CBMM

We introduce and analyze Structured Stochastic Zeroth order Descent (S-SZD), a finite difference approach that approximates a stochastic gradient on a set of $l\le d$ orthogonal directions, where d is the dimension of the ambient space. These directions are randomly chosen and may change at each step. For smooth convex functions we prove almost sure convergence of the iterates and a convergence rate on the function values of the form $O( (d/l) k^{-c})$ for every $c<1/2$, which is arbitrarily close to the one of Stochastic Gradient Descent (SGD) in terms of number of iterations (Garrigos and Gower in Handbook of convergence theorems for (stochastic) gradient methods, arXiv:2301.11235, 2024) . Our bound shows the benefits of using l multiple directions instead of one. For non-convex functions satisfying the Polyak-Łojasiewicz condition, we establish the first convergence rates for stochastic structured zeroth order algorithms under such an assumption. We corroborate our theoretical findings with numerical simulations where the assumptions are satisfied and on the real-world problem of hyper-parameter optimization in machine learning, achieving competitive practical performance.

Recrystallization and grain growth during plate rolling are prevented by Nb addition both with the solute drag and the Nb carbide precipitation. Although a fine microstructure is achieved in the base material, welding heat completely changes the microstructure in the heat affected zone (HAZ). In this study, laboratory simulation of the coarse grain HAZ (CGHAZ) thermal cycle of double submerged arc welded linepipe was carried out using low carbon steels containing different Nb contents. Extraction residue analysis of the simulated CGHAZ samples revealed that almost all the Nb remained in solid solution. To clarify the interaction of Nb carbide dissolution and grain growth on overall simulated HAZ microstructure evolution, additional weld HAZ thermal simulations were performed. It was found that Nb carbides remain undissolved at HAZ peak temperatures up to 1200℃ and showed significant pinning effect to prevent austenite grain growth. Significant grain growth was seen after continuous fast heating to 1350℃ peak temperature, while the higher Nb added steel showed a slower overall austenite grain growth rate, suggesting that grain growth in the HAZ at higher temperature was suppressed by the combined effects of slower coarse Nb carbide dissolution providing some pinning, and the solute drag effect of higher amounts of Nb in solid solution. A pronounced retardation of longer-term isothermal grain growth was identified at 1350℃ at higher levels of solute Nb, confirming the influence of Nb solute drag on high temperature resistance to austenite grain coarsening.

Discrete inverse problems correspond to solving a system of equations in a stable way with respect to noise in the data. A typical approach to select a meaningful solution is to introduce a regularizer. While for most applications the regularizer is convex, in many cases it is neither smooth nor strongly convex. In this paper, we propose and study two new iterative regularization methods, based on a primal-dual algorithm, to regularize inverse problems efficiently. Our analysis, in the noise free case, provides convergence rates for the Lagrangian and the feasibility gap. In the noisy case, it provides stability bounds and early stopping rules with theoretical guarantees. The main novelty of our work is the exploitation of some a priori knowledge about the solution set: we show that the linear equations determined by the data can be used more than once along the iterations. We discuss various approaches to reuse linear equations that are at the same time consistent with our assumptions and flexible in the implementation. Finally, we illustrate our theoretical findings with numerical simulations for robust sparse recovery and image reconstruction. We confirm the efficiency of the proposed regularization approaches, comparing the results with state-of-the-art methods.

We propose an accurate and efficient machine learning approach for monitoring particle detectors in real-time. The goal is to assess the compatibility of incoming experimental data with a reference dataset, characterising the data behaviour under normal circumstances, via a likelihood-ratio hypothesis test. The model is based on a modern implementation of kernel methods, nonparametric algorithms that can learn any continuous function given enough data. The resulting approach is efficient and agnostic to the type of anomaly that may be present in the data. Our study demonstrates the effectiveness of this strategy on multivariate data from drift tube chamber muon detectors.

