XTC (China)

Large Language Models (LLMs) excel in reasoning tasks requiring a single correct answer, but they perform poorly in multi-solution tasks that require generating comprehensive and diverse answers. We attribute this limitation to reasoning overconfidence: a tendency to express undue certainty in an incomplete solution set. To examine the effect, we introduce MuSoBench, a benchmark of multi-solution problems. Experiments show that the conventional short chain-of-thought (Short-CoT) prompting paradigm exhibits pronounced overconfidence, whereas the emerging long chain-of-thought (Long-CoT) approach mitigates it through iterative exploration and self-reflection. We further characterise observable behaviours and influential factors. To probe the underlying cause, we propose the cognitive-rigidity hypothesis, which posits that overconfidence arises when the reasoning process prematurely converges on a narrow set of thought paths. An attention-entropy analysis offers preliminary support for this view. These findings provide tools for assessing the completeness of LLM reasoning and highlight the need to move evaluation beyond single-answer accuracy toward comprehensive exploration.

This paper investigates the mechanical structural design of semiconductor probe stations. It presents the overall system architecture, functional module integration, and optimized layout schemes, including the selection and calculation of moving components and precision-retention mechanisms. A dedicated test platform is established to evaluate system performance. Experimental results demonstrate that a simulation-driven design approach enables the achievement of high-precision performance, effectively reduces thermal drift, and satisfies the testing requirements of semiconductor wafers. The proposed design provides a valuable reference for the research and development of similar semiconductor testing equipment.

Building energy conservation and noise control technologies are key technologies in construction engineering under the background of the new era, which are of great significance to the modernization of urban construction and the sustainable development of the construction industry. With the progress of social economy and the improvement of people’s living standards, urban noise pollution and energy issues have attracted increasing attention. Based on the analysis of the demands for building energy conservation and noise control, this paper proposes targeted application strategies for energy conservation and noise control technologies, providing references for construction engineering.

The displacement platform imposes stringent positioning-accuracy requirements in applications such as cross-scale assembly. However, conventional semi-closed-loop control based on the motor encoder is constrained by nonlinear transmission effects of the ball screw—primarily backlash and friction—and therefore struggles to meet high-precision demands. To address this issue, this paper proposes a full closed-loop control system based on the EtherCAT fieldbus. The system uses the built-in motor encoder as the inner loop and an external high-precision optical linear scale as the outer loop to compensate transmission errors. We present a detailed design and comparison of two control strategies: a semi-closed-loop scheme based on Profile Position (PP) mode, and a full closed-loop scheme that combines PP mode with linear-scale feedback. Experimental results indicate that the proposed full closed-loop approach significantly improves both absolute positioning accuracy and repeatability, providing an effective high-performance solution for precision motion control.

Abstract Robust short-circuit (SC) withstand capability is crucial for the practical applications of SiC metal-oxide-semiconductor field-effect transistor (MOSFETs). The failure mechanisms of SiC MOSFETs vary depending on the applied conditions. In this paper, the SC failure process of 600 V SiC MOSFETs at a low gate voltage is investigated, and an unreported failure mechanism is uncovered: a cascading failure mechanism initiated by gate oxide (GOX) rupture, followed by polysilicon (Poly) thermal expansion, and culminating in Pre-Metal Dielectric (PMD) fracture. Electro–thermal–mechanical simulations establish that defect generation likely precipitates GOX breakdown, and the degradation of Poly and PMD stems from high temperature and thermomechanical stress. This failure frequently occurs under conditions of either a thin GOX or poor GOX quality. Experiments show that improving GOX quality effectively increases the SC time.