Repository Collection: null

Optimization of MuMax3 by Using Claude Code: A CUDA-Graph-Based Case Study in AI-Assisted Performance Engineering

2026-09-29T03:10:15Z

Title: Optimization of MuMax3 by Using Claude Code: A CUDA-Graph-Based Case Study in AI-Assisted Performance Engineering Author(s): You, Chun-Yeol Abstract: MuMax3 is a widely used open-source GPU-accelerated micromagnetic simulator whose computational core - CUDA kernels and cuFFT-based demagnetization convolutions - has changed little since its original release. We report a case study in which an agentic large-language-model coding assistant (Claude Code, Anthropic) was used, under continuous human supervision, to profile and optimize this mature CUDA/Go codebase. Profiling with NVIDIA Nsight Systems revealed that for small and medium grids (64 & times;64 & times;1 to 256 & times;256 & times;1 cells), 75-79 % of step time is spent in CPU-side cuLaunchKernel driver calls rather than in GPU computation, because every solver step launches roughly 27 kernels sequentially on a single CUDA stream. Building on this finding, we implemented a "split-graph" execution strategy: the time-step-independent torque-evaluation kernel sequences of each solver stage are captured once as CUDA Graphs and replayed via cudaGraphLaunch, while the time-step-dependent update, error estimate, and adaptive-step-size logic remain ordinary stream-ordered calls. The optimization is exposed transparently through Run()/Steps(), guarded by a compatibility check (constant excitation, zero thermal field, no custom field terms, time-independent material parameters, mesh size below a tunable threshold) that falls back silently to the original code path when violated. Across five solvers (Heun, RK23, RK45DP, RK56, Backward Euler), the optimization yields up to 5.4 & times; throughput for a 64 & times;64 & times;1 grid with a fixed time step, 2.4-3.6 & times; for adaptive time-stepping, decreasing smoothly to approximate to 1.0 & times; near 10(6) cells, essentially independent of which physical field terms (exchange, anisotropy, DMI) are active. All results were verified bit-for-bit identical (ndiff=0) against the unmodified code, and the full 176-script mumax3 regression suite passes with zero failures. We discuss the workflow itself - including three episodes in which the assistant autonomously diagnosed and corrected its own defects - as a template for AI-assisted optimization of legacy scientific HPC codes.

Deterministic spin-orbit torque switching of epitaxial ferrimagnetic insulator with perpendicular magnetic anisotropy fabricated by on-axis magnetron sputtering

2026-02-10T18:01:23Z

Title: Deterministic spin-orbit torque switching of epitaxial ferrimagnetic insulator with perpendicular magnetic anisotropy fabricated by on-axis magnetron sputtering Author(s): Ngaloy, Roselle T.; Yamashita, Naoto; Zhao, Bing; Kim, Soojung; Yamashita, Kohei; Cools, Ivo P.C.; Agusutrisno, Marlis Nurut; Lee, Soobeom; Kurokawa, Yuichiro; You, Chun-Yeol; Yuasa, Hiromi; Dash, Saroj Prasad Abstract: Current-induced switching of magnetization states in ferromagnet/spin-orbit material heterostructures has attracted significant attention, driven by the increasing need for low power consumption and a more efficient mechanism for magnetization switching. However, current shunting for the used metallic ferromagnets remains challenging in achieving low switching current densities. Thulium iron garnet, Tm3Fe5O12 (TmIG), is promising for such devices as it exhibits strong perpendicular magnetic anisotropy (PMA) and fast magnetization dynamics. However, there still remains a technological challenge in the growth of TmIG films using industry-compatible magnetron sputtering in a simple on-axis geometry for spintronic device applications. Here, we demonstrated the spin-orbit torque (SOT) magnetization switching of TmIG thin film grown by on-axis radio-frequency magnetron sputtering. Robust and deterministic SOT magnetization switching is achieved using TmIG/Pt heterostructures at a current density as low as 0.7×1011A/m2. Anomalous Hall and second harmonic Hall measurements were performed to quantify effective spin-orbit fields. The effective field inducing damping-like torque is estimated to be 21±1 Oe per 107A/cm2, higher than previous reports. These findings show a growth method for ferrimagnetic insulators with strong PMA in industry-compatible on-axis sputtering methods and its utilization for achieving energy-efficient SOT non-volatile memory applications. © 2025 Elsevier B.V., All rights reserved.

