Energy and natural sciences: paper-to-code routes
On this page
All 60 tools · How to contribute
These are construction routes researched by agents. The execution line reports the bounded work actually performed; the comparison section defines what remains before claiming a published result. Human scientific review is pending.
pypsa#
PyPSA — Can the historical European transmission network and its validation be reconstructed from the authors' workflow?
Route: Paper model adaptation. Execution: Not run. Source confidence: medium.
Paper for pypsa#
Jonas Hörsch; Fabian Hofmann; David Schlachtberger; Tom Brown (2018). PyPSA-Eur: An Open Optimisation Model of the European Transmission System. 10.1016/j.esr.2018.08.012.
Publication type: See primary metadata record; not classified here.. Primary metadata. Metadata verification: confirmed. Full-text access: Open author preprint; original metadata refreshed through arXiv after Crossref returned 429.
Target in the paper: Network dataset description and validation; 6001 lines and 3657 substations reported in the abstract. Section 3.4 validation is the first reproduction target.
Code and inputs for pypsa#
Artifact or upstream source. Author model repository and historical release; v0.1.0 README cites the 2018 article but this release is later than the paper, so exact paper inputs remain to establish.
Source inspected: yes. Ref: v0.1.0. Commit: 192b605bb55e74184e7bd1bb4217dd01b25d020f.
Files: README.md; environment.yaml; Snakefile
Terms: Historical source and component licenses must be read at the pin; input data have separate terms. Current PyPSA-Eur is MIT.
Environment and compute: Historical conda environment.yaml, Snakemake, PyPSA, atlite, geospatial packages, and solver versions must be reconstructed. Do not replace with the current Pixi tutorial silently. Geospatial workflow and continental optimization can require substantial RAM and hours; build a prebuilt-network inspection exercise first.
Inputs: GridKit/ENTSO-E topology, OPSD demand, plant database, reanalysis/cutout and eligibility inputs; prefer an archived prebuilt network to redownloading changing services.
Data access: README links archived prebuilt models at https://doi.org/10.5281/zenodo.3601882; archive contents/version-to-paper correspondence not yet inspected.
Construct and compare pypsa#
The sequence below contains upstream commands and proposed setup instructions, as labeled. Only commands identified in the execution evidence were run here; inspect the pinned source and prerequisites before running.
Proposed preparation: git clone https://github.com/PyPSA/pypsa-eur.git; cd pypsa-eur; git checkout 192b605bb55e74184e7bd1bb4217dd01b25d020fUpstream environment: conda env create -f environment.yamlProposed inspection: snakemake --cores 1 --dry-run; select the network-construction target after identifying the archived input bundle
Expected outputs: Historical PyPSA network files, input hashes, configuration, workflow log, topology summary and validation tables.
Comparison with the paper: First compare network line/substation counts to the paper's 6001/3657 and geographic/voltage scope; then reproduce the paper's validation tables/plots. Count agreement alone is insufficient. Numeric tolerance for validation requires reading the complete target table and matching its input snapshots.
Execution result: Sources and construction route inspected; no scientific result reproduced here.
Execution evidence: None recorded.
Blockers: Historical release does not by itself identify the exact 2018 input snapshot.; Full target-table extraction and immutable input-file hashes are pending.
Next action: Inspect the model archive and article methods; recover the paper-era snapshot, then lock one validation comparison before solving.
Gaps: Historical release does not by itself identify the exact 2018 input snapshot.; Full target-table extraction and immutable input-file hashes are pending.
Sources for pypsa#
- Title, authors, year, reported network counts and validation scope. (inspected 2026-11-15).
- Historical model, article relationship and archived-network link. (inspected 2026-11-15).
- Historical workflow source inspected. (inspected 2026-11-15).
pandapower#
pandapower — How do switching states and transformer taps affect losses and operating constraints in the published 10 kV ring network?
Route: Paper result candidate. Execution: Example executed. Source confidence: high.
Paper for pandapower#
Leon Thurner; Alexander Scheidler; Florian Schäfer; Jan-Hendrik Menke; Julian Dollichon; Friederike Meier; Steffen Meinecke; Martin Braun (2018). Pandapower—An Open-Source Python Tool for Convenient Modeling, Analysis, and Optimization of Electric Power Systems. 10.1109/TPWRS.2018.2829021.
Publication type: See primary metadata record; not classified here.. Primary metadata. Metadata verification: confirmed. Full-text access: Open author preprint; Crossref title/authors/year verified.
Target in the paper: Section VII-B switching analysis; VII-D / Figure7 operational results
Code and inputs for pandapower#
Artifact or upstream source. Exact paper-associated case study, explicitly linked by maintainers to the IEEE reference article.
Source inspected: yes. Ref: master. Commit: 6213b92acd265cf7803ff4b8978542957d0c9777.
