Search for: All records

MG-SpaIR is a training-data-free framework for restoring a clean image from a single observation corrupted by a mixture of blur, downsampling, noise, and missing pixels. Building on implicit neural representations (INRs), we introduce a multi-grade residual hierarchy that progressively refines the reconstruction from low to high spatial frequencies across grades, improving representational fidelity and mitigating spectral limitations. To stabilize reconstruction optimization and suppress INR-induced artifacts, we further propose an explicit sparse proximal regularization (e.g.,$$\ell _0$$ ${\ell }_0$ type) applied directly in the high-resolution image domain, which discourages spurious high-frequency patterns while preserving sharp structures. The resulting optimization is solved efficiently via a multi-grade proximal alternating scheme, and we establish convergence guarantees for the associated updates under standard regularity conditions. Experiments on mixed-degradation benchmarks demonstrate that MG-SpaIR consistently outperforms strong training-data-free baselines such as Deep Image Prior, providing a stable, interpretable, and data-efficient alternative to conventional learning-based restoration methods. 

Abstract The Colorado River supplies water to 40 million people, 4.5 million acres of irrigated land, seven US states, over two dozen federally recognized and sovereign tribes and Mexico. River discharge is strongly controlled by snowpack accumulation and melt in the headwater regions, making upper-basin snow dynamics a primary driver of water availability throughout the basin. Therefore, changes in snow water equivalent (SWE) are critical for water resources planning in the Upper Colorado River Basin (UCRB), yet there is considerable uncertainty even in historical SWE trends. We compare historical UCRB SWE trends across various gridded snow products, including new Weather Research and Forecasting simulations at 45, 9, and 3 km resolution, against snow telemetry station data. We find that SWE trends are systematically amplified with higher model resolution and that high resolution (<10 km) is needed to accurately capture not only SWE amounts but also historical SWE declines across high-elevation mountains. Trend amplification primarily results from higher resolution simulations having higher snowfall due to stronger orographic forcing, an effect further modulated by synoptics. This is confirmed in a set of alternative historical realizations from a downscaled climate model large ensemble, which, despite a tendency for SWE declines, highlight the important role of internal variability in historical UCRB SWE trends. These results demonstrate that model resolution plays an important role in shaping assessments of snow-related climate change, with implications for a wide range of impact studies. 

We present Apache Flink 2.0, an evolution of the popular stream processing system's architecture that decouples computation from state management. Flink 2.0 relies on a remote distributed file system (DFS) for primary state storage and uses local disks as a secondary cache, with state updates streamed continuously and directly to the DFS. To address the latency implications of remote storage, Flink 2.0 incorporates an asynchronous runtime execution model. Furthermore, Flink 2.0 introduces ForSt, a novel state store featuring a unified file system that enables faster and lightweight checkpointing, recovery, and reconfiguration with minimal intrusion to the existing Flink runtime architecture. Using a comprehensive set of Nexmark benchmarks and a large-scale stateful production workload, we evaluate Flink 2.0's large-state processing, checkpointing, and recovery mechanisms. Our results show significant performance improvements and reduced resource utilization compared to the baseline Flink 1.20 implementation. Specifically, we observe up to 94% reduction in checkpoint duration, up to 49× faster recovery after failures or a rescaling operation, and up to 50% cost savings. 

Abstract A rapid warming trend was observed in the upper 2,000 m of the subtropical North Atlantic in 2011–2021, with significant heat accumulation not only in the upper ocean layers (0–300 m, 300–700 m) but also in the mid‐depth layer (700–2,000 m). Warming in the mid‐depth layer accounted for approximately 40% of the upper‐2000 m warming, exceeding the global average in this layer. Our analysis indicates that warming in the upper 700 m was primarily driven by isopycnal heave changes, closely linked to anomalous anticyclonic atmospheric circulation. In contrast, warming in the mid‐depth layer was mainly attributed to spice variability, influenced by advection and water mass changes in both the subtropical and subpolar regions. Overall, this study highlights the importance of mid‐depth ocean heat uptake in shaping the ocean heat content variability in the subtropical North Atlantic and reveals the combined impact of local and remote processes. 

The Neurospora crassa genome has a gene cluster for the synthesis of galactosaminogalactan (GAG). The gene cluster includes the following: (1) UDP-glucose-4-epimerase to convert UDP-glucose and UDP-N-acetylglucosamine to UDP-galactose and UDP-N-acetylgalactosamine (NCU05133), (2) GAG synthase for the synthesis of an acetylated GAG (NCU05132), (3) GAG deacetylase (/NCW-1/NCU05137), (4) GH135-1, a GAG hydrolase with specificity for N-acetylgalactosamine-containing GAG (NCU05135), and (5) GH114-1, a galactosaminidase with specificity for galactosamine-containing GAG (NCU05136). The deacetylase was previously shown to be a major cell wall glycoprotein and given the name of NCW-1 (non-GPI anchored cell wall protein-1). Characterization of the polysaccharides found in the growth medium from the wild type and the GAG synthase mutant demonstrates that there is a major reduction in the levels of polysaccharides containing galactosamine and N-acetylgalactosamine in the mutant growth medium, providing evidence that the synthase is responsible for the production of a GAG. The analysis also indicates that there are other galactose-containing polysaccharides produced by the fungus. Phenotypic characterization of wild-type and mutant isolates showed that deacetylated GAG from the wild type can function as an adhesin to a glass surface and provides the fungal mat with tensile strength, demonstrating that the deacetylated GAG functions as an intercellular adhesive. The acetylated GAG produced by the deacetylase mutant was found to function as an adhesive for chitin, alumina, celite (diatomaceous earth), activated charcoal, and wheat leaf particulates. 

A burst buffer is a common method to bridge the performance gap between the I/O needs of modern supercomputing applications and the performance of the shared file system on large-scale supercomputers. However, existing I/O sharing methods require resource isolation, offline profiling, or repeated execution that significantly limit the utilization and applicability of these systems. Here we present ThemisIO, a policy-driven I/O sharing framework for a remote-shared burst buffer: a dedicated group of I/O nodes, each with a local storage device. ThemisIO preserves high utilization by implementing opportunity fairness so that it can reallocate unused I/O resources to other applications. ThemisIO accurately and efficiently allocates I/O cycles among applications, purely based on real-time I/O behavior without requiring user-supplied information or offline-profiled application characteristics. ThemisIO supports a variety of fair sharing policies, such as user-fair, size-fair, as well as composite policies, e.g., group-then-user-fair. All these features are enabled by its statistical token design. ThemisIO can alter the execution order of incoming I/O requests based on assigned tokens to precisely balance I/O cycles between applications via time slicing, thereby enforcing processing isolation. Experiments using I/O benchmarks show that ThemisIO sustains 13.5--13.7% higher I/O throughput and 19.5--40.4% lower performance variation than existing algorithms. For real applications, ThemisIO significantly reduces the slowdown by 59.1--99.8% caused by I/O interference.