This paper investigates the application of Multi-Agent Reinforcement Learning (MARL) to optimize software development processes. Traditional software development methodologies often struggle with adaptability and efficiency, particularly in complex projects. This research proposes a novel approach leveraging MARL to dynamically adjust and refine the development workflow. The core idea involves modeling the software development process as a multi-agent system, where each agent represents a distinct stage or activity. These agents learn optimal strategies through interaction and reward signals, leading to improved development efficiency and quality. We present a framework for formulating this problem, detailing the agent architecture, state space, action space, and reward function. The effectiveness of the MARL approach is demonstrated through a theoretical analysis and conceptual design, highlighting its potential to overcome limitations of static, rule-based methodologies. Future work will focus on implementing and testing this framework within a simulated software development environment.
Jincheng Zhang· Zenodo (CERN European Organi...· 0 citations
This software package contains the Musiconis Model Context Protocol (MCP) server and companion FastAPI/Gradio application developed for the article “Model Context Protocol Server for the Musiconis Medieval Music Iconography Database,” submitted to Digital Medievalist. The package enables LLM-assisted exploration of the Musiconis medieval music iconography database. It provides MCP tools for searching and retrieving iconographic records, accessing reference data and database statistics, generating summaries, and asking natural-language questions about the collection. The repository includes the Python application, MCP server, API routes, database bootstrap scripts, tests, technical documentation, SQL and JSON runtime fixtures, Docker configuration, and installation instructions. The large Musiconis JPEG image corpus is intentionally excluded from this software deposit.
Xavier Fresquet· Zenodo (CERN European Organi...· 0 citations
When two researchers analyze the same data with the same statistical model but different software, they expect identical results. This expectation is often violated. We identify and catalog 37 “hidden decisions”—default settings and undocumented implementation choices—that vary across five major statistical packages (SPSS, R, SAS, Stata, and Python/statsmodels) for five widely used methods: the t-test, logistic regression, Poisson GLM, linear mixed models, and Cox proportional hazards regression. Through controlled stress tests on carefully constructed “boundary datasets” where the reference p-value is near 0.05, and through verified five-software follow-up analyses, we show how hidden decisions can flip statistical significance in boundary cases. The most impactful verified decisions are: (1) scale or dispersion handling in Poisson GLMs, (2) tie-handling method in Cox regression (Breslow vs. Efron), (3) standard-error computation for noncanonical GLMs, (4) finite-sample degrees-of-freedom vs. Wald-z inference in mixed models, and (5) boundary-fit diagnostics in mixed models. For each hidden decision, we trace the discrepancy to source code where available, or to documentation and diagnostic behavior otherwise, quantify the effect size on p-values, standard errors, and significance verdicts, and provide practical recommendations. We argue that the reproducibility crisis in empirical research has an underappreciated computational component that can be addressed through transparency requirements in reporting.
This dataset contains processed numerical outputs supporting the figures, tables, validation tests, and Supplemental Material of the article "Linear Stability Analysis of Supercritical Water in Channel: Strongly Non-Ideal Effects." It includes equation-of-state and base-flow figure data, modal spectra and eigenfunctions, critical-layer locations, neutral curves, critical Reynolds-number scaling, modal and transient kinetic-energy budgets, non-modal optimal-growth results, wavenumber-parameter contours, numerical validation cases, and Supplemental Material animations. Data are primarily stored as MATLAB v5 `.mat` files, with selected `.csv`, `.xlsx`, `.fig`, `.png`, and `.gif` files. The numerical solver and license-restricted third-party software are not included.
Peitong Li, Hui Jin, Liejin Guo et al.· Zenodo (CERN European Organi...· 0 citations
Reach audiences
Advertise in front of researchers, engineers, and readers.
