Our Work

Overcoming the middle-income trap is one of the central challenges facing developing economies undergoing structural transformation. For Uzbekistan, this question has become particularly pressing amid…

Overcoming the middle-income trap is one of the central challenges facing developing economies undergoing structural transformation. For Uzbekistan, this question has become particularly pressing amid the liberalization of foreign trade, the expansion of export potential, and the transition away from a predominantly resource-based model toward higher value-added production. This study assesses the role of export diversification in reducing the risk of the middle-income trap in Uzbekistan. It applies comparative and statistical analysis of export-structure dynamics together with econometric time-series modelling of merchandise exports and the export concentration index over 1995–2022. Of seven models tested (linear and parabolic trend, moving-average and exponential smoothing, AR(1), MA, and ARIMA(0,1,0)), the parabolic trend model produced the best fit by RMSE and MAE (R² ≈ 80%) and was used to generate forecasts through 2030. The analysis took the data from the World Bank, UNCTAD, the World Trade Organization, the Statistics Agency under the President of the Republic of Uzbekistan, the Central Bank of Uzbekistan and other national and international sources. The analysis is based on the data of the World Bank, UNCTAD, the World Trade Organization, the Statistics Agency under the President of the Republic of Uzbekistan, the Central Bank of Uzbekistan and other national and international sources. The results indicate that structural diversification remains limited, despite a significant increase in export volumes since the beginning of economic reforms. The UNCTAD export diversification index was 0.729 in 2022, compared to 0.8 in 2019, and the econometric forecast for the Herfindahl–Hirschman-type concentration index points to a further increase, to 0.474 by 2030, implying continued commodity concentration if current trends continue. These results

This report presents the findings of a comprehensive systematic review and meta-analysis commissioned by the Education Endowment Foundation (EEF) to identify how educational technology (EdTech) interv…

This report presents the findings of a comprehensive systematic review and meta-analysis commissioned by the Education Endowment Foundation (EEF) to identify how educational technology (EdTech) interventions impact pupil attainment, particularly for disadvantaged learners. The study addresses a critical need for clarity amidst growing EdTech usage and mixed evidence on effectiveness, especially following increased reliance on digital learning during and after the Covid-19 pandemic. The research was conducted in three sequential phases. Phase 1 involved stakeholder engagement with educators and researchers to shape a conceptual framework grounded in practical experience. Phase 2 synthesised global research through systematic review and meta-analysis to assess the impact of EdTech interventions across varied educational contexts. Phase 3 included a structured community review where practitioners validated and refined the findings, ensuring alignment with real-world implementation. Nine high-potential mechanisms were identified as central to effective EdTech use, including aligning technology with pedagogy, creating dynamic feedback loops, supporting adaptive teaching, fostering self-regulation, and ensuring inclusive, accessible design. These mechanisms were synthesised into "building blocks" that underpin effective EdTech models and interventions. Importantly, the review found that while EdTech can benefit all students, disadvantaged pupils may derive disproportionately greater benefit when interventions are well-designed and contextually implemented—though barriers to access and implementation risks remain. This mixed-methods review uniquely integrates quantitative evidence with practitioner insights, offering a practical and nuanced framework for designing, implementing, and evaluating EdTech interventions. The report concludes with actionable guidance for educators, policymakers, and EdTech developers and highlights the need for future research to deepen understanding of contextual factors, mechanisms of change, and sustainable implementation practices. An output of the Open Development & Education, https://opendeved.net/ Available from https://docs.opendeved.net/lib/Z4TK5P7S

In middle income Asian economies, export diversification plays a crucial role in determining both sustainable economic growth and economic resilience, since relying on a narrow export base makes econo…

In middle income Asian economies, export diversification plays a crucial role in determining both sustainable economic growth and economic resilience, since relying on a narrow export base makes economies more susceptible to external shocks and price fluctuations. Using panel data for 27 middle income Asian countries (2001–2024), this study examines how trade policy and import structure affect export concentration, applying Pooled OLS, Fixed Effects, Random Effects and Bayesian Model Averaging. Natural resource rents are found to be the strongest driver of export concentration, supporting the resource curse hypothesis, while manufacturing value added and tertiary education are the main forces for diversification. Aggregate trade policy variables, such as tariffs and trade openness, are not found to be robust predictors once country heterogeneity is accounted for. The findings highlight the importance of industrial policy, higher education investment and resource revenue management for sustainable export diversification.

