[{"data":1,"prerenderedAt":518},["ShallowReactive",2],{"content-query-uKHYGlxwzV":3},{"_path":4,"_dir":5,"_draft":6,"_partial":6,"_locale":7,"title":8,"description":9,"date":10,"cover":11,"type":12,"category":13,"body":14,"_type":512,"_id":513,"_source":514,"_file":515,"_stem":516,"_extension":517},"/technology-blogs/zh/2775","zh",false,"","MindSpore AI科学计算系列 | 以MindSpore Elec实践为例的智能电磁计算若干进展综述","东南大学崔铁军院士团队近期发表了一篇“智能电磁计算的若干进展”的综述论文[1]，里面详尽地描述了人工智能在电磁计算领域的进展，为读者入门并了解该领域最新的研究成果提供有益帮助。","2023-10-09","https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2023/10/12/92811b13f4f441a796464715949af4f6.png","technology-blogs","大V博文",{"type":15,"children":16,"toc":507},"root",[17,25,44,52,57,67,72,77,85,90,99,104,109,114,119,126,131,139,144,151,156,164,169,177,182,189,196,203,213,218,223,237,242,249,254,268,273,278,285,290,300,305,310,329,334,339,344,349,354,364,369,378,383,396,407,412,417,422,427,432,437,442,447,452,457,462,467,477,487,497],{"type":18,"tag":19,"props":20,"children":22},"element","h1",{"id":21},"mindspore-ai科学计算系列-以mindspore-elec实践为例的智能电磁计算若干进展综述",[23],{"type":24,"value":8},"text",{"type":18,"tag":26,"props":27,"children":28},"p",{},[29,31,37,39],{"type":24,"value":30},"**作者：**",{"type":18,"tag":32,"props":33,"children":34},"strong",{},[35],{"type":24,"value":36},"于璠",{"type":24,"value":38}," ",{"type":18,"tag":32,"props":40,"children":41},{},[42],{"type":24,"value":43},"来源：知乎",{"type":18,"tag":26,"props":45,"children":46},{},[47],{"type":18,"tag":32,"props":48,"children":49},{},[50],{"type":24,"value":51},"背景",{"type":18,"tag":26,"props":53,"children":54},{},[55],{"type":24,"value":56},"东南大学崔铁军院士团队近期发表了一篇“智能电磁计算的若干进展”的综述论文[1]，里面详尽地描述了人工智能在电磁计算领域的进展，为读者入门并了解该领域最新的研究成果提供有益帮助。昇思MindSpore[2]是最早提出电磁仿真套件的AI框架，本文将结合昇思MindSpore Elec电磁仿真套件[3]进行论文分析。论文从算法层面分别介绍了智能电磁计算在正向电磁仿真和逆向电磁成像上的最新研究成果，随后从软硬件或数字物理相结合的系统层面介绍了基于信息超材料的智能计算新体制和相关应用，最后对全文进行了总结并预测智能电磁计算的发展方向。本文主要围绕算法软件层面介绍。",{"type":18,"tag":26,"props":58,"children":59},{},[60,62],{"type":24,"value":61},"**1、**",{"type":18,"tag":32,"props":63,"children":64},{},[65],{"type":24,"value":66},"正向智能电磁仿真",{"type":18,"tag":26,"props":68,"children":69},{},[70],{"type":24,"value":71},"正向电磁仿真技术在电磁兼容分析、电子器件设计、信号处理、通信网络设计等领域发挥着举足轻重的作用。掌握自主可控的精准快速的正向电磁仿真技术是衡量一个国家科学技术水准以及工业制造能力的重要指标。",{"type":18,"tag":26,"props":73,"children":74},{},[75],{"type":24,"value":76},"正向电磁仿真计算方法主要包括有限差分法、有限元法、矩量法等全波仿真方法，以及弹跳射线法等高频渐近方法。但面临实时化、多尺度等要求仍然相距甚远。因此，需提出一种全新的计算范式来解决传统方法面临的计算效率难题。智能计算可实现正向仿真效率提升，其本质是通过学习输入到输出的映射关系，提取有效物理信息，从而构建等效神经网络模型代替传统数值算子，在保证计算精度基本不变前提下实现计算复杂度的降阶。正向智能电磁计算主要分为数据驱动以及物理驱动两类。