SearcharxivSearch

arXiv subjects

Michael Tran

Publications and source records attributed to Michael Tran.

9 recordsLinked to original sources

Characterizing the Quality Profile of AI-Generated C++ in Production

The widespread integration of AI coding assistants offers undeniable boosts to engineering velocity. Yet, recent studies point to a growing trade-off, revealing persistent challenges with code quality and maintainability. Industry leaders, including frontier AI labs, echo these concerns. As large language models are increasingly relied upon to author production code, understanding their impact on shipped software quality has become a critical priority. However, assessing these effects in industrial workflows remains difficult due to observability barriers. We study the impact of AI-generated code on production quality within a large enterprise operating global products relied upon by billions of users daily. Driven by this scale and user trust, the organization values code quality and has built thorough observability for every line of code deployed into production, enabling us to overcome measurement barriers to assess these effects. This study presents a large-scale empirical analysis of AI-generated C++ code from April 2025 to April 2026, tracking 3.52 million code changes across this enterprise's brownfield codebase. The core purpose is to understand the quality, performance, and maintenance characteristics of AI-generated code compared to human-written code in a production environment at scale. We find that AI-generated C++ code has a distinct quality profile, showing higher rates of interface and coupling burdens, copy and allocation overheads, and a reliance on explicit loops over optimized standard APIs. These issues translate into tangible downstream costs, including increased review effort and a 5-8% increase in compute resource consumption. However, we demonstrate that providing models with targeted, taxonomy-informed feedback can mitigate these effects, leading to an 11.1% reduction in targeted static analysis warnings and improved computational efficiency.

cs.SE

Implementation of a Binary Neural Network on a Passive Array of Magnetic Tunnel Junctions

The increasing scale of neural networks and their growing application space have produced demand for more energy- and memory-efficient artificial-intelligence-specific hardware. Avenues to mitigate the main issue, the von Neumann bottleneck, include in-memory and near-memory architectures, as well as algorithmic approaches. Here we leverage the low-power and the inherently binary operation of magnetic tunnel junctions (MTJs) to demonstrate neural network hardware inference based on passive arrays of MTJs. In general, transferring a trained network model to hardware for inference is confronted by degradation in performance due to device-to-device variations, write errors, parasitic resistance, and nonidealities in the substrate. To quantify the effect of these hardware realities, we benchmark 300 unique weight matrix solutions of a 2-layer perceptron to classify the Wine dataset for both classification accuracy and write fidelity. Despite device imperfections, we achieve software-equivalent accuracy of up to 95.3 % with proper tuning of network parameters in 15 x 15 MTJ arrays having a range of device sizes. The success of this tuning process shows that new metrics are needed to characterize the performance and quality of networks reproduced in mixed signal hardware.

cs.ET

Origin of the resistance-area product dependence of spin transfer torque switching in perpendicular magnetic random access memory cells

We report on an experimental study of current induced switching in perpendicular magnetic random access memory (MRAM) cells with variable resistance-area products (RAs). Our results show that in addition to spin transfer torque (STT), current induced self-heating and voltage controlled magnetic anisotropy also contribute to switching and can explain the RA dependencies of switching current density and STT efficiency. Our findings suggest that thermal optimization of perpendicular MRAM cells can result in significant reduction of switching currents.

cond-mat.mes-hall

Magnetization dynamics and its scattering mechanism in thin CoFeB films with interfacial anisotropy

Studies of magnetization dynamics have incessantly facilitated the discovery of fundamentally novel physical phenomena, making steady headway in the development of magnetic and spintronics devices. The dynamics can be induced and detected electrically, offering new functionalities in advanced electronics at the nanoscale. However, its scattering mechanism is still disputed. Understanding the mechanism in thin films is especially important, because most spintronics devices are made from stacks of multilayers with nanometer thickness. The stacks are known to possess interfacial magnetic anisotropy, a central property for applications, whose influence on the dynamics remains unknown. Here, we investigate the impact of interfacial anisotropy by adopting CoFeB/MgO as a model system. Through systematic and complementary measurements of ferromagnetic resonance (FMR), on a series of thin films, we identify narrower FMR linewidths at higher temperatures. We explicitly rule out the temperature dependence of intrinsic damping as a possible cause, and it is also not expected from existing extrinsic scattering mechanisms for ferromagnets. We ascribe this observation to motional narrowing, an old concept so far neglected in the analyses of FMR spectra. The effect is confirmed to originate from interfacial anisotropy, impacting the practical technology of spin-based nanodevices up to room temperature.