Files: grid.ipynb; switch_evaluation.ipynb; timeseries_simulation.ipynb; requirements.txt; data/; results/
Terms: pandapower BSD-3-Clause; artifact redistribution terms not yet independently established, so link rather than vendor notebooks/data.
Environment and compute: Author README requires pandapower 1.4.3; requirements.txt is otherwise unpinned (pandapower, python-igraph, xlrd, numba, notebook). Reconstruct a compatible legacy Python/pandas/numpy environment before installing. Small CPU network; switching/time-series batch can take longer than one power-flow solve.
Inputs: Bundled 10 kV network, load profiles and author reference results; do not replace the legacy network with a current built-in test network.
Data access: Data and reference results are present in the public author repository; data licensing and numeric reference files still need audit.
Construct and compare pandapower#
The sequence below contains upstream commands and proposed setup instructions, as labeled. Only commands identified in the execution evidence were run here; inspect the pinned source and prerequisites before running.
Upstream: git clone https://github.com/e2nIEE/pandapower-paperProposed pin: cd pandapower-paper; git checkout 6213b92acd265cf7803ff4b8978542957d0c9777Upstream: pip install -r requirements.txt (only after adding the documented pandapower==1.4.3 constraint in an isolated compatible environment)Upstream: jupyter notebook; execute grid.ipynb, switch_evaluation.ipynb, then timeseries_simulation.ipynb
Expected outputs: Executed notebooks, feasible switch-state table, one-day losses/line-loading/transformer-loading/voltage time series, and differences against results/.
Comparison with the paper: Successful exact bounded switching-target comparison; the aggregate declared comparison is NOT marked passed because complete historical Figure7 claims are not all matched. Initial raw loss difference is a documented power/energy unit mismatch, not solver failure. Article says9 selected states; original notebook says11 and current run matches11.
Execution result: Published switching counts match exactly; full-day97-point author calculation completed. Eleven selected states match author notebook/JSON but differ from article text9. Operational constraints hold. Source-defined kW→kWh conversion reconciles loss energy; transformer convention unresolved.
Execution evidence: Author-code attempt; Run record
Blockers: Full-day paper text9states versus author notebook11states.; Saved transformer-loading reference is0.1 of new derated percentage; data df=0.1 and source calculation explain a plausible convention, but reference generation convention is undocumented.
Next action: Use the bounded switching guide now; resolve the two full-day source conventions before claiming complete Figure7 reproduction.
Gaps: Full-day paper text9states versus author notebook11states.; Saved transformer-loading reference is0.1 of new derated percentage; data df=0.1 and source calculation explain a plausible convention, but reference generation convention is undocumented.
Sources for pandapower#
- Article identity verified. (inspected 2026-11-15).
- Author attribution, three case-study notebooks and required pandapower 1.4.3. (inspected 2026-11-15).
- Actual dependency list; unpinned environment limitation. (inspected 2026-11-15).
calliope#
Calliope — Can the study's baseline carbon-neutral European energy system and near-optimal alternatives be regenerated?
Route: Paper result candidate. Execution: Saved author outputs analysed; simulator not run. Source confidence: high.
Paper for calliope#
Bryn Pickering; Francesco Lombardi; Stefan Pfenninger (2022). Diversity of options to eliminate fossil fuels and reach carbon neutrality across the entire European energy system. 10.1016/j.joule.2022.05.009.
Publication type: See primary metadata record; not classified here.. Primary metadata. Metadata verification: confirmed. Full-text access: Open full text and author archives; metadata verified.
Target in the paper: Table1 metric ranges; Figure3 describes441 configurations; Resource availability identifies model/result archives.
Code and inputs for calliope#
Artifact or upstream source. Exact article-associated workflow tag identified by the authors' README, plus prebuilt model DOI 10.5281/zenodo.5774988 and result archive DOI 10.5281/zenodo.6546817.
Source inspected: yes. Ref: 2021-11-29. Commit: f81453a454424e36f45eb0a304f28f82f6e126d7.
Files: README.md; environment.yaml; Snakefile; config/; run_scripts/; euro-calliope (submodule)
Terms: MIT workflow; archived data/results and solver terms separate.
Environment and compute: Historical customized Calliope 0.6.6 fork and submodules; archive includes model definitions. Original optimization used ETH Euler; obtain solver/version and license from the archived configuration before execution. Large CPU optimization; full ensemble used an HPC cluster. Begin with archive-only analysis and one baseline solve.
Inputs: Prebuilt sector-coupled 98-region model with 2018 input data, scenario overrides, SPORES settings and result archive (~1.1 GB).
Data access: Paper explicitly deposits model (5774988) and all results (6546817) on Zenodo. Full downloads were not performed.
Construct and compare calliope#
The sequence below contains upstream commands and proposed setup instructions, as labeled. Only commands identified in the execution evidence were run here; inspect the pinned source and prerequisites before running.