브릴루앙 광 산란 실험을 이용한 버퍼층의 변화에 따른 30 nm 두께의 Ni80Fe20 박막의 자성특성연구

2026-04-15T08:13:32Z

Title: 브릴루앙 광 산란 실험을 이용한 버퍼층의 변화에 따른 30 nm 두께의 Ni80Fe20 박막의 자성특성연구 Author(s): 김정원; 김수정; 민성현; 유천열; 조재훈 Abstract: 본 연구에서는 다양한 버퍼층 구조가 Ni20Fe20(30 nm) 박막의 자성특성에 미치는 영향을 분석하였다. Si/SiO2 기판 위에 DC 마그네트론 스퍼터링법을 이용하여 Ta (5 nm), Ta (5 nm)/Ru (5 nm)/Ta (5 nm), Ta (5 nm)/Ru (10 nm)/Ta (5 nm), Ta (5 nm)/Ru (10 nm)/Ta (10 nm)의 버퍼층을 각각 형성하고, 그 위에 30 nm 두께의 Ni20Fe20 박막을 증착하였다. 버퍼층의 형태학적 특성은 원자힘현미경(AFM)으로, 결정 구조는 X선 회절(XRD)로 분석하였다. 그 결과, 모든 버퍼층의 표면 거칠기 편차는 0.03 nm 이하로 Ni의 원자 단층 두께(0.2 nm)보다 작아, Ni20Fe20 박막의 미세구조에는 미미한 영향을 미치는 것으로 판단되었다. 자화 동역학 특성은 브릴루앙 광산란(Brillouin light scattering; BLS)을 이용하여 측정하였으며, Damon-Eshbach 모드와 첫번째 덩어리 모드로부터 포화자화(Mₛ)와 교환 뻣뻣함 상수(Aₑₓ)를 산출하였다. 분석 결과, Mₛ 값은 버퍼층에 따라 큰 차이를 보이지 않았으며, Aₑₓ 값은 버퍼층 구조에 따라 최대 약 3%의 차이가 나지만, 결정된 값들의 표준편차는 0.2 pJ/m 정도로 최소 제곱법으로 결정된 값들의 오차와 큰 차이가 없다. Aex 값 또한 버퍼층의 적층 형태에 따라 큰 차이를 보이지 못하였다. 이는 버퍼층의 결정학적 특성이 Ni80Fe20 박막의 덩어리 스핀파 모드 특성에는 영향을 끼치지 않는 것으로 생각된다.

Revealing the Altermagnetism in Hematite via XMCD Imaging and Anomalous Hall Electrical Transport

2025-12-03T18:01:07Z

Title: Revealing the Altermagnetism in Hematite via XMCD Imaging and Anomalous Hall Electrical Transport Author(s): Galindez-Ruales, Edgar; Gonzalez-Hernandez, Rafael; Schmitt, Christin; Das, Shubhankar; Fuhrmann, Felix; Ross, Andrew; Golias, Evangelos; Akashdeep, Akashdeep; Luenenbuerger, Laura; Baek, Eunchong; Yang, Wanting; Smejkal, Libor; Krishna, Venkata; Jaeschke-Ubiergo, Rodrigo; Sinova, Jairo; Rothschild, Avner; You, Chun-Yeol; Jakob, Gerhard; Klaeui, Mathias Abstract: Altermagnets are a class of magnetic materials that exhibit unconventional transport properties, such as an anomalous Hall effect (AHE), despite having compensated sublattice magnetic moments. In this study, fundamental experimental evidence of the altermagnetic nature of hematite (alpha-Fe2O3), is reported combining electrical transport with advanced X-ray photoemission electron microscopy (XPEEM) imaging with linear and circular dichroism contrast. These measurements directly visualize the N & eacute;el vector's coupling to the crystal orientation, confirming hematite's altermagnetic order and its symmetry-driven transport behavior. The transport measurements reveal an anisotropic AHE with a pronounced crystal orientation dependence, including a sign inversion for specific N & eacute;el vector alignments. Supported by first-principles theoretical calculations, how the interplay between collinear spin and crystal symmetry breaking drives the observed AHE is explained. These findings establish hematite as an altermagnet, paving the way for experimental identification of altermagnetic materials and their integration into spintronic technologies.