While lexico-semantic tropes have been extensively investigated in prose and poetry, their significance in proverbs has explored insufficiently. Consequently, the stylistic and semantic value of lexic…

While lexico-semantic tropes have been extensively investigated in prose and poetry, their significance in proverbs has explored insufficiently. Consequently, the stylistic and semantic value of lexico-semantic tropes in proverbs gained limited scholarly attention. Therefore, the main aim of this study is to evaluate and present the linguopoetic features of 6 primary lexico-semantic tropes: metaphor, metonymy, synecdoche, personification, hyperbole and litotes within the most significant category of the English paremiological units - proverbs. By employing linguopoetic, continuous sampling, stylistic analysis and cognitive-semantic mapping methods, the results of the study demonstrates that these tropes function not only just as stylistic decoration, but also they contribute to cognitive economy, memorability, didactic efficacy and cultural encoding of the proverbs.

Quantum state certification concerns a resource-efficient way to verify whether a quantum device produces a prescribed target state without reconstructing the state in full, making it a fundamental pr…

Quantum state certification concerns a resource-efficient way to verify whether a quantum device produces a prescribed target state without reconstructing the state in full, making it a fundamental primitive for validating quantum systems. We consider the private-coin model with only quantum communication, where the parties have no shared randomness or preshared entanglement, and no additional classical communication is allowed. We establish the optimal sample complexity, thereby closing the logarithmic gap between the previously established upper and lower bounds in this model. Remarkably, our upper bound attains the log-free optimum using a fixed family of Clifford encoders. Thus, optimal distributed state certification under private coins and quantum-only communication does not require general quantum encoders.

We are pleased to release version 1.1.2 of the Input Redundancy Toolbox. 📦 What's Included This initial release provides a complete MATLAB environment to analyze input redundancy and implement geometr…

We are pleased to release version 1.1.2 of the Input Redundancy Toolbox. 📦 What's Included This initial release provides a complete MATLAB environment to analyze input redundancy and implement geometric control allocation for over-actuated systems. +geometric namespace: Core algorithms for geometric control theory. ir* functions: Practical tools for dynamic control allocation and redundancy analysis. 🚀 How to Install Download the input-redundancy-toolbox.mltbx file attached below. Double-click the downloaded file directly from your OS, or drag and drop it into the MATLAB workspace. Click Install when prompted. The toolbox functions and namespaces will be automatically added to your MATLAB path. 📚 Citation If you use this toolbox in your academic work, please refer to the BibTeX citation provided in the README.md file.

# MSFK MATLAB implementation and reproducible validation This package accompanies **Modified Stockwell Frequency-Wavenumber Workflow for Surface-Wave Dispersion Representation** by Hanbing Ai, Jiangta…