智能化被视为计算电磁领域未来最为重要的发展方向之一。",{"type":18,"tag":26,"props":78,"children":79},{},[80],{"type":18,"tag":32,"props":81,"children":82},{},[83],{"type":24,"value":84},"1.1****数据驱动的正向电磁计算",{"type":18,"tag":26,"props":86,"children":87},{},[88],{"type":24,"value":89},"数据驱动的电磁计算大致分为结果学习（即直接学习从电磁参数到期望计算结果的映射，包括场值和电流等）和过程学习（即用神经网络代替传统仿真方法中的某一个中间环节，实现计算效率提升），如图1所示。",{"type":18,"tag":26,"props":91,"children":92},{},[93],{"type":18,"tag":94,"props":95,"children":98},"img",{"alt":96,"src":97},"image.png","https://fileserver.developer.huaweicloud.com/FileServer/getFile/cmtybbs/e64/154/b38/90a1d5d431e64154b387b3660e356ff5.20231011021852.54212967324889654979046032817964:50541011062818:2400:ECCBFF5D54BD40EF29EE20EDF7CC237AC503259E7AE3A08DDE4EE8C927F79EA3.png",[],{"type":18,"tag":26,"props":100,"children":101},{},[102],{"type":24,"value":103},"图1. 数据驱动的正向电磁计算分类",{"type":18,"tag":26,"props":105,"children":106},{},[107],{"type":24,"value":108},"结果学习是一种最直接的策略。如文献[4]采用CNN代替频域有限差分法(FDFD)求解Helmholtz方程(见图2a所示)。文献[5]基于注意力机制设计的等效求解器表现优异（见图2b），在其所给测试数据集上RCS预测正确率超过98%，且相比于矩量法获得接近100倍的计算加速比。结果学习直观、高效，但由于缺乏物理规律指导，求解精度和泛化能力往往不尽如人意。",{"type":18,"tag":26,"props":110,"children":111},{},[112],{"type":24,"value":113},"以智能手段加速中间环节的过程学习同样受到了关注，如文献[6,7]提出了“智能吸收边界” 方案，分别采用RNN以及长短期记忆神经网络(Long Short Term Memory, LSTM)代替理想匹配层吸收边界条件(Perfectly Matched Layer, PML)，能够在单层智能边界的条件下达到多层PML的吸收效果，其中RNN方案更快，能够实现约2倍的计算提速，但吸收效果不及LSTM方案。",{"type":18,"tag":26,"props":115,"children":116},{},[117],{"type":24,"value":118},"过程学习方案相比结果学习虽然引入了更多物理信息，整体泛化能力得到了改善，但计算效率增益通常会大打折扣，很少能够观测到出现超过1个数量级的提升。如何进一步降低过程学习方案计算复杂度，也是需要在未来深入研究的问题。",{"type":18,"tag":26,"props":120,"children":121},{},[122],{"type":18,"tag":94,"props":123,"children":125},{"alt":96,"src":124},"https://fileserver.developer.huaweicloud.com/FileServer/getFile/cmtybbs/e64/154/b38/90a1d5d431e64154b387b3660e356ff5.20231011021914.98508671398060732132458575112520:50541011062818:2400:3BD103E908614418B15D6CC2CF421FFE27F940F0ECE39380583124AE227134C4.png",[],{"type":18,"tag":26,"props":127,"children":128},{},[129],{"type":24,"value":130},"图2. 部分数据驱动正向电磁计算研究成果",{"type":18,"tag":26,"props":132,"children":133},{},[134],{"type":18,"tag":32,"props":135,"children":136},{},[137],{"type":24,"value":138},"1.2****物理驱动的正向电磁计算",{"type":18,"tag":26,"props":140,"children":141},{},[142],{"type":24,"value":143},"物理驱动的深度神经网络以PINNs(Physics-Informed Neural Network, PINN)为代表，该方法在提升网络逼近能力的同时减少了数据依赖，特别适用于解决小样本学习问题。如文献[8]在U-Net架构基础上引入频域电场方程作为损失函数，提出了求解自由空间散射光场的MaxwellNet，如图3(b)所示，这一成果被应用于指导光学透镜设计[9]。",{"type":18,"tag":26,"props":145,"children":146},{},[147],{"type":18,"tag":94,"props":148,"children":150},{"alt":96,"src":149},"https://fileserver.developer.huaweicloud.com/FileServer/getFile/cmtybbs/e64/154/b38/90a1d5d431e64154b387b3660e356ff5.20231011021936.74111083423661065299321609077481:50541011062818:2400:57719145B053B8CA883BFDD75FC751E5B6FF348113E610D7A731F77DBE7A5685.png",[],{"type":18,"tag":26,"props":152,"children":153},{},[154],{"type":24,"value":155},"图3. 