Our Work

County-level indices of the intensity of China's 1959–1961 famine (cohort size shrinkage index, CSSI), computed from the IPUMS International 1% samples of the 1982, 1990 and 2000 population cens…

County-level indices of the intensity of China's 1959–1961 famine (cohort size shrinkage index, CSSI), computed from the IPUMS International 1% samples of the 1982, 1990 and 2000 population censuses, with prefecture- and province-level indices, a crosswalk from IPUMS county codes to national administrative codes (GB/T 2260, 1990), and the analysis code of the manuscript "Prenatal exposure to the Chinese famine of 1959–1961 and schizophrenia-related disability: a county-level gradient analysis of two national surveys" (Yihe Wangding and Cuntong Wang). Contents: data/ (county, prefecture and province indices with standard errors and empirical-Bayes estimates; persons per county and birth year in the 1990 sample; a summary of precision and reliability; the IPUMS–GB/T 2260 crosswalk); code/ (scripts s02–s32; CODE_INDEX.md gives the purpose and status of each script); sources/ (Ministry of Civil Affairs county code tables for 1987, 1990 and 2006); README.md; MANIFEST_sha256.txt. The indices and the crosswalk can be regenerated with the code from an IPUMS International extract (not redistributed; available from IPUMS International) and public code tables. The survey analyses of the paper cannot be reproduced from this package: they use the 1987 and 2006 National Sample Surveys on Disability, which were used under a data use agreement and are not included. The package contains no individual-level data and no files derived from those surveys. During peer review the files are restricted; editors and reviewers can access them through the link provided with the submission. Version 1.3 (2026-10-04). Changes from version 1.2: scripts s06 and s26 are corrected (the shrinkage target and the prefecture index of Shanghai's sample counties are now found through the IPUMS prefecture of the linked census county), and all survey analyses of the paper were re-run; the analysis scripts now write unrounded estimates, which the tables of the paper round once. No data file was changed.

A faster matrix-free product, zero-allocation Polyester threading, assembly of a grid-function coefficient and of πₕ across two meshes on a device space, and a set of host assembly fixes (milestones v…

A faster matrix-free product, zero-allocation Polyester threading, assembly of a grid-function coefficient and of πₕ across two meshes on a device space, and a set of host assembly fixes (milestones v3.21.0 to v3.23.0; v3.21.0 and v3.22.0 were not tagged on their own). This minor release has two breaking changes, listed first. Breaking MatrixFreeOperator gains a type parameter (7fe7b7de) The operator caches per-axis geometry for separable terms, and its type carries the cache's type. Code that names the operator's full type, rather than MatrixFreeOperator, needs the extra parameter. getindex and jacobi_preconditioner on the operator now require a Number element type (aa3101ab). A linear form's πₕ(uₕ) source refuses a moved mesh (d8080b79, 797f60e1) The source is read at every fill. A source whose mesh moved since the form was built now throws instead of being read through stale points. Wrong-grid destinations of the vector-calculus ! forms and short mixed sparsity patterns are refused too. Features Matrix-free product The serial product at 262144 unknowns takes 0.42× the assembled sparse product's time in 2D and 0.64× in 3D, from 1.9× and 3.2× at v3.22.0's code. It is still 1.09× and 1.17× the Kronecker product's. Weights are read by Cartesian index, the interior is gathered row by row along the first axis, and per-axis geometry is cached for separable terms and rebuilt when the mesh moves. Zero-allocation Polyester threading Twelve CpuPolyester paths allocate nothing on a warm call: differences, averages, avgₕ!, shifts, divₕ!, curlₕ!, εₕ!, broadcasts, innerₕ, inner₊ₓ, masked innerₕ and the Kronecker product. Forms carry at most 512 B per call (#437 follows them up). Kronecker products and the explicit right-hand side run under @batch, and loading Polyester no longer invalidates Bramble's cached hook callers. The Polyester extension precompiles every path whose first call took 100 ms or more. First calls drop from 148–1833 ms to 4–124 ms, and using Bramble, Polyester takes 1.04× as long. Device-backed forms Forms with a grid-function coefficient, and forms with πₕ across two device meshes, assemble on a device space (Metal). jacobian_pattern widens coefficient dependencies through πₕ and reads test-side πₕ rows. Fixes Δₕ on a composite element is the Laplacian of each component, not only of the first. jacobian_pattern and the host refill agree with assemble on cross-mesh and mixed composite forms. The GMG coarse solve allocates nothing at --optimize=1. JET's package report is empty again. Documentation The matrix-free example says when each route is preferred, with measured ratios. The Kronecker structure behind the memory-scaling example, the solver decision trees and the policy crossover are explained. The site moved to MaterialDocs, and the worked examples are grouped. The benchmarks page compares alternatives as well as each benchmark against v3.0.0. Tests The suite is split by concern into unit, slow, quality, ext, examples, AD and GPU groups, and runs at --optimize=1. Aqua and JET run on the tests themselves, and every src/ file is covered to at least 90%. Full changelog: https://github.com/gpena/Bramble.jl/compare/v3.20.0...v3.23.0