Upstream clone, HTTPS equivalent: git clone --recurse-submodules https://github.com/calliope-project/sector-coupled-euro-calliope.gitProposed paper pin: cd sector-coupled-euro-calliope; git checkout f81453a454424e36f45eb0a304f28f82f6e126d7; git submodule update --init --recursiveCurrent README route, which must be checked against the historical tag: conda env create -f environment.yamlCurrent README route: snakemake --use-conda --cores 1 (build model); inspect historical optimization entry point before running it
Expected outputs: Baseline model/result NetCDF, exact scenario YAML, solver log, SPORES result NetCDF ensemble and figures.
Comparison with the paper: Independent paper-table extraction agrees with author data; does not reproduce or validate a new solver run. Original optimization remains blocked by a verified expired solver license.
Execution result: Completed archived-data analysis:3969rows/441configurations, all9Table1ranges agree within rounding. Separate fresh open-solver follow-up confirmed GLPK5.0/Pyomo6.0.1 interface compatibility, then attempted the exact original model constructor; it failed a12GiB address-space budget with an8.05GiB Unicode allocation before native inputs or any study solve.
Execution evidence: Author-code attempt; Run record
Blockers: Exact author input constructor failed under a scoped12GiB process address-space limit:8.05GiB Unicode array allocation shape6491x8760,dtype<U38; peakRSS8993384KiB. This is a bounded resource outcome, not proof the host or an open solver inherently cannot solve the LP.; No compiled input arrays/LP exist in the two author archives, and native input construction did not complete; a fresh same-model backend/optimization was not reached.; Earlier originalGurobi9.1.2 prerequisite failed with an expired license. Existing GLPK5.0 is compatible with authorPyomo6.0.1; license expiry is not a general open-solver blocker.
Next action: Qualify unchanged input construction in a provisioned environment or verify a memory representation improvement, persist native arrays, then run the same LP using the declared open-solver protocol and compare objective/residuals/capacities. Keep archived-data analysis separate.
Gaps: Exact author input constructor failed under a scoped12GiB process address-space limit:8.05GiB Unicode array allocation shape6491x8760,dtype<U38; peakRSS8993384KiB. This is a bounded resource outcome, not proof the host or an open solver inherently cannot solve the LP.; No compiled input arrays/LP exist in the two author archives, and native input construction did not complete; a fresh same-model backend/optimization was not reached.; Earlier originalGurobi9.1.2 prerequisite failed with an expired license. Existing GLPK5.0 is compatible with authorPyomo6.0.1; license expiry is not a general open-solver blocker.
Sources for calliope#
- Article experimental procedures and explicit code/model/result deposits. (inspected 2026-11-15).
- README identifies exact 2021-11-29 paper workflow and subsequent assumption changes. (inspected 2026-11-15).
- Published result archive identity, filename and size. (inspected 2026-11-15).
- Historical tagged workflow and bespoke Calliope fork. (inspected 2026-11-15).
sumo#
SUMO — Can the published Luxembourg traffic-demand validation be repeated with the original scenario and simulator?
Route: Paper result candidate. Execution: Example executed. Source confidence: high.
Paper for sumo#
Lara Codeca; Raphael Frank; Sebastien Faye; Thomas Engel (2017). Luxembourg SUMO Traffic (LuST) Scenario: Traffic Demand Evaluation. 10.1109/MITS.2017.2666585.
Publication type: See primary metadata record; not classified here.. Primary metadata. Metadata verification: confirmed. Full-text access: Metadata and author README inspected; full validation tables not yet read.
Target in the paper: Author scenario and traffic-demand evaluation; full article target unavailable (institutional LuST2.pdf requires copy request).
Code and inputs for sumo#
Artifact or upstream source. Author scenario repository explicitly cites the paper and warns that validation used SUMO 0.26.
Source inspected: yes. Ref: master. Commit: 5edb7ecb9ad196c39172b6eb95d19ed789f4b1a6.
Files: README.md; dua.static.sumocfg; dua.actuated.sumocfg; due.static.sumocfg; due.actuated.sumocfg
Terms: MIT scenario; SUMO licensing follows its chosen historical release.
Environment and compute: SUMO 0.26 legacy executable and scenario pin; modern SUMO runs are adaptations requiring a new comparison. 24-hour city simulation on CPU; historical binary/build support required.
Inputs: Author network, demand, routes, rerouting configuration and original detector/traffic-count validation data.
Data access: Scenario source public; original observation/validation data availability must be checked in the article.
Construct and compare sumo#
The sequence below contains upstream commands and proposed setup instructions, as labeled. Only commands identified in the execution evidence were run here; inspect the pinned source and prerequisites before running.
Proposed source pin: git clone https://github.com/lcodeca/LuSTScenario.git; cd LuSTScenario; git checkout 5edb7ecb9ad196c39172b6eb95d19ed789f4b1a6Upstream (using SUMO 0.26): sumo -c dua.static.sumocfgUpstream alternatives: sumo -c dua.actuated.sumocfg; sumo -c due.static.sumocfg; sumo -c due.actuated.sumocfg
Expected outputs: Configured SUMO output files, simulation log, completed vehicle/trip counts, travel-time profiles and detector-flow comparison tables.
