main grid and EBs,while the traffic paths of all EBs are optimized with the goal of maximizing the restoration loads by the EBs in the lower-level considering time consumption and energy consumption during movement.The PSO and a genetic algorithm are used to solve the proposed bi-level optimization problem.Simulation studies are performed to verify the superiority of the proposed method.

Changes in productivity and carbon storage of grasslands in china under future global warming scenarios of 1.5℃ and 2℃

Aims the impacts of future global warming of 1.5℃ and 2℃ on the productivity and carbon(c)storage of grasslands in china are not clear yet,although grasslands in china support~45 million agricultural populations and more than 238 million livestock populations,and are sensitive to global warming.Methods this study used a process-based terrestrial ecosystem model named ORcHIDEE to simulate c cycle of alpine meadows and temperate grasslands in china.this model was driven by high-resolution(0.5°×0.5°)climate of global specific warming levels(SWL)of 1.5℃ and 2℃(warmer than pre-industrial level),which is downscaled by Ec-EARtH3-HR v3.1 with sea surface temperature and sea-ice concentration as boundary conditions from IPSL-cM5-LR(low spatial resolution,2.5°×1.5°)Earth system model(ESM).Important Findingscompared with baseline(1971-2005),the mean annual air temperature over chinese grasslands increased by 2.5℃ and 3.7℃ under SWL1.5 and SWL2,respectively.the increase in temperature in the alpine meadow was higher than that in the temperate grassland under both SWL1.5 and SWL2.Precipitation was also shown an increasing trend under SWL2 over most of the chinese grasslands.Strong increases in gross primary productivity(GPP)were simulated in the chinese grasslands,and the mean annual GPP(GPP_(MA))increased by 19.32%and 43.62%under SWL1.5 and SWL2,respectively.the c storage increased by 0.64 Pg c and 1.37 Pg c under SWL1.5 and SWL2 for 50 years simulations.the GPP_(MA) was 0.67_(0.39)^(0.88)(0.82)(model mean_(min) ^(max) (this study)),0.85_(0.45)^(1.24)(0.97)and 0.94_(0.61)^(1.30)(1.17)Pg C year^(−1) under baseline,SWL1.5 and SWL2 modeled by four cMIP5 ESMs(phase 5 of the coupled Model Inter-comparison Project Earth System Models).In contrast,the mean annual net biome productivity was−18.55_(−40.37)^(4.47)(−3.61),18.65_(−2.03)^(64.03)(10.29)and 24.15_(8.38)^(38.77)(24.93)Tg C year^(−1) under base-line,SWL1.5 and SWL2 modeled by the four cMIP5 ESMs.Our results indicated that the chinese grasslands would have higher productivity than the baseline and can mitigate climate change through increased C sequestration under future global warming of 1.5℃ and 2℃ with the increase of precipitation and the global increase of atmospheric CO_(2) concentration.

Phase-shifting technique applied to circular harmonic-based joint transform correlator

The phase-shifting technique is applied to the circular harmonic expansion-based jointtransform correlator. Computer simulation has shown that the light efficiency and the discrimination capability are greatly enhanced, and the full rotation invariance is preserved after the phase-shifting technique has been used. A rotation-invariant optical pattern recognition with high discrimination capability and high light efficiency is obtained. The influence of the additive noise on the performance of the correlator is also investigated. However, the anti-noise capability of this kind of correlator still needs improving.