部分物理驱动以及算子学习正向计算研究成果",{"type":18,"tag":26,"props":157,"children":158},{},[159],{"type":18,"tag":32,"props":160,"children":161},{},[162],{"type":24,"value":163},"1.3****基于算子学习的正向电磁计算",{"type":18,"tag":26,"props":165,"children":166},{},[167],{"type":24,"value":168},"DeepONet、FNO是目前较流行的神经算子模型，FNO在解决流体问题上的成功也带给了电磁计算启发。文献[10]提出用于求解频域自由空间散射问题的改进FNO，相较简单的U-Net等效求解器，不论是计算精度，还是训练以及推理速度都出现了显著提升。文献[11]提出求解频域Maxwell方程组的扩展FNO，相较于FDFD获得了超过100倍的加速比。",{"type":18,"tag":26,"props":170,"children":171},{},[172],{"type":18,"tag":32,"props":173,"children":174},{},[175],{"type":24,"value":176},"1.4****计算可微分正向电磁计算",{"type":18,"tag":26,"props":178,"children":179},{},[180],{"type":24,"value":181},"FDTD算法本身可微分，能够直接嵌入不同功能的可微分系统；另一方面，可以借助现有深度学习平台对于并行计算的支持，加速正向仿真过程，如图4(a)[12]。针对不可微分算法(如高频方法)。如图4(b)所示，文献[13]提出了一套可微分合成孔径雷达(Synthetic Aperture Radar, SAR)渲染系统，能够利用梯度下降算法从目标二维图像中推断出三维信息。",{"type":18,"tag":26,"props":183,"children":184},{},[185],{"type":18,"tag":94,"props":186,"children":188},{"alt":96,"src":187},"https://fileserver.developer.huaweicloud.com/FileServer/getFile/cmtybbs/e64/154/b38/90a1d5d431e64154b387b3660e356ff5.20231011021954.52773092409602244471660907486992:50541011062818:2400:21A26D57432676DB444CD51E3363F5A8B54135E3612C7E329F5091F9C111F3B0.png",[],{"type":18,"tag":26,"props":190,"children":191},{},[192],{"type":18,"tag":94,"props":193,"children":195},{"alt":96,"src":194},"https://fileserver.developer.huaweicloud.com/FileServer/getFile/cmtybbs/e64/154/b38/90a1d5d431e64154b387b3660e356ff5.20231011022016.94154652807595011375002797355177:50541011062818:2400:51056F96BFCF52427BC9F44F6B3EEE816EE7C6C96799FB55E5647C8F2CC2D499.png",[],{"type":18,"tag":26,"props":197,"children":198},{},[199],{"type":18,"tag":94,"props":200,"children":202},{"alt":96,"src":201},"https://fileserver.developer.huaweicloud.com/FileServer/getFile/cmtybbs/e64/154/b38/90a1d5d431e64154b387b3660e356ff5.20231011022028.84673980990732831181341101652260:50541011062818:2400:A1CB1909CE3B1B69DCA53E333C89EF82C19905C38F2FF48DE20EB8D503B3C719.png",[],{"type":18,"tag":26,"props":204,"children":205},{},[206,208],{"type":24,"value":207},"**2、**",{"type":18,"tag":32,"props":209,"children":210},{},[211],{"type":24,"value":212},"逆向智能电磁成像",{"type":18,"tag":26,"props":214,"children":215},{},[216],{"type":24,"value":217},"电磁逆散射成像已经被广泛应用于无损探伤、地质勘探、癌症检测、安全检查等。但由于逆散射问题本身所固有的非线性和病态性，寻找合适的逆散射成像映射关系是一个非常具有挑战性的问题，尤其是在高噪声的环境中。",{"type":18,"tag":26,"props":219,"children":220},{},[221],{"type":24,"value":222},"逆向智能电磁成像的优势在于，它可以从数据中学习映射规律，因此免去了复杂的电磁模型推理和构造的过程，也免去了优化算法中的迭代过程，极大提升了成像的效率。