Comparison with the paper: New numerical outputs from actual historical simulator and complete city inputs; duration bounded. Completed-trip mean is censored, and three seeds do not validate traffic demand. No 2017 paper-target comparison could be made.
Execution result: Three author-model runs completed. Each inserted878vehicles; completed trips757/761/756, still running121/117/122. Conditional completed-trip means486.023778/487.896189/487.310847seconds.
Execution evidence: Author-code attempt; Run record
Blockers: 2017 paper LuST2.pdf and original observation/validation series unavailable; institutional record requires copy request.; Full-day empirical comparison not attempted without its target/data.
Next action: Obtain Codeca etal2017 DOI10.1109/MITS.2017.2666585 full text and validation data, declare a statistic/uncertainty rule, then extend run duration and repeat enough seeds.
Gaps: 2017 paper LuST2.pdf and original observation/validation series unavailable; institutional record requires copy request.; Full-day empirical comparison not attempted without its target/data.
Sources for sumo#
- Title/authors/year verified. (inspected 2026-11-15).
- Author command list, paper citation and explicit SUMO 0.26 validation restriction. (inspected 2026-11-15).
energyplus#
EnergyPlus — How closely do a building simulator's cooling-system outputs match the analytical BESTEST test cases?
Route: Paper model adaptation. Execution: Not run. Source confidence: medium.
Paper for energyplus#
Robert H. Henninger; Michael J. Witte; Drury B. Crawley (2004). Analytical and comparative testing of EnergyPlus using IEA HVAC BESTEST E100–E200 test suite. 10.1016/j.enbuild.2004.01.025.
Publication type: See primary metadata record; not classified here.. Primary metadata. Metadata verification: confirmed. Full-text access: Publisher abstract/metadata inspected; LBNL validation reports openly available.
Target in the paper: Single-zone DX cooling analytical/comparative E100–E200 tests; first candidate is the base cooling case. Current Standard 140-2020 successor tests must be kept separate from 2004 results.
Code and inputs for energyplus#
Artifact or upstream source. Official EnergyPlus/OpenStudio BESTEST workflow implementing later versions of the same benchmark family, with published result release assets; not the exact 2004 article code.
Source inspected: yes. Ref: v25.2. Commit: 80c3f2e56a501da996676086573f4c7f29fe9c5c.
Files: README.md; run_all_generate_reports.rb; results/; resources/
Terms: Read repository and bundled standard-resource terms before redistributing; EnergyPlus BSD-style license, OpenStudio dependency terms separate.
Environment and compute: OpenStudio 3.11.0 (includes EnergyPlus 25.2.0), Ruby 3.2.2, rubyXL and Parallel gems as documented at tag v25.2. CPU; upstream README estimates one test case under a minute and full suite under 30 minutes on typical computers; these estimates were not measured here.
Inputs: Packaged test-generation measures, weather/resource inputs, standard reporting spreadsheets and release reference results.
Data access: Official workflow and release outputs public. ASHRAE standard and resource-file reuse conditions separate; current test suite does not include every Standard 140 case.
Construct and compare energyplus#
The sequence below contains upstream commands and proposed setup instructions, as labeled. Only commands identified in the execution evidence were run here; inspect the pinned source and prerequisites before running.
Proposed pin: git clone https://github.com/NatLabRockies/BESTEST-GSR.git; cd BESTEST-GSR; git checkout 80c3f2e56a501da996676086573f4c7f29fe9c5cUpstream: ruby run_all_generate_reports.rb
Expected outputs: Generated OSM and IDF models, EnergyPlus results, workflow_results.csv, populated RESULTS5-*.xlsx and per-case result ZIPs.
Comparison with the paper: For the current-version exercise compare cooling energy/capacity outputs with the pinned release's corresponding test-case reference spreadsheet and analytical benchmark range. To reproduce the 2004 article recover EnergyPlus 1.0.x inputs and E100–E200 reference tables; a modern BESTEST pass cannot establish that historical reproduction.
Execution result: Sources and construction route inspected; no scientific result reproduced here.
Execution evidence: None recorded.
Blockers: Exact 2004 source/input files and numerical reference targets not recovered.; OpenStudio and Ruby dependencies are not installed here.
Next action: Build a current pinned BESTEST cooling-case exercise first; inspect the 2004 full text and LBNL reports to determine whether a historical reproduction is feasible.
Gaps: Exact 2004 source/input files and numerical reference targets not recovered.; OpenStudio and Ruby dependencies are not installed here.
Sources for energyplus#
- Article identity verified. (inspected 2026-11-15).
- Published E100–E200 analytical test scope and historical EnergyPlus versions. (inspected 2026-11-15).