# MSFK MATLAB implementation and reproducible validation This package accompanies **Modified Stockwell Frequency-Wavenumber Workflow for Surface-Wave Dispersion Representation** by Hanbing Ai, Jiangtao Li, Yunus Levent Ekinci, Yang Hu, Ling Ning, and Yingwei Yan. The MSFK algorithm was developed by **Hanbing Ai**. It selects the dominant complex coefficient of the standard Stockwell transform at each frequency and receiver and passes those coefficients to a spatial Fourier transform. The standard Stockwell transform is unchanged. MSFK requires neither group-velocity steering nor repeated slant stacking. ## Version and citation status - Prepared package version: **1.1.0**.- Preparation and validation date: **2 October 2026**.- Primary developer: Hanbing Ai, hanbingai@foxmail.com.- Validation environment: MATLAB R2024b, PCWIN64.- Version 1.1.0 DOI: https://doi.org/10.5281/zenodo.23121000.- Original version DOI: https://doi.org/10.5281/zenodo.22108446. The authors reserved the version 1.1.0 DOI in Zenodo on 3 October 2026. Zenodo registers this DOI when the record is published. The original version DOI remains the identifier of the earlier archive. Original source headers retain that historical DOI; use the version 1.1.0 citation below for the complete package and added validation. No publication date is asserted in this package. Recommended citation: Ai, H., Li, J., Ekinci, Y. L., Hu, Y., Ning, L., and Yan, Y. (2026). Modified Stockwell Frequency-Wavenumber Workflow for Surface-Wave Dispersion Representation (Version 1.1.0) [Computer software]. Zenodo. https://doi.org/10.5281/zenodo.23121000. ## Files and quick start The original implementation and example are retained without changes: - `msfk_dispersion.m`: MSFK dispersion-image implementation.- `msfk_stockwell_transform.m`: documented complex Stockwell transform.- `Stockwell_FK.m`: compatibility wrapper.- `test.m`: original synthetic example.- `OUTPUT_SG_DWM_20260323_300Hz.mat`: original synthetic input. Run the original example from the package directory: ```matlabtest``` The second-round validation is in `validation_R2/`. Its copies of the MSFK core and synthetic input are byte-identical to the original files above. This duplication allows the validation folder to run independently as journal supplementary code. ```matlabaddpath('validation_R2')run_round2_testscalculate_band_RMSE``` These commands write into `validation_R2/results/`. Example outputs are already supplied there; copy the package to a writable directory before rerunning it. `validation_R2/README.txt` describes the tests and their parameters. ## Additional validation and interpretation The package adds known-phase single-mode controls, overlapping and separated two-mode controls, an unnormalized input-spectrum and Parseval audit, Gaussian/Hann coefficient-window ablation, and five paired noise realizations at each noise level. The noise results compare each method with its own noise-free image. They measure normalized-image sensitivity, not velocity RMSE, repeated-human-pick uncertainty, or modal detection probability. The point audit retains the supplied picks and frozen mode assignments. Theoretical velocities are linearly interpolated from saved Shi et al. SASEM branches at each pick frequency, without extrapolation or bridging missing segments. No new solver call is made at each point. The calculator independently reconstructs those residuals and applies the same 5-80 Hz eligibility band to every method, with no residual-based rejection. Verification results: - Both 40 Hz controls peak at the prescribed 320 m/s on the spatial FFT lattice; maximum inter-receiver phase residuals are approximately 4.12e-14 and 1.35e-14 rad.- The independent Parseval check has relative error approximately 1.34e-16.- Independent re-interpolation agrees with cached residuals to a maximum difference of approximately 4.12e-13 m/s.- Separate portable control and eight-method ensemble runs reproduce the validation statistics; the band-table difference is below 1e-12 m/s. The controls support phase consistency for the stated narrowband cases and retention of concurrent contributions; weaker temporally separated arrivals can be suppressed. The supplied Figure 2 M5-M6 picks coexist at 99-102 Hz in a weak-input-energy interval. At 100 Hz, their velocities are 284.8 and 301.2 m/s, compared with SASEM reference values of 285.36 and 301.45 m/s. This supports local ridge separation and reference agreement, without validating every high-frequency feature. The M5 labels within the supplied mixed-mode file are nearest-admissible-branch assignments; the M6 labels are explicitly