# See Sodium to Treat (SSTT) — Analysis Code This repository contains the statistical analysis code for the SSTT study,which investigated the effect of oral potassium chloride (KCl) supplementationon muscle sodium content and blood pressure in patients with hypertension. ## Contents | File | Description ||------|-------------|| `SSTT_script_upload.R` | Main analysis script (R). Reproduces Table 1, Table 2, and the Figure 3 / Online Supplemental Figure 1 regression statistics. || `sessionInfo.txt` | R and package versions used to produce the reported results. | ## Data The analysis requires the following data files, placed in the working directory: | File | Description ||------|-------------|| `SSTT_wide.RData` | Processed wide-format data (one row per participant; outcomes at Visit 1 and Visit 2). || `SSTT_long.RData` | Processed long-format data (one row per participant-visit). || `SSTT_data.xlsx` | Source data including demographics, medications, and aldosteronism subtype (PA/SA). | **Data availability:** The study data are individual-level clinical data and areavailable to qualified researchers on request to the study sponsor with a validresearch protocol (see the Data Availability statement in the manuscript). Thedata files are **not** included in this repository. ## How to run 1. Install R (version 4.6.0 or compatible) and the required packages:```r install.packages(c("tidyr", "dplyr", "readxl", "table1", "lme4", "emmeans", "pbkrtest"))```2. Place the three data files (above) in your working directory.3. Set the working directory in `SSTT_script_upload.R` (uncomment and edit the `setwd()` line), or run R from that directory.4. Run the script:```r source("SSTT_script_upload.R")``` ## What the script produces - **Table 1** — baseline comparisons (Student's t-test for continuous variables; Fisher's exact test for categorical variables; Welch and Mann-Whitney U are shown as a non-determinative robustness check).- **Prescribed KCl dose** — between-group Welch two-sample t-test.- **Table 2** — linear mixed-effects model contrasts (within-group effect of KCl, Visit 2 vs Visit 1; between-group HA vs EH at each visit), with Kenward-Roger degrees of freedom (via pbkrtest) and exact p-values.- **Figure 3 / Online Supplemental Figure 1** — linear regressions of the change in muscle Na relative to changes in plasma K and the plasma Na/K ratio, overall and by group.- **Session information** — R and package versions (`sessionInfo()`). ## Software versions Analyses were performed in R version 4.6.0. Exact package versions are recordedin `sessionInfo.txt` and at the end of the R script.
Adriana Marton, Thị Ngọc Hiếu Trần, Norihiko Morisawa et al.· Zenodo (CERN European Organi...· 0 citations
This record contains the replication package for the manuscript “Beyond Plans and Permissions: Enforcing Physical Effect Closure in Infrastructure Automation.” The package provides the EffectSeal reference implementation, executable verification tests, controlled Terraform/AWS-provider experiment configurations and frozen evidence, historical AWS Lambda and S3 provider reproductions, the AWS authorization-projection census and validity-hardening data, counterfactual stability and divergence analyses, approval-contract renderings, operational-cost accounting, and ordered-monitor scalability measurements. It also contains the frozen genuine-AWS CloudWatch Logs mediation witness and sanitized AWS CloudTrail corroboration. Reproduction instructions are provided in the package README and reproduction guide. The artifact supports the bounded empirical and reproducibility claims reported in the manuscript; it is not presented as production-wide cloud deployment validation or as a measurement of production end-to-end overhead.Licensing: Original EffectSeal software is released under the MIT License. Original data, documentation, and research artifacts produced by the authors are released under the Creative Commons Attribution 4.0 International License. Third-party materials and software remain subject to their respective copyright and license terms and are not relicensed by the authors.