- Actual current workflow commands, dependencies, outputs and missing-case limitations. (inspected 2026-11-15).
- Official historical validation reports. (inspected 2026-11-15).
wrf#
WRF — Does an idealized atmospheric dynamical core produce the Held–Suarez benchmark circulation?
Route: Paper model adaptation. Execution: Not run. Source confidence: high.
Paper for wrf#
Isaac M. Held; Max J. Suarez (1994). A Proposal for the Intercomparison of the Dynamical Cores of Atmospheric General Circulation Models. 10.1175/1520-0477(1994)075<1825:APFTIO>2.0.CO;2.
Publication type: See primary metadata record; not classified here.. Primary metadata. Metadata verification: confirmed. Full-text access: Publisher metadata and abstract plus official WRF source inspected; NOAA hosts author full text.
Target in the paper: Held–Suarez long-term circulation benchmark, presented as latitude/vertical cross sections of time-averaged zonal means.
Code and inputs for wrf#
Artifact or upstream source. Official WRF idealized implementation explicitly identifies the Held–Suarez article; a different dynamical core and coarser default resolution.
Source inspected: yes. Ref: master. Commit: 06d4240ae989cc3e50af412bb472df3d9048783c.
Files: test/em_heldsuarez/README.heldsuarez; test/em_heldsuarez/namelist.input
Terms: UCAR public-domain source notice; dependencies have their own terms.
Environment and compute: WRF ARW, supported C/Fortran compilers and NetCDF; compile idealized em_heldsuarez case. Serial configuration is available; MPI is optional for this controlled case. Long CPU integration; README notes its coarse default can run on a single processor. Higher resolution requires more compute.
Inputs: Bundled namelist and idealized dry global initial state; no real meteorological forcing or WPS geographic preprocessing required.
Data access: All idealized initial inputs bundled; numerical reference-field extraction from the article is still required.
Construct and compare wrf#
The sequence below contains upstream commands and proposed setup instructions, as labeled. Only commands identified in the execution evidence were run here; inspect the pinned source and prerequisites before running.
Proposed pin: git clone https://github.com/wrf-model/WRF.git; cd WRF; git checkout 06d4240ae989cc3e50af412bb472df3d9048783cOfficial idealized workflow: ./configure; ./compile em_heldsuarezOfficial idealized workflow: cd test/em_heldsuarez; ./ideal.exe; ./wrf.exe
Expected outputs: wrfinput file, time-indexed wrfout NetCDF and solver logs; postprocess zonal/time means of temperature and zonal wind.
Comparison with the paper: Default namelist runs 1000 days. Official README says averages use the last few hundred days and warns that default resolution is much coarser than the article. Compare latitude-pressure circulation structure and sensitivity to resolution/dissipation; no exact solution exists, so a pointwise exact-match tolerance is inappropriate. Fix averaging window and refinement comparison before claiming a benchmark reproduction.
Execution result: Sources and construction route inspected; no scientific result reproduced here.
Execution evidence: None recorded.
Blockers: No Fortran/NetCDF build environment qualified here.; Default resolution does not match the paper.; Averaging window, refinement settings and comparison criteria remain to be fixed.
Next action: Compile the idealized case in a dedicated scientific project, run a short setup check, then the documented climatology and a refinement study.
Gaps: No Fortran/NetCDF build environment qualified here.; Default resolution does not match the paper.; Averaging window, refinement settings and comparison criteria remain to be fixed.
Sources for wrf#
- Publisher title/authors/year/DOI and benchmark purpose. (inspected 2026-11-15).
- Exact paper link, 1000-day route, resolution and qualitative-comparison limits. (inspected 2026-11-15).
- Official idealized compile/initialize/run workflow. (inspected 2026-11-15).
opendrift#
OpenDrift — Where does a 5000-particle unballasted life-raft ensemble drift over 48 hours?
Route: Paper model adaptation. Execution: Not run. Source confidence: medium.
Paper for opendrift#
Knut-Frode Dagestad; Johannes Röhrs; Øyvind Breivik; Bjørn Ådlandsvik (2018). OpenDrift v1.0: a generic framework for trajectory modelling. 10.5194/gmd-11-1405-2018.
Publication type: See primary metadata record; not classified here.. Primary metadata. Metadata verification: confirmed. Full-text access: Open article; metadata and paper Leeway listing/Figure 2 description inspected.
Target in the paper: Section 3.1 OpenLeeway code listing and Figure 2: 5000 objectType 26 particles near longitude 4.5, latitude 60, radius 1000 m at 2017-07-01 12:00 for 48 hours.
Code and inputs for opendrift#
Artifact or upstream source. Current official Leeway example implements the same model family, but uses 2015 forcing and is not identical to the 2017 paper listing.
Source inspected: yes. Ref: master. Commit: ffd8e68b4be39ed42ed7187a875dd8eda65a3322.
Files: examples/example_leeway.py
Terms: GPL-2.0 code; ocean/wind forcing datasets and coastline have separate terms.