supplied. Assignments alone are not independent modal ground truth. The common 5-80 Hz table band is a conservative evaluation choice for these records, not a universal MSFK frequency limit. No inversion experiment is included. ### Publication palette `validation_R2/publication_colormap.mat` stores the exact 254-row RGB palette used by manuscript Figure 2. `format_round2_publication.m` applies it explicitly to each normalized spectral axes, with colour limits [0,1] and colour-bar ticks at 0, 0.5, and 1. This keeps the supplementary spectra consistent with the synthetic and field comparisons. The presentation correction does not change spectral arrays, theoretical curves, supplied picks, or numerical metrics. Grayscale gathers and line/error-bar plots retain their distinct encodings; the original example script is preserved unchanged. ## External comparator implementations To rerun the eight-method noise diagnostic: ```matlabrun_round2_ensemble(benchmarkDir, sfkDir)``` `benchmarkDir` must contain the original `phase_kai.m`, `FK_Dispersion.m`, `Radon_conjgrad.m`, and helpers; `sfkDir` must contain SFK and its dependencies. These third-party functions are not redistributed here. F-J and F-H use Kai Zhang's MATLAB `phase_kai.m` from [PassiveFW](https://github.com/jordankai/PassiveFW/tree/main/active%20seismic%20modeling%20by%20DWM). The supplied wrapper records the original calls: time-by-receiver data; offsets 3:1:62 m; velocities 50:2:600 m/s; dt = 0.0005 s; fmax = 150 Hz; df_factor = 1; vertical component `v`; kernels `bessel` and `hankel`. The additional ensemble uses fmax = 80 Hz. [CC-FJpy](https://github.com/ColinLii/CC-FJpy) provides related Python implementations; it was not timed in this study. SFK is obtained from the source associated with Serdyukov et al. (2019). Obtain comparator code from its original providers and comply with their terms. Restricted legacy `st.m` and `seidisp.m` are excluded and are not needed by the self-contained controls. ## Attribution and license Original source headers, authorship, creation dates, and copyright notices are preserved. Algorithm development is credited to Hanbing Ai; the added validation scripts document tests prepared for this study on 2 October 2026. See `LICENSE` and `THIRD_PARTY_NOTICES.txt` for the software-license scope and exclusions. No new ownership or license is asserted over third-party software or field datasets. No field seismic records or published field curves are redistributed here. ## References Ai, H., Li, J., Ekinci, Y. L., Hu, Y., Ning, L., and Yan, Y. (2026). Modified Stockwell Transform-Driven Frequency-Wavenumber Method for Reliable Surface-Wave Dispersion Energy Representation [Computer software]. Zenodo. https://doi.org/10.5281/zenodo.22108446. Original archive title, retained for historical citation. Stockwell RG, Mansinha L and Lowe RP (1996). Localization of the complex spectrum: The S transform. IEEE Trans Signal Process 44(4):998-1001. https://doi.org/10.1109/78.492555. Shi C, Ren H, Li Z and Chen X (2022). Calculation of normal and leaky modes for horizontal stratified models based on a semi-analytical spectral element method. Geophys J Int 230(3):1928-1947. https://doi.org/10.1093/gji/ggac163. Wang J, Wu G and Chen X (2019). Frequency-Bessel transform method for effective imaging of higher-mode Rayleigh dispersion curves from ambient seismic noise data. J Geophys Res Solid Earth 124(4):3708-3723. https://doi.org/10.1029/2018JB016595. Yang Z, Sun YC, Zhang D, Han P and Chen X (2024). A frequency-Hankel transform method to extract multimodal Rayleigh wave dispersion spectra from active and passive source surface wave data. Geophysics 89(2):KS69-KS81. https://doi.org/10.1190/geo2023-0189.1. Xi C, Xia J, Mi B, Dai T, Liu Y and Ning L (2021). Modified frequency-Bessel transform method for dispersion imaging of Rayleigh waves from ambient seismic noise. Geophys J Int 225(2):1271-1280. https://doi.org/10.1093/gji/ggab008. Li Z, Zhou J, Wu G, Wang J, Zhang G, Dong S, Pan L, Yang Z, Gao L, Ma Q, Ren H and Chen X (2021). CC-FJpy: A Python Package for Extracting Overtone Surface-Wave Dispersion from Seismic Ambient-Noise Cross Correlation. Seismol Res Lett 92(5):3179-3186. https://doi.org/10.1785/0220210042. Serdyukov AS, Yablokov AV, Duchkov AA, Azarov AA and Baranov VD (2019). Slant f-k transform of multichannel seismic surface wave data. Geophysics 84(1):A19-A24. https://doi.org/10.1190/geo2018-0430.1. Zhang K (2021). Active and passive seismic full-wavefield modeling based on discrete wavenumber method (Version 1.0) [Computer software]. Zenodo. https://doi.org/10.5281/zenodo.5559375.