Hamad Alsawalqah, Ahmad Abadleh· Zenodo (CERN European Organi...· 0 citations
Version 1.3 accompanies version 1.1 of Part II of the series, "The exact row of the Toeplitz–Cayley model: coefficient law, Binet–Laguerre remainder, and certified positivity at finite rank" (https://doi.org/10.5281/zenodo.22285620; concept DOI https://doi.org/10.5281/zenodo.22276540). New in this version: the research checkpoint frontier_v0_57 — the post-publication audit of Part II version 1.0 (blind re-derivation of the theorems by scripts that import no code of the project, every prose number recomputed, and two adversarial blocks: an independent end-to-end reproduction of a certified eigenvalue of Table 4 by a Cauchy-integral route, and the envelope, the budget and the transfer tested on that independent row at N = 625), the certified saturation check of the enveloping theorem in Arb ball arithmetic (N = 300, 625, 5000; K = 1..7: supremum of the ratio, attainment at d = 1, sign, two-sided row sums; the source of every number of the new Remark 3.7), the verbatim reports of six adversarial reviews (two of the revision, two complete re-readings of the notes, two verification passes) and the finding-by-finding review record, the q = 11/4 certificate of Part I replayed at N = 1250, 2500, 5000, and the deposited PDF and LaTeX sources of Part II version 1.1 (https://doi.org/10.5281/zenodo.22285620) and Part I version 1.3 (https://doi.org/10.5281/zenodo.22286746). The earlier versions of the notes, shipped in versions 1.0–1.2 of this archive inside the sealed checkpoints, are not shipped in version 1.3 because their title footnote carried a personal address (PACKAGE_REDACTIONS.md lists them with their hashes); no script, data file or replay report is affected. Replay: sh recovered_work/frontier_v0_57/REPRODUCE.sh /path/to/python [--n5000] (283 s, plus about 5 min with --n5000), or sh REPRODUCE.sh /path/to/python --audit from the top level. The cached exact rows are unchanged: the seven row archives of this record are those of version 1.2. Companion research software and audit archive for the two notes. It contains the scripts, the pinned historical inputs, the pre-computed replay reports, SHA-256 manifests of every file, and the notes themselves with their LaTeX sources. Every table entry, figure and certified enclosure displayed in Part II is produced by a script of this package, and every prose number is traceable to one of its replay reports. Requirements: Python 3.10 or later for Part I and 3.11 or later for Part II (the versions the replays were run on), python-flint 0.9.0 (Arb ball arithmetic), mpmath, numpy 2.x, matplotlib. Full replay of Part I: sh REPRODUCE.sh /path/to/python (a few minutes); of Part II: sh recovered_work/frontier_v0_54/REPRODUCE.sh and sh recovered_work/frontier_v0_55/REPRODUCE.sh (about three hours each on one core). Statements of a definite sign are decided only in ball arithmetic, on the appropriate one-sided endpoint; float64 scripts are labelled as probes and decide nothing. Reference environment: Linux x86-64, Python 3.11, python-flint 0.9.0, mpmath, NumPy 2.4.4, matplotlib; Part I additionally replayed on macOS 15 arm64 (Python 3.14.6) and on Linux x86-64 (Python 3.10.12). The artifact is distinct from the notes and has no direct implication for the Riemann hypothesis. Original code is MIT-licensed; project-authored documentation is CC BY 4.0, as detailed in the licence files inside the archive. Anthropic Claude assisted with preparation; Julien Lange is the sole author and assumes responsibility for the release. Versions 1.0, 1.1 and 1.2 of this archive, which contained those earlier versions of the notes, are withdrawn from the repository by the author on the day of this deposit.
Julien Lange· Zenodo (CERN European Organi...· 0 citations
This dataset contains reproducible network-performance experiments generated with QUICompare Auto-Research in an Oracle VirtualBox Type-2 hypervisor environment. The infrastructure consists of two virtual machines: a QUICompare VM, responsible for workload generation and measurements, and a Device Under Test (DUT) VM The DUT was configured with 5 vCPUs while its virtual memory was varied between 1 GB and 4 GB vRAM. The QUICompare benchmarking tool obey the RFC 2544/6815. For each memory configuration, experiments were executed with 10, 50, and 100 concurrent users, producing six principal workload/resource scenarios: 1 GB × 10 users; 1 GB × 50 users; 1 GB × 100 users; 4 GB × 10 users; 4 GB × 50 users; and 4 GB × 100 users. The dataset evaluates 15 Linux TCP congestion-control algorithms under equivalent virtualized conditions, varying Web-server and protocol configurations based on Apache2, NGINX HTTP/2, and NGINX QUIC/HTTP/3. TCP congestion-control algorithms are directly applied to TCP-based Apache2 and NGINX HTTP/2 scenarios. QUIC/HTTP/3 is analyzed as a separate comparative branch because QUIC operates over UDP and implements congestion control independently from the Linux TCP stack. Experiments were automated using QUICompare Auto-Research mode, with FFmpeg-based traffic generation