Maqolada o‘zbek ilmiy diskursining leksik tarkibi va diskursiv xususiyatlari korpusga asoslangan lingvostatistik yondashuv asosida tadqiq etildi. Tadqiqot materiali O‘zbekiston Milliy univ…

Maqolada o‘zbek ilmiy diskursining leksik tarkibi va diskursiv xususiyatlari korpusga asoslangan lingvostatistik yondashuv asosida tadqiq etildi. Tadqiqot materiali O‘zbekiston Milliy universiteti ilmiy nashrlaridan tanlangan 150 ta maqoladan shakllantirilgan 233475 tokenli korpusdan iborat. Chastota va nisbiy chastota tahlili asosida yuqori chastotali leksik birliklar hamda ilmiy argumentatsiyada faol qo‘llanuvchi diskurs markerlari aniqlanib, funksional guruhlar bo‘yicha tasniflandi. Tadqiqot natijalari  korpusga asoslangan lingvostatistik yondashuv o‘zbek ilmiy diskursining leksik va diskursiv xususiyatlarini empirik jihatdan tavsiflashda samarali vosita ekanligini ko‘rsatadi.

The crushing and pressing of oilseeds generates substantial quantities of press cake, which is rich in fibre,carbohydrates, and protein and is therefore of interest for further valorisation, tradition…

The crushing and pressing of oilseeds generates substantial quantities of press cake, which is rich in fibre,carbohydrates, and protein and is therefore of interest for further valorisation, traditionally in the animalfeed sector. In Europe, five oilseeds are predominantly used for vegetable oil production: rapeseed (colza),soybeans, sunflower seed, flaxseed (linseed), and maize germ. Other feedstocks are also used, but were notconsidered further due to their comparatively minor relevance. To establish a comprehensive dataset of press cake quantities from vegetable oil production, an in-depthdesk research was conducted covering the largest industrial oil mills in the ten largest producing countriesin Europe. In total, 84 industrial oil mills were identified. Data on production was generally found oncompany webpages, newspaper articles or governmental documents. For two facilities, neither productiondata nor a sufficiently robust method for estimating the quantities could be identified, therefore, thesefacilites were assigned NAs as input and output values in the dataset. For a small number of facilities,production quantities had to be estimated. The respective estimation methods are documented in the datasheet. Press cake quantities were subsequently calculated using seed-specific residue-to-product ratios (RPRs), withthe corresponding sources documented in the metadata sheet. Where a facility processed multiple oilseedtypes but the available data did not specify the quantities of the individual feedstocks, equal shares of therespective feedstocks were assumed. Finally, the estimated press cake quantities of individual oil mills wereaggregated from NUTS-3 to NUTS-0 level. Several limitations arise from the nature and availability of the underlying data. For some facilities, specificproduction data were available, whereas for others, only general production capacities were reported. Inmost cases, the year to which the reported data referred was not specified. Therefore, the year of theresearch was used as the reference year (2025), although the actual year represented by the data may differ.Where data for several years were available, the most recent year was used. All biomass potentials arereported in terms of annual tonnes of fresh matter. Where production figures were reported as dailyquantities, an operating period of 360 days per year was assumed to calculate the corresponding annualquantities. Furthermore, where several plants were located within the same NUTS entity, their biomasspotentials and corresponding source references were aggregated. If no information on a given plant couldbe identified, its biomass potential was not included in the respective NUTS entity. Such cases are indicatedin the column “COMPLETE”, where the value is set to “false”. Consequently, the reported biomass potentialfor these NUTS entities should be understood as a minimum value. All the data on biomass potential is provided in the corresponding CSV file. A detailed overview of theavailable data and sources is provided in the metadata file for this dataset. The work was done within the project SURFsUp. The project is supported by the Circular Bio-based EuropeJoint Undertaking and its members under Grant Agreement No. 101157586 and UKRI Grant Agreement No.10115826.