Tailings disposal in the form of diluted slurries has a tendency for particle size segregation, where coarse particles settle near the discharge point, and finer particles are carried by the water flux to more distant regions. This causes a loss of reservoir capacity due to voids between the coarser particles and increased water content in the deposit. This work aimed to evaluate the feasibility of reaching non-segregable high-density slurries with fine tailings from the niobium oreflotation process and measure its disposal parameters. The innovation is to achieve increased solids percentage in the settled deposit and to avoid particle size segregation along the slurry path with niobium tailings. The study involved physical, chemical, and mineralogical characterization and semi-pilot thickening tests to produce enough volume of underflow with different bed heights and solids flux rates. Slump, rheology, and flume tests were performed to evaluate underflow disposal characteristics. The results indicated that the thickener bed height did not significantly influence the underflow solids content, yield stress, or slump. The solids flux rate, on the other side, had a greater influence—the higher it was, the lower the solids content, yield stress, and disposal angle, along with a higher slump. In flume tests, a high density of non-segregable tailings slurry was achieved with 1.96 t/m3, corresponding to an underflow with 66.8% solids, 43.9 Pa of yield stress with 0.5 (t/h)/m2, and 0.5 m of bed height.

Recrystallization and grain growth during plate rolling are prevented by Nb addition both with the solute drag and the Nb carbide precipitation. Although a fine microstructure is achieved in the base material, welding heat completely changes the microstructure in the heat affected zone (HAZ). In this study, laboratory simulation of the coarse grain HAZ (CGHAZ) thermal cycle of double submerged arc welded linepipe was carried out using low carbon steels containing different Nb contents. Extraction residue analysis of the simulated CGHAZ samples revealed that almost all the Nb remained in solid solution. To clarify the interaction of Nb carbide dissolution and grain growth on overall simulated HAZ microstructure evolution, additional weld HAZ thermal simulations were performed. It was found that Nb carbides remain undissolved at HAZ peak temperatures up to 1200°C and showed significant pinning effect to prevent austenite grain growth. Significant grain growth was seen after continuous fast heating to 1350°C peak temperature, while the higher Nb added steel showed a slower overall austenite grain growth rate, suggesting that grain growth in the HAZ at higher temperature was suppressed by the combined effects of slower coarse Nb carbide dissolution providing some pinning, and the solute drag effect of higher amounts of Nb in solid solution. A pronounced retardation of longer-term isothermal grain growth was identified at 1350°C at higher levels of solute Nb, confirming the influence of Nb solute drag on high temperature resistance to austenite grain coarsening. Fullsize Image

Obtaining high levels of mechanical properties in steels is directly linked to the use of special mechanical forming processes and the addition of alloying elements during their manufacture. This work presents a study of a hot-rolled steel strip produced to achieve a yield strength above 600 MPa, using a niobium microalloyed HSLA steel with non-stoichiometric titanium (titanium/nitrogen ratio above 3.42), and rolled on a Steckel mill. A major challenge imposed by rolling on a Steckel mill is that the process is reversible, resulting in long interpass times, which facilitates recrystallization and grain growth kinetics. Rolling parameters whose aim was to obtain the maximum degree of microstructural refinement were determined by considering microstructural evolution simulations performed in MicroSim-SM® software and studying the alloy through physical simulations to obtain critical temperatures and determine the CCT diagram. Four ranges of coiling temperatures (525–550 °C/550–600 °C/600–650 ° C/650–700 °C) were applied to evaluate their impact on microstructure, precipitation hardening, and mechanical properties, with the results showing a very refined microstructure, with the highest yield strength observed at coiling temperatures of 600–650 °C. This scenario is explained by the maximum precipitation of titanium carbide observed at this temperature, leading to a greater contribution of precipitation hardening provided by the presence of a large volume of small-sized precipitates. This paper shows that the combination of optimized industrial parameters based on metallurgical mechanisms and advanced modeling techniques opens up new possibilities for a robust production of high-strength steels using a Steckel mill. The microstructural base for a stable production of high-strength hot-rolled products relies on a consistent grain size refinement provided mainly by the effect of Nb together with appropriate rolling parameters, and the fine precipitation of TiC during cooling provides the additional increase to reach the requested yield strength values.