and configurable acquisition density through the `--sample-rate` parameter. The workflow systematically varies congestion-control algorithm, service/protocol, concurrent users, DUT vRAM, and experiment repetition. The experimental matrix can be summarized as: Congestion Control × Service/Protocol × Users × vRAM × Repetition The resulting files support analysis of throughput, transferred data, latency, jitter, packet behavior, TCP/UDP characteristics, and temporal network-performance series. When the corresponding QUICompare modes are enabled, the data can also support studies involving long-range dependence, self-similarity, fractal behavior, heavy-tailed traffic, Hurst-related measures, sensitivity analysis, machine learning, and Data Envelopment Analysis (DEA). The dataset was designed for reproducibility and controlled comparison of congestion-control mechanisms, Web protocols, workload intensity, and virtualized computational resources. Tutorials, research articles, software, preconfigured virtual machines, installation instructions, datasets, and reproducibility resources are available at the official QUICompare website: https://lhraphael.github.io/quicompare/quicompare.html Our more recent FRANCISCO DEA network predictor/comparison tool is available at URL: https://israellmt.shinyapps.io/Francisco-3-0/
Francisco Daladier Marques Júnior· Mendeley Data· 0 citations
Open-source, unit-tested Python tool for building watershed-scale EPA SWMM 5.2 green-infrastructure (GI/LID) models from public and incomplete municipal data, with a Gaussian-process-emulator global (Sobol') sensitivity analysis and a transferable forcing/parameter calibration-diagnostic protocol. Companion code for the manuscript prepared for Environmental Modelling & Software. Contributions: Config-driven pipeline handling incomplete municipal storm-sewer data (orphan-outfall reconnection, DEM-Strahler channel augmentation) and SWMM's easily-missed multi-LID stacking constraint. A forcing/parameter calibration-diagnostic protocol (forcing-swap + GP-Sobol' + roughness sweep) localizing model error to forcing, losses, or conveyance, validated on a synthetic twin with known truth (5/5 correct, including the no-error control) and with quantified detection floors. An explicit mass-conservation precondition check. A node-flooding sink, which arises from SWMM's default configuration, can mask a real loss error: holding one loss error fixed while enlarging the sink, a genuine 27% volume error reports as +0.3% at a 21% sink so the protocol returns 'none, within tolerance', and beyond about 60% the sign inverts. Routing continuity error stays at 0.00% throughout and never registers it. Reproducible via tools/twin_masking.py. GP-emulator global sensitivity analysis: channel Manning's n dominates outlet peak Q (total-order index 0.94-0.98 among four sampled parameters). A controlled twin experiment varying only the channel size shows the ranking order is invariant from a free-draining network to 88% flooding loss, while the shares are not: surcharge drives the loss parameter's total-order index from 0.17 to 0.00. A paired 100-member ensemble on the repaired configuration then tests the same question on the real basin. Removing the mass sink leaves the outlet ranking almost unchanged (total-order index for channel n 0.978 to 0.968), so roughness leads there because the reach routes the flood wave rather than because surcharge suppressed the alternatives; at the headwater the loss parameter's share does rise, 0.652 to 0.872. Controlled two-basin build-and-diagnose demonstration (Leon Creek + Lower Medina, San Antonio, TX) plus third-basin generalization: the FOSS (WhiteboxTools/pysheds) pipeline reproduces the reference arcpy channel network within ~1% on length and drainage density (89% spatial overlap) and builds an independent third basin (Headwaters Salado Creek) with no code changes, delineating 98.5% of its mapped area. GI scenario performance and its cross-basin transferability are quantified in a separate companion paper. The demonstration model is reported candidly as uncalibrated for absolute prediction. Acting on the protocol's own verdict shows the raw diagnostics understated the error, because the flooding sink was removing most of the excess runoff; the true event-volume deficit is roughly sixfold. Four candidate mechanisms are then excluded by measurement (gauge-to-node drainage area, unrepresented detention, channel transmission loss, closed-depression storage removed by DEM filling). A fifth, Edwards Aquifer recharge-zone infiltration, is confirmed as a contributor and narrows the bias further. A second masked error then surfaces, since the calibration hyetographs were themselves synthetic: under basin-average radar forcing the wettest-antecedent event closes to within 7% of observed volume. The residual orders by antecedent wetness and is attributed to the model's single fixed soil-moisture state, which continuous simulation would track. Results rest on pipeline validation, sensitivity attribution, and diagnostic error attribution rather than absolute discharge.