同时，针对特定的逆散射问题，深度学习网络能够学习出暗含几何先验信息的映射关系，可提升成像的精度，甚至实现突破成像分辨率极限的超分辨成像。",{"type":18,"tag":26,"props":224,"children":225},{},[226,231,232],{"type":18,"tag":32,"props":227,"children":228},{},[229],{"type":24,"value":230},"2.1",{"type":24,"value":38},{"type":18,"tag":32,"props":233,"children":234},{},[235],{"type":24,"value":236},"纯数据驱动的逆向智能电磁成像",{"type":18,"tag":26,"props":238,"children":239},{},[240],{"type":24,"value":241},"文献[14]使用U-Net网络，进一步学习并训练了3种成像映射关系，这3种映射的输出都是目标图像输入分别为原始散射电场回波测量数据、由BP算法生成的初步图像和由主成分分析法获得的感应电流数据。在文中作者分别称这3种映射关系为直接反演模式、反向传播模式和主成分电流模式。经过测试，反向传播模式和主成分电流模式均能够生成较为理想的目标图像，但直接反演模式的成像效果不佳，如图5(b)所示。",{"type":18,"tag":26,"props":243,"children":244},{},[245],{"type":18,"tag":94,"props":246,"children":248},{"alt":96,"src":247},"https://fileserver.developer.huaweicloud.com/FileServer/getFile/cmtybbs/e64/154/b38/90a1d5d431e64154b387b3660e356ff5.20231011022047.11069467716348560159372788619132:50541011062818:2400:E9B70EA4FD9CDBD5B35B9BFAF01CB12D7A94A545DDEAAB82AB392BC19E26AF8E.png",[],{"type":18,"tag":26,"props":250,"children":251},{},[252],{"type":24,"value":253},"图5. 基于U-Net结构的逆向智能电磁成像",{"type":18,"tag":26,"props":255,"children":256},{},[257,262,263],{"type":18,"tag":32,"props":258,"children":259},{},[260],{"type":24,"value":261},"2.2",{"type":24,"value":38},{"type":18,"tag":32,"props":264,"children":265},{},[266],{"type":24,"value":267},"电磁物理驱动的逆向智能电磁成像",{"type":18,"tag":26,"props":269,"children":270},{},[271],{"type":24,"value":272},"将电磁物理机理或方程引入逆散射深度学习网络的结构设计和误差函数设计中，为逆散射问题定制专用的深度学习模型，能够更容易地学习到输入和输出之间的非线性关系。",{"type":18,"tag":26,"props":274,"children":275},{},[276],{"type":24,"value":277},"文献[15]将逆散射迭代优化算法的方程结构引入深度学习网络的结构设计中，级联多层复值残差卷积神经网络模块构建出了专用于逆向智能电磁成像的深度神经网络，称之为DeepNIS，如图6(a)所示。通过仿真和实测都证实了DeepNIS在生成图像质量和计算时间方面都显著优于传统的非线性逆散射方法。",{"type":18,"tag":26,"props":279,"children":280},{},[281],{"type":18,"tag":94,"props":282,"children":284},{"alt":96,"src":283},"https://fileserver.developer.huaweicloud.com/FileServer/getFile/cmtybbs/e64/154/b38/90a1d5d431e64154b387b3660e356ff5.20231011022108.87595175638234369025403493714053:50541011062818:2400:598FD2633A7858B464C89C48B57418DADF9C37930FED679572AF57F0B321EB2F.png",[],{"type":18,"tag":26,"props":286,"children":287},{},[288],{"type":24,"value":289},"图6. 借鉴迭代优化算法的端到端逆向智能电磁成像",{"type":18,"tag":26,"props":291,"children":292},{},[293,295],{"type":24,"value":294},"**3、**",{"type":18,"tag":32,"props":296,"children":297},{},[298],{"type":24,"value":299},"昇思MindSpore Elec实践",{"type":18,"tag":26,"props":301,"children":302},{},[303],{"type":24,"value":304},"昇思MindSpore Elec基本涵盖了正向智能电磁仿真和逆向智能电磁成像。",{"type":18,"tag":26,"props":306,"children":307},{},[308],{"type":24,"value":309},"正向智能电磁仿真：",{"type":18,"tag":26,"props":311,"children":312},{},[313,315,320,322,327],{"type":24,"value":314},"a) 数据驱动：终端手机的AI电磁仿真，仿真精度媲美传统科学计算软件，同时",{"type":18,"tag":32,"props":316,"children":317},{},[318],{"type":24,"value":319},"性能提升10倍",{"type":24,"value":321},"（结果学习）。