A new, 19 π‐delocalized electrons planar Blatter radical building block was developed and used to obtain paramagnetic bent‐core liquid crystals. The mesogens were investigated by optical, thermal, powder XRD and DFT methods in the pure form and as binary mixtures. Comparison of their properties with those of the classical Blatter radical analogues revealed that planarization of the central angular element results in a significantly higher stability of the mesophases and increased molecular organization suitable for the formation of ordered banana and columnar mesophases with tighter π–π interactions. These results indicate access to a new, potentially rich class of functional paramagnetic soft materials.

Today’s market for API pipeline steels requires higher strength and fracture toughness performance in heavier gauges and larger pipe diameters. Meeting basic average mechanical properties of strength, TCVN and DWTT performance has been achieved by many with newer powerful plate mills and cooling systems. However, stability/standard deviations of plate/pipe strength and toughness along with pipe seam/girth weld toughness continues to be challenging for many. Current focus has been on pipe seam and girth weld HAZ toughness stability which has proven to be difficult to achieve. Optimization of the alloy and processing design are critical components of creating the desired final through thickness microstructure, grain size/homogeneity and a volume fraction of proper size Nb precipitates in the hot rolled condition and corresponding final base and weld metal ductility properties. Optimized processing of two Nb levels, 0.055% and 0.075%, Nanjing Iron and Steel Company successfully improved base metal plate strength and toughness stability that translated into improved pipe strength and toughness stability including pipe seam/girth welds HAZ toughness. The optimized 0.075% Nb girth weld HAZ fusion line average was 339 J @ −80 °C with a 9.3 J standard deviation while the optimized 0.055% Nb girth weld HAZ fusion line was 307 J @ −80 °C with a 55.2 J standard deviation.

Characterizing the function spaces corresponding to neural networks can provide a way to understand their properties. In this paper we discuss how the theory of reproducing kernel Banach spaces can be used to tackle this challenge. In particular, we prove a representer theorem for a wide class of reproducing kernel Banach spaces that admit a suitable integral representation and include one hidden layer neural networks of possibly infinite width. Further, we show that, for a suitable class of ReLU activation functions, the norm in the corresponding reproducing kernel Banach space can be characterized in terms of the inverse Radon transform of a bounded real measure, with norm given by the total variation norm of the measure. Our analysis simplifies and extends recent results in [45], [36], [37].

Optimization in machine learning typically deals with the minimization of empirical objectives defined by training data. The ultimate goal of learning, however, is to minimize the error on future data (test error), for which the training data provides only partial information. In this view, the optimization problems that are practically feasible are based on inexact quantities that are stochastic in nature. In this paper, we show how probabilistic results, specifically gradient concentration, can be combined with results from inexact optimization to derive sharp test error guarantees. By considering unconstrained objectives, we highlight the implicit regularization properties of optimization for learning.

Base excision repair (BER) removes damaged bases by generating single‐strand breaks (SSBs), gap‐filling by DNA polymerase β (POLβ), and resealing SSBs. A base‐damaging agent, MMS is widely used to study BER. BER increases cellular tolerance to MMS, anti‐cancer base‐damaging drugs, Temozolomide, Carmustine, and Lomustine, and to clinical poly(ADP ribose)polymerase (PARP) poisons, Olaparib and Talazoparib. The poisons stabilize PARP1/SSB complexes, inhibiting access of BER factors to SSBs. PARP1 and XRCC1 collaboratively promote SSB resealing by recruiting POLβ to SSBs, but XRCC1‐/‐ cells are much more sensitive to MMS than PARP1‐/‐ cells. We recently report that the PARP1 loss in XRCC1‐/‐ cells restores their MMS tolerance and conclude that XPCC1 facilitates the release of PARP1 from SSBs by maintaining its autoPARylation. We here show that the PARP1 loss in XRCC1‐/‐ cells also restores their tolerance to the three anti‐cancer base‐damaging drugs, although they and MMS induce different sets of base damage. We reveal the synthetic lethality of the XRCC1‐/‐ mutation, but not POLβ‐/‐, with Olaparib and Talazoparib, indicating that XRCC1 is a unique BER factor in suppressing toxic PARP1/SSB complex and can suppress even when PARP1 catalysis is inhibited. In conclusion, XRCC1 suppresses the PARP1/SSB complex via PARP1 catalysis‐dependent and independent mechanisms.

