Testing LLMs on superconductivity research questions
Education Innovation
More from the blog
From Atari to EVE Online: Building on 15 Years of AI Research in Games
Google DeepMind partners with game studios to prototype breakthrough AI gameplay.
Paving the way for greener ammonia production
New MIT research could lead to better materials for a fossil-fuel-free process for making the chemical that's essential to fertilizer and other products.
How Much Memory Does Your Agent Actually Need?
A Blog post by IBM Research on Hugging Face
Thinking of ACE? We Can Do It with Fewer Tokens
A Blog post by IBM Research on Hugging Face
Related papers
Exception handling bugs in Python: An empirical study of root causes, fix patterns, and anti-patterns
Analysis of exception handling bugs in Python projects reveals systematic relationships between root causes and repair strategies, indicating that exception handling bugs often follow predictable patterns.
Multiscale Fresh Tea Leaves Sorting Device with Drum-axial Airflow Coupling: Design and Performance Test
To address the problems of low sorting accuracy and poor operation stability caused by physical leaf entanglement in traditional drum screening of fresh tea leaves, a multiscale fresh tea leaf sorting system with drum-axial airflow coupling based on intelligent control was designed. The axial moving distances of fresh tea leaves of different scales at wind speeds of 5, 7, and 9 m/s were calibrated through bench tests, and the optimal wind speed parameter for secondary fine screening was determined. A numerical model of the axial airflow field inside the drum was established via Fluent software, and a coupled sorting test platform was built to compare the sorting performance of the traditional pure drum screening mode and that of the coupled intelligent sorting mode. The results showed that 7 m/s is the optimal axial airflow velocity for secondary fine screening of multiscale fresh tea leaves during this test, which can realize effective back-blowing of small-scale materials and accurate screening of large-scale materials. At this velocity, the flow field is evenly distributed, and the effective thrust area highly matched the sorting demand. The average sorting efficiency of the coupled intelligent sorting mode reached 84.8%, which is 25.6% higher than that of traditional pure drum screening, with favorable sorting accuracy and operation stability. These findings can provide a theoretical basis and technical reference for the optimization, upgrading, and intelligent transformation of high-efficiency fresh tea leaf sorting equipment.
Critique of Lower-Margin Equations in Concrete Design Codes and of Their Effect on Structural Safety Software
It is generally agreed that engineered structures, whether bridges or aircraft, should be designed to have failure probability no higher than 10 − 6 per lifetime. The safety analysis of concrete structures based on the current design codes cannot guarantee meeting this goal. While sophisticated probabilistic models have been developed to deal with the randomness of loads, the uncertainty of material failure has been relegated to empirical understrength (or capacity reduction) factors. The problem is that the design equations of all design codes have traditionally been formulated as lower-margin equations, set at the lower margin of the test data cloud (which lies, depending on structure size, 25% to 40%, below the data mean, in the case of shear strength of RC beams). The load factors are applied to these lower-margin equations while the offset of the mean and the variance of the database remain buried in the code committee documents. Moreover, probabilistic modeling of the mechanics of failure processes, which determines structural strength, has been incorrectly employed, and the probability density function (pdf) required to extrapolate to 10 − 6 has been chosen arbitrarily, often as the lognormal pdf for mathematical convenience. Although the lognormal pdf may be an acceptable approximation for a database of concretes with very different strengths, it is shown to be physically impossible to model the strength distribution of one-and-the-same concrete (i.e., a concrete of the same design strength and composition). These traditional concepts have rendered the current failure probability predictions of the structural safety and reliability software for reinforced concrete structures meaningless. However, experience of many decades shows that the frequency of structural failures has not been excessive. The explanation is that many designs must have had excessive safety margins, thus becoming uneconomical, while the benefit of sophisticated commercial software applicable to the randomness of applied loads gets wasted. A sine qua non of the remedy is that the values of the coefficient of variation of the database and of the offset of the database mean from the code equation accompanying each design code equation must be revealed. This could be done in the code Commentary.
Geometric Reconstruction-Driven Load Capacity Evaluation for Locally Buckled Steel Members
Conventional contact-based inspection techniques face significant challenges in accurately modeling the complex 3D geometry of locally buckled steel members, which hinder reliable assessment of their residual load-carrying capacity. To overcome these limitations, this study proposes a method for analyzing the bearing capacity of locally buckled steel members using geometric reconstruction models, enabling precise evaluation of their load-carrying capacity. The research methodology encompasses three primary aspects: (1) model preprocessing; (2) load-bearing capacity analysis of specimens; and (3) experimental validation. Model preprocessing involves three key tasks: point cloud model reconstruction and optimization; evaluation of the effect of external factors on model accuracy; and parametric modeling with verification of geometric accuracy. The load-bearing capacity of damaged specimens was analyzed by predicting the residual capacity of the corresponding parametric models using finite element software. Finally, axial compression tests on equal-leg single-angle steel specimens were conducted to validate the accuracy of the finite element analysis results, thereby demonstrating the effectiveness of the proposed method. Key findings include: (1) the overlap ratio has the most significant influence on model accuracy; at an overlap level of 18, the comprehensive mean absolute error is below 0.005, and model-specimen similarity between the angle steel parametric model and the angle steel specimen reaches 0.99 (no significant difference at 95% confidence level); (2) among the extracted key feature data, the elastic stiffness, peak load, and peak displacement all exhibit relative errors less than 10%, while peak strain in deformation zones shows larger deviations; and (3) among the 44 statistically key characteristic data points, 41 exhibit relative errors less than 10%, confirming the method’s high reliability for practical engineering applications.