“金陵.电磁脑” AI电磁仿真基础模型精度媲美传统方法，",{"type":18,"tag":32,"props":323,"children":324},{},[325],{"type":24,"value":326},"效率提升10+倍",{"type":24,"value":328},"，而且随着目标规模的增大，该提升将会更加显著（过程学习）。",{"type":18,"tag":26,"props":330,"children":331},{},[332],{"type":24,"value":333},"b) 物理驱动：基于PINNs方法求解二维时域MaxWell方程时，通过高斯分布函数平滑、多通道残差网络结合sin激活函数的网络结构以及自适应加权的多任务学习策略，使得求解精度和性能均明显优于其他框架及方法。",{"type":18,"tag":26,"props":335,"children":336},{},[337],{"type":24,"value":338},"c) 计算可微分正向电磁计算: 时域有限差分（FDTD）方法求解麦克斯韦方程组的过程等价于一个循环卷积网络（RCNN）。利用MindSpore的可微分算子重写更新流程，便可得到端到端可微分FDTD。三维贴片天线S参数仿真精度与BenchMark一致。",{"type":18,"tag":26,"props":340,"children":341},{},[342],{"type":24,"value":343},"逆向智能电磁成像:",{"type":18,"tag":26,"props":345,"children":346},{},[347],{"type":24,"value":348},"a) 纯数据驱动: 基于GPRMAX软件生成探地雷达（Ground Penetrating Radar，GPR）反演的训练数据，利用AI模型通过输入电磁波信号快速精确获得目标结构。",{"type":18,"tag":26,"props":350,"children":351},{},[352],{"type":24,"value":353},"b) 电磁物理驱动: 基于端到端可微FDTD求解二维TM模式的电磁逆散射问题。反演得到的相对介电常数SSIM达0.9635，与目标（下图右）高度吻合；使用物理辅助对抗生成网络（Physics-assisted GAN）的AI方法针对超表面全息成像设计进行非监督学习，避免了数据集的制作过程，并且和传统的GS算法相比在指标和视觉感受上效果更优。",{"type":18,"tag":26,"props":355,"children":356},{},[357,359],{"type":24,"value":358},"**4、**",{"type":18,"tag":32,"props":360,"children":361},{},[362],{"type":24,"value":363},"展望",{"type":18,"tag":26,"props":365,"children":366},{},[367],{"type":24,"value":368},"昇思MindSpore Elec在智能电磁方面已经开展了许多工作，我们也欢迎广大的科学计算爱好者和研究者加入，共同拓展和维护昇思MindSpore Elec套件。",{"type":18,"tag":370,"props":371,"children":373},"h2",{"id":372},"参考文献",[374],{"type":18,"tag":32,"props":375,"children":376},{},[377],{"type":24,"value":372},{"type":18,"tag":26,"props":379,"children":380},{},[381],{"type":24,"value":382},"[1] LIU Che, YANG Kaiqiao, BAO Jianghan, et al. Recent progress in intelligent electromagnetic computing[J]. Journal of Radars, 2023, 12(4): 1–27. doi: 10.12000/JR23133",{"type":18,"tag":26,"props":384,"children":385},{},[386,388],{"type":24,"value":387},"[2] ",{"type":18,"tag":389,"props":390,"children":394},"a",{"href":391,"rel":392},"https://mindspore.cn",[393],"nofollow",[395],{"type":24,"value":391},{"type":18,"tag":26,"props":397,"children":398},{},[399,401],{"type":24,"value":400},"[3] ",{"type":18,"tag":389,"props":402,"children":405},{"href":403,"rel":404},"https://gitee.com/mindspore/mindscience/tree/master/MindElec",[393],[406],{"type":24,"value":403},{"type":18,"tag":26,"props":408,"children":409},{},[410],{"type":24,"value":411},"[4] QI