Among post-translational modifications of proteins, ADP-ribosylation has been studied for over fifty years, assigning to this post-translational modification (PTM) a large set of functions, including DNA repair, transcription and cell signaling. This review presents an update on the function of a large set of enzyme writers, the readers that are recruited by the modified targets, and the erasers that reverse the modification to the original amino acid residue, removing the covalent bonds formed. In particular, the review provides details on the involvement of the enzymes performing monoADP-ribosylation/ polyADP-ribosylation (MAR/PAR) cycling in cancers. Of note, there is potential for application of the inhibitors developed for cancer also in the therapy of non-oncological diseases such as the protection against oxidative stress, suppression of inflammatory responses, and the treatment of neurodegenerative diseases. This field of studies is not concluded, since novel enzymes are being discovered at a rapid pace.

Annually over 500 million tons worldwide of flat and long commodity grade structural steel products are produced for applications in the construction sector. Major costs to produce these commodity grade structural steels lies in alloys, energy and depending on region labor. Production costs for these commodity grades, which typically have low margins for profits, continue to rise worldwide driven by ferro alloy costs. Improved operational efficiencies (productivity, reduced energy consumption, improved melt to finished yields, reduced consumable consumption, etc.) can be realized along with lowered alloy costs with a proper understanding of a new metallurgical strategy to alloy design. These operational efficiencies along with cost savings can be accomplished with proper alloy design in conjunction with the mills processing capabilities to achieve the desired end metallurgy/mechanical properties. Requirements of strength and ductility for any given structural steel microstructure are obtained from three metallurgical mechanisms or "building blocks": a) grain size refinement, b) solid solution and c) precipitation. Overall operation costs including alloy costs can be minimized if better engineering of these contributions can be realized. The correct use of these factors brings improved process/mechanical property stability. Use of practical metallurgical modeling tools along with mill data to determine process control capabilities can also assist in optimization of the metallurgical designs for cost effective structural steel production.

Abrasion resistant steel plates with hardness equal to or greater than 360 BHN and high strength structural steel plates/coils with yield strength requirements equal to or greater than 690 MPa are typically produced by the quench and tempering (Q&T) process. The Q&T microstructures thus produced by either reheating, quench or tempering (RQT) or direct quench and/or tempering (DQ and/or DQT) lead to very high hardness and strength levels as a result of tempered martensite/bainite or auto-tempered and/or lower bainite microstructures. Using these steels in the earth moving, mining equipment and in the general materials processing industry, in addition to strength and toughness other major requirements are weldability (low CE), good elongation and formability. Hence, to achieve optimum balance of properties in these steels, it is very important to understand the microstructural evolution from the interaction of thermo-mechanical control rolling of plate/coil followed by the RQT/DQ/DQT processes. It is shown in the present paper that with careful design of the steel chemistry (C, Mn, Si, Cr, Mo) while utilising optimized microalloying, especially niobium, titanium and boron complemented by proper rolling, cooling, and Q&T process design, uniform cross sectional microstructure and optimum balance of strength, hardness, toughness, ductility and weldability are achieved.

The objective of this work was to observe the significant factors for the dehydration reaction of xylose to furfural and to optimize the processes using experimental design. The studied variables were temperature, time, initial percentage of xylose mass, and catalyst/xylose ratio. Temperature and initial percentage of xylose mass were considered statistically significant, while the maximum point for furfural selectivity was at 160 °C and 2% of initial xylose mass. Using niobic acid and niobium phosphate (1:1) (NbP/NbA), 44.05% xylose conversion and 74.71% furfural selectivity were obtained. The results showed that the mixture of catalysts with Brönsted acid and Lewis acid sites improved the selectivity of furfural from the xylose dehydration reaction. NbP/NbA catalysts were very stable under the investigated condition after 5 continuous recycles.