Environment and compute: OpenDrift with NetCDF/geospatial/plotting dependencies; for a historical result recover v1.0-era model and exact meteorological/ocean reader files. CPU ensemble; plotting/animation dependencies and external forcing downloads can dominate setup.
Inputs: Paper: 5000 life-rafts, 1000 m release radius, prescribed release time, 48-hour integration. Archive full wind/ocean forcing and coastline version; do not depend on a latest.nc endpoint.
Data access: Paper uses transient MET Norway THREDDS endpoints; exact historical files must be located. Current example uses different archived dates.
Construct and compare opendrift#
The sequence below contains upstream commands and proposed setup instructions, as labeled. Only commands identified in the execution evidence were run here; inspect the pinned source and prerequisites before running.
Proposed pin: git clone https://github.com/OpenDrift/opendrift.git; cd opendrift; git checkout ffd8e68b4be39ed42ed7187a875dd8eda65a3322Current upstream example: python examples/example_leeway.pyProposed paper route: transcribe Section 3.1 listing into a dedicated run script and replace transient reader endpoints only with byte-identical archived forcing; record all differences
Expected outputs: Particle trajectory NetCDF (add explicit export), leeway_example.png and animation; preserve seed, reader datasets and object-property table.
Comparison with the paper: Compare Figure 2 endpoint distribution, drift extent and stranding statistics after matching forcing and release settings. The article listing lacks a fixed random seed, so use ensemble distribution/uncertainty criteria, not a trajectory-by-trajectory byte comparison. Numeric target/tolerance requires figure extraction or original output.
Execution result: Sources and construction route inspected; no scientific result reproduced here.
Execution evidence: None recorded.
Blockers: Paper-era forcing snapshots and output not recovered.; Random seed was not specified in the printed listing.; Current example cannot reproduce Figure 2 without matching the original inputs.
Next action: Retrieve historical forcing, freeze release/model settings and a statistical comparison; otherwise publish a clearly labeled current-data adaptation.
Gaps: Paper-era forcing snapshots and output not recovered.; Random seed was not specified in the printed listing.; Current example cannot reproduce Figure 2 without matching the original inputs.
Sources for opendrift#
- Article identity and open access. (inspected 2026-11-15).
- Indexed Section 3.1 listing and Figure 2 relationship inspected through web; PDF not downloaded. (inspected 2026-11-15).
- Actual current official example and different forcing dates. (inspected 2026-11-15).
copasi#
COPASI — Can the published PER/TIM circadian-clock model be rerun before exploring chaotic and birhythmic regimes?
Route: Paper model adaptation. Execution: Not run. Source confidence: medium.
Paper for copasi#
Jean-Christophe Leloup; Albert Goldbeter (1999). Chaos and birhythmicity in a model for circadian oscillations of the PER and TIM proteins in drosophila. 10.1006/jtbi.1999.0924.
Publication type: See primary metadata record; not classified here.. Primary metadata. Metadata verification: confirmed. Full-text access: PubMed metadata verified; author-hosted 15-page full text inspected through web, not downloaded. COPASI native/SBML model inspected.
Target in the paper: Section 2 equations (1a)–(1j); Figure 2 on article p.449 reports sustained oscillations close to a 24-hour period and gives the full baseline parameter set.
Code and inputs for copasi#
Artifact or upstream source. Maintainer-distributed model explicitly cites Leloup and Goldbeter 1999. This is a later COPASI encoding, not an identified original author code archive.
Source inspected: yes. Ref: develop. Commit: ed1677836f03f1b56a2da227f3fa8ba793dd125a.
Files: TestSuite/distribution/CircadianClock.cps; TestSuite/sbml/CircadianClockExportedFromCopasi.xml
Terms: Artistic-2.0 COPASI repository; preserve model/article attribution and inspect model redistribution notices before vendoring.
Environment and compute: COPASI GUI or CopasiSE, pinned version. Existing .cps has deterministic LSODA time course: 96 hours, 960 steps, 0.1 hour step, relative tolerance 1e-6 and absolute tolerance 1e-12. Small deterministic CPU model; parameter-scan and attractor analysis need longer controlled runs.
Inputs: Bundled CPS initial state, PER/TIM kinetic parameters and reaction definitions; task is initially scheduled=false and has no report target.
Data access: Executable model source is openly inspectable in the official COPASI tree; author full text located at https://utc.ulb.be/ARTICLES/1999_Leloup_JTB.pdf but not downloaded.
Construct and compare copasi#
The sequence below contains upstream commands and proposed setup instructions, as labeled. Only commands identified in the execution evidence were run here; inspect the pinned source and prerequisites before running.