Shutong, WANG Yinpeng, LI Yongzhong, et al. Twodimensional electromagnetic solver based on deep learning technique[J]. IEEE Journal on Multiscale and Multiphysics Computational Techniques, 2020, 5: 83–88. doi: 10.1109/JMMCT.2020.2995811",{"type":18,"tag":26,"props":413,"children":414},{},[415],{"type":24,"value":416},"[5] KONG Dehua, ZHANG Wenwei, HE Xiaoyang, et al. Intelligent prediction for scattering properties based on multihead attention and target inherent feature parameter[J]. IEEE Transactions on Antennas and Propagation, 2023, 71(6): 5504–5509. doi: 10.1109/TAP. 2023.3262341.",{"type":18,"tag":26,"props":418,"children":419},{},[420],{"type":24,"value":421},"[6] YAO Heming and JIANG Lijun. Machine-learning-based PML for the FDTD method[J]. IEEE Antennas and Wireless Propagation Letters, 2019, 18(1): 192–196. doi: 10.1109/LAWP.2018.2885570.",{"type":18,"tag":26,"props":423,"children":424},{},[425],{"type":24,"value":426},"[7] YAO Heming and JIANG Lijun. Enhanced PML based on the long short term memory network for the FDTD method[J]. IEEE Access, 2020, 8: 21028–21035. doi: 10.1109/ACCESS.2020.2969569",{"type":18,"tag":26,"props":428,"children":429},{},[430],{"type":24,"value":431},"[8] LIM J and PSALTIS D. MaxwellNet: Physics-driven deep neural network training based on Maxwell's equations[J]. APL Photonics, 2022, 7(1): 011301. doi: 10.1063/5.0071616.",{"type":18,"tag":26,"props":433,"children":434},{},[435],{"type":24,"value":436},"[9] GIGLI C, SABA A, AYOUB A B, et al. Predicting nonlinear optical scattering with physics-driven neural networks[J]. APL Photonics, 2023, 8(2): 026105. doi: 10.1063/5.0119186.",{"type":18,"tag":26,"props":438,"children":439},{},[440],{"type":24,"value":441},"[10] AUGENSTEIN Y, REPÄN T, and ROCKSTUHL C. Neural operator-based surrogate solver for free-form electromagnetic inverse design[J]. ACS Photonics, 2023,10(5): 1547–1557. doi: 10.1021/acsphotonics.3c00156.",{"type":18,"tag":26,"props":443,"children":444},{},[445],{"type":24,"value":446},"[11] PENG Zhong, YANG Bo, XU Yixian, et al. Rapid surrogate modeling of electromagnetic data in frequency domain using neural operator[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 2007912. doi: 10.1109/TGRS.2022.3222507",{"type":18,"tag":26,"props":448,"children":449},{},[450],{"type":24,"value":451},"[12] GUO Liangshuai, LI Maokun, XU Shenheng, et al. Electromagnetic modeling using an FDTD-equivalent recurrent convolution neural network: Accurate computing on a deep learning framework[J]. IEEE Antennas and Propagation Magazine, 2023, 