cond-mat.mtrl-sci

Influence of Ta insertions on the magnetic properties of MgO/CoFeB/MgO films probed by ferromagnetic resonance

We show by vector network analyzer ferromagnetic resonance measurements that low Gilbert damping α down to 0.006 can be achieved in perpendicularly magnetized MgO/CoFeB/MgO thin films with ultra-thin insertions of Ta in the CoFeB layer. While increasing the number of Ta insertions allows thicker CoFeB layers to remain perpendicular, the effective areal magnetic anisotropy does not improve with more insertions, and also comes with an increase in α.

cond-mat.mtrl-sci

Asymmetry in effective fields of spin-orbit torques in Pt/Co/Pt stacks

Measurements of switching via spin-orbit coupling (SOC) mechanisms are discussed for a pair of inverted Pt/Co/Pt stacks with asymmetrical Pt thicknesses. Taking into account the planar Hall effect contribution, effective fields of spin-orbit torques (SOT) are evaluated using lock-in measurements of the first and second harmonics of the Hall voltage. Reversing the stack structure leads to significant asymmetries in the switching behavior, including clear evidence of a nonlinear current dependence of the transverse effective field. Our results demonstrate potentially complex interplay in devices with all-metallic interfaces utilizing SOT.

cond-mat.mes-hall

Effects of lateral device size and material properties on the ferromagnetic resonance response of spinwave eigen-modes in magnetic devices

We analyze the effects of lateral device size and magnetic material parameters on the ferromagnetic resonance (FMR) response. Results presented are directly relevant to widely used FMR experimental techniques for extracting magnetic parameters from thin films, the results of which are often assumed to carry over to corresponding nanometer-sized patterned devices. We show that there can be significant variation in the FMR response with device size, and that the extent of the variation depends on the magnetic material properties. This explains, for example, why different experiments along these lines have yielded different size-dependent trends from damping measurements. Observed trends with increasing size and different material parameters are explained through the evolution of three distinct eigen-modes, demonstrating the respective roles of demagnetization and exchange. It is also shown that there is a crossover of dominant eigen-modes in the response signal, accompanied by conjugating edge-type modes, leading to evident effects in measured linewidth and damping. Among the sizes considered, in higher saturation magnetization, we observe as much as a 40% increase in apparent damping, due solely to device size variation.

cond-mat.mes-hall

Optical properties of Bi2Te2Se at ambient and high pressure

The temperature dependence of the complex optical properties of the three-dimensional topological insulator Bi2Te2Se is reported for light polarized in the a-b planes at ambient pressure, as well as the effects of pressure at room temperature. This material displays a semiconducting character with a bulk optical gap of 300 meV at 295 K. In addition to the two expected infrared-active vibrations observed in the planes, there is additional fine structure that is attributed to either the removal of degeneracy or the activation of Raman modes due to disorder. A strong impurity band located at 200 cm^{-1} is also observed. At and just above the optical gap, several interband absorptions are found to show a strong temperature and pressure dependence. As the temperature is lowered these features increase in strength and harden. The application of pressure leads to a very abrupt closing of the gap above 8 GPa, and strongly modifies the interband absorptions in the mid-infrared spectral range. While ab initio calculations fail to predict the collapse of the gap, they do successfully describe the size of the band gap at ambient pressure, and the magnitude and shape of the optical conductivity.

cond-mat.mtrl-sci

Strain-Control of the magnetic anisotropy in (Ga,Mn)(As,P) ferromagnetic semiconductor layers

A small fraction of phosphorus (up to 10 %) was incorporated in ferromagnetic (Ga,Mn)As epilayers grown on a GaAs substrate. P incorporation allows reducing the epitaxial strain or even change its sign, resulting in strong modifications of the magnetic anisotropy. In particular a reorientation of the easy axis toward the growth direction is observed for high P concentration. It offers an interesting alternative to the metamorphic approach, in particular for magnetization reversal experiments where epitaxial defects stongly affect the domain wall propagation.

cond-mat.mtrl-sci