Proposed pin: git clone https://github.com/copasi/COPASI.git; cd COPASI; git checkout ed1677836f03f1b56a2da227f3fa8ba793dd125aGUI route: open TestSuite/distribution/CircadianClock.cps; inspect Time-Course settings; run the task and export concentrationsProposed batch route after explicitly scheduling the time-course task and defining a report in a copy: CopasiSE CircadianClock-run.cps
Expected outputs: 96-hour concentration table and plots; saved run CPS containing scheduled task/report path; solver/version provenance.
Comparison with the paper: Target Figure 2 baseline concentration trajectories and period near 24 hours. The inspected COPASI exported SBML has transcription repression K=0.9 for Mp/Mt, whereas the Figure 2 caption specifies K_IP=K_IT=1 nM; change these only in a documented paper-matched copy after auditing every other parameter and initial condition. Then compare oscillation period, amplitudes and phase relations, with transient treatment and figure-derived tolerances fixed before running. The article describes these parameter values as arbitrary physiological-range choices, so execution is not empirical biological validation.
Execution result: Sources and construction route inspected; no scientific result reproduced here.
Execution evidence: None recorded.
Blockers: Native distributed model does not exactly match Figure 2: transcription repression K is 0.9 rather than the paper's 1 nM. Other parameters and initial conditions still need a complete audit.; Figure-reference numeric trajectories/tolerances not recovered.; CPS task/report must be configured before batch execution; blindly running the unscheduled source file does not generate the target trace.
Next action: Create a paper-matched copy for Figure 2, document the K_IP/K_IT correction and all initial conditions, configure a numeric report, and compare period/amplitudes/phase with digitized or author-supplied reference data.
Gaps: Native distributed model does not exactly match Figure 2: transcription repression K is 0.9 rather than the paper's 1 nM. Other parameters and initial conditions still need a complete audit.; Figure-reference numeric trajectories/tolerances not recovered.; CPS task/report must be configured before batch execution; blindly running the unscheduled source file does not generate the target trace.
Sources for copasi#
- Verified publication identity and chaos/birhythmicity scope. (inspected 2026-11-15).
- Explicit 1999 paper annotation, actual time-course duration/steps/tolerances, initially unscheduled task. (inspected 2026-11-15).
- Equivalent exported SBML model inspected. (inspected 2026-11-15).
- Author full text inspected: Section 2 model equations and Figure 2 baseline parameters, target period and concentration traces. (inspected 2026-11-15).
tellurium#
Tellurium — Can a standards-based archive regenerate the paper's Drosophila cell-cycle simulations?
Route: Paper result candidate. Execution: Selected paper target compared. Source confidence: high.
Paper for tellurium#
J. Kyle Medley; Kiri Choi; Matthias König; Lucian Smith; Stanley Gu; Joseph Hellerstein; Stuart C. Sealfon; Herbert M. Sauro (2018). Tellurium notebooks—An environment for reproducible dynamical modeling in systems biology. 10.1371/journal.pcbi.1006220.
Publication type: See primary metadata record; not classified here.. Primary metadata. Metadata verification: confirmed. Full-text access: Open full text and actual COMBINE archive inspected.
Target in the paper: Figure4B Wee1c selected10points; exact archive from reference54
Code and inputs for tellurium#
Artifact or upstream source. Paper-author archive repository named in the article. Paper links a former demos/ path; current files reside in real-models/ and were actually inspected.
Source inspected: yes. Ref: master. Commit: eff025846c5b12ee4350f043623332158e05f262.
Files: real-models/calzone-fig1-fig3.omex; real-models/calzone-feedback-study.omex; real-models/calzone-log-plot.omex
Terms: Tellurium software Apache-2.0; individual model/COMBINE archive attribution and license must be checked independently.
Environment and compute: Tellurium, libRoadRunner, libSEDML, libCombine and plotting dependencies; paper-era Python 3.6 environment or separately labeled modern compatibility execution. Small biochemical ODE model on CPU; binary package compatibility may limit setup.
Inputs: Archive embeds SBML model and SED-ML model variants/solver protocol; inspected time courses use 250 and 180 units, 1000 points, KISAO:0000019; variant alters ksstg/kdstg and StgPc.
Data access: SBML and SED-ML bundled inside a public COMBINE archive; no clinical or human data needed.
Construct and compare tellurium#
The sequence below contains upstream commands and proposed setup instructions, as labeled. Only commands identified in the execution evidence were run here; inspect the pinned source and prerequisites before running.
Proposed pin: git clone https://github.com/0u812/tellurium-combine-archive-test-cases.git; cd tellurium-combine-archive-test-cases; git checkout eff025846c5b12ee4350f043623332158e05f262Tellurium execution API route to check against the chosen version: python -c "import tellurium as te; te.executeCombineArchive('real-models/calzone-fig1-fig3.omex', workingDir='run-calzone')"
Expected outputs: Executed SED-ML trajectories/plots and extracted model/experiment files, plus full environment, archive SHA256 and numeric result exports.
Comparison with the paper: Selected10-point Wee1c trace only; full archive ran but other5panels not numerically accepted.