65(1): 93–102. doi: 10.1109/MAP.2021.3127514.",{"type":18,"tag":26,"props":453,"children":454},{},[455],{"type":24,"value":456},"[13] FU Shilei and XU Feng. Differentiable SAR renderer and image-based target reconstruction[J]. IEEE Transactions on Image Processing, 2022, 31: 6679–6693. doi: 10.1109/TIP.2022.3215069.",{"type":18,"tag":26,"props":458,"children":459},{},[460],{"type":24,"value":461},"[14] WEI Zhun and CHEN Xudong. Deep-learning schemes for full-wave nonlinear inverse scattering problems[J]. IEEE Transactions on Geoscience and Remote Sensing, 2019, 57(4): 1849–1860. doi: 10.1109/TGRS.2018.2869221.",{"type":18,"tag":26,"props":463,"children":464},{},[465],{"type":24,"value":466},"往期回顾",{"type":18,"tag":26,"props":468,"children":469},{},[470],{"type":18,"tag":389,"props":471,"children":474},{"href":472,"rel":473},"http://mp.weixin.qq.com/s?__biz=MzkxMTM2MjMzNg==&mid=2247606710&idx=1&sn=6375a788218ccdd3f275e305bff292be&chksm=c11e31f9f669b8ef3af56d04541d885904905734f5bbe69c8c9d57dd6c91b1e9bdd93d579b75&scene=21#wechat_redirect",[393],[475],{"type":24,"value":476},"MindSpore AI科学计算系列 | 最新综述文章梳理量子到宏观尺度AI4S共性介绍",{"type":18,"tag":26,"props":478,"children":479},{},[480],{"type":18,"tag":389,"props":481,"children":484},{"href":482,"rel":483},"http://mp.weixin.qq.com/s?__biz=MzkxMTM2MjMzNg==&mid=2247605234&idx=1&sn=401b0c7d2546e1aea395cdba4466db80&chksm=c11e4bbdf669c2ab34a07da59513f19b1be187569b66acb45290e1eb9a49f8930210bc7bd377&scene=21#wechat_redirect",[393],[485],{"type":24,"value":486},"MindSpore AI科学计算系列 | 基于PhyCRNet求解时空域PDEs",{"type":18,"tag":26,"props":488,"children":489},{},[490],{"type":18,"tag":389,"props":491,"children":494},{"href":492,"rel":493},"http://mp.weixin.qq.com/s?__biz=MzkxMTM2MjMzNg==&mid=2247604729&idx=1&sn=cbe8da44dd3e9c886afde74e6f3a482a&chksm=c11e49b6f669c0a0056b1ba1ab2f2db05081af81023ad3bb209872074470e3461a8d4c969444&scene=21#wechat_redirect",[393],[495],{"type":24,"value":496},"MindSpore AI科学计算系列 | 薛定谔方程的巧妙求解，基于FermiNet高精度求解波函数",{"type":18,"tag":26,"props":498,"children":499},{},[500],{"type":18,"tag":389,"props":501,"children":504},{"href":502,"rel":503},"http://mp.weixin.qq.com/s?__biz=MzkxMTM2MjMzNg==&mid=2247601859&idx=1&sn=2698890322ae1de76749b809f48c6b13&chksm=c11e5e8cf669d79ae4d075a04a23d1ad0b75e65d4a3e72797c3950e0d3afecc572ed7cc58a1a&scene=21#wechat_redirect",[393],[505],{"type":24,"value":506},"MindSpore AI科学计算系列 | 对称性与AI for Science，游走于人工智能与科学领域的几何深度学习",{"title":7,"searchDepth":508,"depth":508,"links":509},4,[510],{"id":372,"depth":511,"text":372},2,"markdown","content:technology-blogs:zh:2775.md","content","technology-blogs/zh/2775.md","technology-blogs/zh/2775","md",1776506123177]