Reproduce all the pictures; including HRRR/ERA5 wind + EPA AQS PM2.5 cache input, see README for usage.

Signal Processing Explained: Systems, Analysis & Digital Communication Technologies is a publication-grade Open Educational Resource (OER) module covering the mathematical foundations, discrete-ti…

Signal Processing Explained: Systems, Analysis & Digital Communication Technologies is a publication-grade Open Educational Resource (OER) module covering the mathematical foundations, discrete-time systems, spectral transforms, and digital filter architectures governing signal analysis across electrical and electronic engineering. Serving as an integral core module within the Electrical and Electronic Engineering curriculum on Prep4Uni.Online, this text bridges analog front-end conditioning and Nyquist-Shannon sampling with contemporary AI-era computational paradigms, neural operators, and edge digital signal processors. Key Features & Pedagogical Additions:• Theoretical Grounding & Contemporary Paradigms: Integrates classical continuous and discrete-time theory (Fourier transforms, z-transforms, Nyquist sampling, quantization noise, convolution, spectral windowing) with Physics-Informed Neural Networks (PINNs) solving acoustic Helmholtz wave equations for inverse tomography, Fourier Neural Operators (FNOs) for non-stationary spectrogram modeling, and Lie-group SO(3) Invariant Extended Kalman Filters (IEKF) for multi-axis sensor fusion.• Systems Engineering Architecture & Trade-Off Analysis: Structured comparison matrix contrasting Finite Impulse Response (FIR) against Infinite Impulse Response (IIR) filters, Fixed-Point (Q15/Q31) against Floating-Point arithmetic, Discrete Wavelet Transforms (DWT) against Short-Time Fourier Transforms (STFT), and Recursive Least Squares (RLS) against Least Mean Squares (LMS) adaptive filtering.• Worked Numerical Engineering Application: A complete, multi-step engineering calculation detailing an acoustic vibration monitoring chain—evaluating Nyquist oversampling margins, ADC quantization step sizes (V_LSB), theoretical 16-bit SQNR (98.08 dB), linear-phase FIR low-pass filter tap sizing via Bellanger's approximation, and a 204.8x computational speedup analysis comparing direct DFT against Radix-2 FFT algorithms.• Curricular Assessment & Analytical Inquiries: Contains 25 fully resolved foundational review questions, deep analytical design inquiries, and step-by-step numerical calculations spanning anti-aliasing cutoff frequencies, ADC quantization signal-to-noise ratios, Radix-2 FFT operation counts, FIR linear convolution output lengths, first-order discrete-time RC filter cutoff derivations, active notch filter decibel attenuations, and uncompressed audio bitrate compression ratios.

Demonstration video of pilot experiment for "Investigating the Intersection of Consciousness and Emotion"

Project Sentinel is an unalterable, mathematically synchronized socioeconomic manifesto detailing a transition from old-world capitalist extraction to a permanent, sustainable cosmic civilization. The…

Project Sentinel is an unalterable, mathematically synchronized socioeconomic manifesto detailing a transition from old-world capitalist extraction to a permanent, sustainable cosmic civilization. The framework completely eradicates systemic poverty and generational dynasties by establishing a free universal baseline birthright alongside a strict zero-growth mandate. Traditional currency is entirely replaced by a 1:1 time-equivalent credit system where all human efforts—including creative arts, scientific innovation, and health-centric activities like dance and martial arts—are compensated equally. Monitored by an incorruptible AI Peer System, all luxury status and service credits remain strictly transient, ensuring society remains perfectly fair, balanced, and uncorruptible for a millennium.  

В статье проверяется распространённое допущение о том, что развитие цифровой инфраструктуры автоматически повышает уровень жизни населения. Данные по пяти странам Центральной Азии за 2018–2025 г…

В статье проверяется распространённое допущение о том, что развитие цифровой инфраструктуры автоматически повышает уровень жизни населения. Данные по пяти странам Центральной Азии за 2018–2025 годы это допущение опровергают. Казахстан — региональный лидер по цифровым рейтингам (индекс развития электронного правительства 0,9009, 24-е место в мире) — в 2025 году зафиксировал снижение реальных доходов населения на 2,7 %. Узбекистан при более скромной цифровой оснащённости (индекс 0,7999) сократил бедность с 14,0 % в 2022 году до 5,8 % в 2025 году, а неравенство доходов (коэффициент Джини) снизил с 0,35 до 0,33. Цель работы — объяснить это расхождение и предложить практический инструмент управления.