Fahmida Yeasmin Sami· Zenodo (CERN European Organi...· 0 citations
A read-only, zero-dependency tool that measures, over any git repository, whether a declared version label identifies exactly one state of the producing code at the points where that repository publishes. Addressing a published artefact by (input, version) is sound only if the label is injective over closures. Nothing enforces that: the label is a string a human edits. When two code states share a label, one address denotes two outputs, and the system cannot detect it, because the label is the only thing it recorded. The tool distinguishes publication regimes. At tags (the default) it measures the case of released software; at commits it measures continuously published output. Measured against four widely used open-source projects (click, requests, packaging, httpx) it reports no drift; against a system publishing a daily edition under a hand-maintained label it reports one label covering six distinct closures. The failure belongs to continuous publication, not to versioning in general. Each report stamps the commit measured and the hash of the tool that measured it, because a count over repository history is a function of repository state. Version 0.4.0: the measurement script is byte-identical to 0.3.0; this version adds a negative test fixture (label-only ledger: must be refused at tags and reported as drift at commits), the documented limitation (the detector checks addressing, not re-execution of originating code states), and new reference results over three public repositories selected under a pre-registered rule. Version 0.5.0: the measurement script is byte-identical to 0.3.0 and 0.4.0, and no reference result was re-run. What changes is what the deposit says about itself, and what it asks for. RESULTS.md is a table for measurements produced by someone other than the author, on repositories the author does not control; it is published empty, because as of this release nobody outside the author has run the tool and reported a result, and omitting the section would let a reader assume otherwise. It states what a line must carry to count — the version DOI of the deposit used, the stamp block as emitted, and the publication-point setting — and records that results contradicting the detector are wanted on the same terms as results confirming it. The report carries counts, labels and hashes and never file contents, so a private repository can be measured without anything leaving the machine. SCOPE.md states what the tool does and, explicitly, what it will not be extended to do: the (A) the record is well-formed / (B) the artefact can be re-produced boundary, restated as a commitment rather than a caveat. NOTICE records that the author has patent applications pending; it adds no condition to the licence, and commercial use carries no royalty and no payment obligation. The README now leads with the measured result and adds two sections: why an unambiguous address is a precondition of reproducibility rather than a part of it, and what this tool is not — it never rebuilds and it issues no attestation.
Luiz F. Nunes da Silva· Zenodo (CERN European Organi...· 0 citations
Licensing and attribution correction for the publication software release. Changes Added a repository-level third-party notices document. Retained the complete upstream BSD 3-Clause licence for slanColor/slanCM. Added clear slanColor/slanCM attribution and provenance to the plotting helper and relevant documentation. Removed an intervening, invalid uncommented licence block from slanCM.m. The executable MATLAB statements in slanCM.m are unchanged from v1.2.1-publication_zenodo. This release does not change network architecture, training, testing, numerical processing, plotting behaviour, datasets or reported results.
K. O. Mason, Sonia Sennik, Claudia Clopath et al.· Zenodo (CERN European Organi...· 0 citations
This archive contains the complete replication package accompanying the manuscript "Local Consistency and Global Failure: A Cohomological Theory of Realization." The package includes the source code, verification library, automated test suite, and a single-command runner that reproduces all computational results reported in the manuscript. It verifies the theoretical propositions on randomized exchange networks, reproduces all numerical examples, and regenerates every computational claim reported in the paper. The software is released under the CC0 1.0 Universal licence.
Soumik Sarkar, Sampriti Pal· Zenodo (CERN European Organi...· 0 citations
A USAF cadet and a Lincoln Laboratory researcher found AI chatbots can help nontechnical service members produce viable software applications for their unique problems.