Project Summary: Compelling evidence exists for two classes of black holes (BHs), stellar-mass and supermassive. The Milky Way is expected to have 10^8 stellar-mass BHs, ranging in mass from ~10^…

Project Summary: Compelling evidence exists for two classes of black holes (BHs), stellar-mass and supermassive. The Milky Way is expected to have 10^8 stellar-mass BHs, ranging in mass from ~10^1 to 10^2 times that of the Sun. Supermassive black holes (SMBHs) are found in the centers of nearly every large galaxy, including the Milky Way, and range from ~10^6 to 10^9 solar masses. Observations by the James Webb Space Telescope have revealed galaxies that host SMBHs within the first 5x10^8 years after the Big Bang. However, the origin of SMBHs is unclear since they must grow rapidly in a relatively short time. Clues to the origin of common SMBHs may lie in the rare galaxies harboring central intermediate-mass black holes (IMBHs) having masses between stellar-mass and SMBHs (~10^3 to 10^5 solar masses). The discovery and analysis of these galaxies could elucidate the co-evolution of SMBHs and their galaxies. Observationally confirming an IMBH in a galaxy center requires significant telescope resources, so targeting likely candidates is imperative. One method of identifying potential targets is by using easily observable parameters of a galaxy that link to the central BH mass. One such relation is between SMBH mass and the degree of winding of spiralarms in a galaxy (known as pitch angle or PA). Spiral galaxies with tightly wrapped arms, or small PA, are found to have larger SMBH mass and vice versa. The existence of large sets of all-sky survey image data makes measuring PA a more time efficient process for estimating the central BH mass of a spiral galaxy. Previously, we ran a participatory science project called Spiral Graph (SG) that asked volunteers to trace the spiral structure in relatively low contrast all-sky survey images of tens of thousands of galaxies. Those tracings effectively increased the signal-to-noise ratio of the arms relative to the background, allowing us to measure the PA via software called P2DFFT. Since many spiral galaxies are not perfectly symmetric, a single fit to all the arms is not realistic and our results showed us that greater accuracy is needed in determining a "global" PA. This can be achieved by measuring the PA of each arm individually and combining the weighted results into a global value. Disentangling the independent tracings from multiple volunteers of each galaxy is not trivial so we developed a line clustering algorithm (LCA), based on a set of hyperparameters (HPs), that is able to group similar tracings into individual arm clusters while eliminating outliers. The LCA produces good results with most clear cases, but it requires dynamic adjustments to the HPs in order to fit a variety of cases correctly. This proposal is for the development of a participatory science project called Spiral Graph: Cluster Buster (SG:CB) on the Zooniverse platform. Specifically, we will: (1) run the LCA with a set of HPs on a series of galaxy tracings obtained via SG; (2) color code the line clusters and present both clustered and unclustered tracing images to volunteers via SG:CB; (3) ask the volunteers to evaluate if the LCA properly clustered all the perceived arms; and (4) adjust the HPs for reapplying steps (1) - (4). This will lead to a comprehensive dataset of LCA HPs for galaxy tracings, allowing us to develop and train an AI algorithm that dynamically selects appropriate LCA HPs. This will result in proper clustering for any galaxy arm tracings and allow PA measurements of individual arms. This proposed project directly addresses the goal of the CSSFP in contributing to the building of a scientifically literate nation by providing opportunities for all participatory scientists (PS), encouraging volunteers who can benefit NASA, leveraging existing PS communities, and connecting PSs with subject matter experts. It also aligns with NASA's Astrophysics Division's interests in understanding the history of galaxies, revealing the extremes of nature, and public engagement with astronomy learners.

[Last updated: October 3, 2026 (Version 5)] I had previously included the proof for the limit limn→∞[n-li(n{logn+loglogn-1+(loglogn-2)/logn})]=0—which forms the core of my paper&mdash…

[Last updated: October 3, 2026 (Version 5)] I had previously included the proof for the limit limn→∞[n-li(n{logn+loglogn-1+(loglogn-2)/logn})]=0—which forms the core of my paper—in Chapter 5. In addition, in the previous version (Version 4), I verified the validity of this result through numerical calculations using PARI/GP. Furthermore, in this current version (Version 5), I have provided a rigorous proof by correctly employing methods involving continuous functions and the Intermediate Value Theorem. This has enabled me to conclusively derive that the Riemann Hypothesis is false based on the fact that limn→∞[n-li(n{logn+loglogn-1+(loglogn-2)/logn})]=0 holds. For details, please refer to pages 40–41 of Section 6.2 ("FINAL COMMENT") in Chapter 6.  Additionally, as my paper resolves the Riemann Hypothesis by examining the error term O(1) in the asymptotic formula for prime numbers Pn=n(logn+loglogn+O(1)), I have decided to change the title of the paper. I have also revised the abstract; please note that while the content itself remains unchanged, the abstract has been updated. 【ABSTRACT】This paper investigates the error term O(1) in the classical asymptotic formula for the n-th prime number, Pn=n{logn+loglogn+O(1)}. By analyzing Cipolla’s asymptotic expansion alongside the explicit prime number estimation formulas established by Rosser and Schoenfeld, Dusart, and Axler, I demonstrate that the error term O(1) must be treated rigorously as an arithmetic function. Incorporating Littlewood ’s theorem, I show that the behavior of this error term is governed by a dual structure consisting of Dusart–Axler-type bounds and a complementary framework derived from the broader scope of the Rosser–Schoenfeld formula. I prove that these two constituent formulas exhibit mutual exclusivity, implying that any given prime number can satisfy at most one framework at a given scale. Finally, by evaluating these structural constraints within the context of von Koch's theorem regarding the equivalence between the error term for prime numbers and the Riemann Hypothesis, I demonstrate the falsehood of the Riemann Hypothesis.