Execution result: Exact author COMBINE archive executed unchanged, both1001-point timegrids and all21 data-generators finite. Selected Figure4B Wee1c comparison RMSE0.0026733313,maxabs0.0062600522; within prospective.03/.06 bounds. Current Tellurium runtime differs from2018 runtime.
Execution evidence: Author-code attempt; Run record
Blockers: Human review of raster calibration/tolerances is pending.; Original archive/model redistribution license not found in inspected archive/repository; input is fetched from author rather than relicensed/repackaged.
Next action: Review the10-point digitization and tolerance, then extend independent numerical reference comparisons to the other paper panels or obtain original numeric figure data.
Gaps: Human review of raster calibration/tolerances is pending.; Original archive/model redistribution license not found in inspected archive/repository; input is fetched from author rather than relicensed/repackaged.
Sources for tellurium#
- Full-text figure mapping, code availability and references 54/55. (inspected 2026-11-15).
- Actual current archive path; original paper's demos/ URL no longer locates this file. (inspected 2026-11-15).
- Downloaded and unzipped actual SBML/SED-ML protocol inspected. (inspected 2026-11-15).
brian2#
Brian 2 — Can the article's closed-loop eye-movement model be executed from the authors' notebook?
Route: Paper result candidate. Execution: Example executed. Source confidence: high.
Paper for brian2#
Marcel Stimberg; Romain Brette; Dan FM Goodman (2019). Brian 2, an intuitive and efficient neural simulator. 10.7554/eLife.47314.
Publication type: See primary metadata record; not classified here.. Primary metadata. Metadata verification: confirmed. Full-text access: Open full text, figure descriptions and author notebook inspected.
Target in the paper: Figure3b eye/object position trace; model code in Figure3c
Code and inputs for brian2#
Artifact or upstream source. Exact author repository explicitly identified by the eLife article for reproducing Figures 1–7; original notebook uses seed 79620.
Source inspected: yes. Ref: master. Commit: 73045cd61322e923e96883dc97c0848154f79f8c.
Files: example_2_eye_movements.ipynb; environment.yml; README.md
Terms: Author code/data dual licensed CC BY 4.0 or MIT; Brian 2 simulator CeCILL-2.1.
Environment and compute: Author environment pins Python 3.6, Brian2 2.2.2.1, scipy 1.2, sympy 1.3 and Plotly 3.6. A modern Python/Brian run must be labeled compatibility execution and compared to historical output. Small CPU model; optional compiler backend, no GPU required.
Inputs: Notebook equations/units, 20 retinal neurons, 2 motor neurons, one eye, seed 79620, 10-second integration; no external dataset.
Data access: All inputs in the public author notebook; Figure 8 performance data separately supplied by eLife and should not be confused with Figure 3 outputs.
Construct and compare brian2#
The sequence below contains upstream commands and proposed setup instructions, as labeled. Only commands identified in the execution evidence were run here; inspect the pinned source and prerequisites before running.
Upstream: git clone https://github.com/brian-team/brian2_paper_examples.gitProposed pin: cd brian2_paper_examples; git checkout 73045cd61322e923e96883dc97c0848154f79f8cUpstream: conda env create -f environment.ymlUpstream interactive route: open example_2_eye_movements.ipynb and execute cells in order; headless numeric execution can omit only the final Plotly presentation cell while preserving all model cells
Expected outputs: 10-second eye/object/target position arrays, retinal/motor spike trains and Figure 3-style plots; save model-cell hashes, seed, dt and runtime versions.
Comparison with the paper: Post-hoc exploratory comparison after prior modern output and first historical run; v2 corrects legend contamination. Historical deterministic repeat does not make overall analysis prospective.
Execution result: Unchanged eight author model cells ran on paper-version Brian2.2.2.1/Python3.6. Repeated historical arrays identical. Exploratory Figure3b comparison: eye RMSE0.0342156/max0.100732,object RMSE0.0195218/max0.0527203; within exploratory bounds. Code and actual published figure both use500ms; initial50ms transcription corrected.
Execution evidence: Author-code attempt; Run record
Blockers: Prospective paper-reference acceptance analysis not established; current raster analysis is explicitly exploratory.; Human scientific review of digitization, calibration and scope pending.
Next action: Review the reference extraction independently, then predeclare any stronger new experiment before new stochastic/model outputs; do not rename this exploratory comparison a full-paper reproduction.
Gaps: Prospective paper-reference acceptance analysis not established; current raster analysis is explicitly exploratory.; Human scientific review of digitization, calibration and scope pending.
Sources for brian2#
- Figure 3 case-study and simulation description; separate Figure 8 benchmarks. (inspected 2026-11-15).
- Article explicitly links author and eLife-archived code repositories. (inspected 2026-11-15).
- Actual equations, seed and 10-second run inspected. (inspected 2026-11-15).
- Original environment pin list. (inspected 2026-11-15).