Dataset Viewer
The dataset viewer is not available for this split.
Cannot load the dataset split (in streaming mode) to extract the first rows.
Error code: StreamingRowsError
Exception: CastError
Message: Couldn't cast
pump_h8: struct<success: bool, wr: double, gi: double, residual: double, nfev: int64, m: int64, p: int64, nz: (... 31 chars omitted)
child 0, success: bool
child 1, wr: double
child 2, gi: double
child 3, residual: double
child 4, nfev: int64
child 5, m: int64
child 6, p: int64
child 7, nz: int64
child 8, kx: double
child 9, kz: double
pump_h9: struct<success: bool, wr: double, gi: double, residual: double, nfev: int64, m: int64, p: int64, nz: (... 31 chars omitted)
child 0, success: bool
child 1, wr: double
child 2, gi: double
child 3, residual: double
child 4, nfev: int64
child 5, m: int64
child 6, p: int64
child 7, nz: int64
child 8, kx: double
child 9, kz: double
pump_h10: struct<success: bool, wr: double, gi: double, residual: double, nfev: int64, m: int64, p: int64, nz: (... 31 chars omitted)
child 0, success: bool
child 1, wr: double
child 2, gi: double
child 3, residual: double
child 4, nfev: int64
child 5, m: int64
child 6, p: int64
child 7, nz: int64
child 8, kx: double
child 9, kz: double
daughter_d1: struct<x: double, minimum_singular_value: double, optimal_gain: double, helical_gain: double, work_e (... 145 chars omitted)
child 0, x: double
child 1, minimum_singular_value: double
child 2, optimal_gain: double
child 3, helical_gain: double
child 4, work_e: double
child 5, work_d: double
child 6, work_t: double
child 7, work_alpha: double
child 8, d_positive_fraction: double
child 9, m:
...
hars omitted)
child 0, background_electron_density_m_minus_3: double
child 1, background_magnetic_field_tesla: double
child 2, deuteron_temperature_eV: double
child 3, electron_temperature_eV: double
child 4, grid_cells: int64
child 5, minority_proton_fraction: double
child 6, minority_ring_energy_eV: double
child 7, phase_code_modulus: int64
child 8, physical_claim_scope: string
child 9, pump_mode_numbers: list<item: int64>
child 0, item: int64
child 10, purpose: string
child 11, smoke_target_steps: int64
child 12, spatial_model: string
current_position: struct<constant_density_control_valid: bool, custom_assay_blocker: string, custom_assay_produced_phy (... 246 chars omitted)
child 0, constant_density_control_valid: bool
child 1, custom_assay_blocker: string
child 2, custom_assay_produced_physical_observables: bool
child 3, field_probe_is_sufficient_cause_of_failure: bool
child 4, lower_level_ladder_executed: bool
child 5, minority_ring_is_sufficient_cause_of_failure: bool
child 6, next_scientific_claim_available: bool
child 7, upstream_software_runtime_demonstrated: bool
provisional_assay_deck: struct<byte_count: int64, present: bool, sha256: string>
child 0, byte_count: int64
child 1, present: bool
child 2, sha256: string
prepared_but_not_run: list<item: struct<branch_id: string, purpose: string>>
child 0, item: struct<branch_id: string, purpose: string>
child 0, branch_id: string
child 1, purpose: string
to
{'assay_scope': {'background_electron_density_m_minus_3': Value('float64'), 'background_magnetic_field_tesla': Value('float64'), 'deuteron_temperature_eV': Value('float64'), 'electron_temperature_eV': Value('float64'), 'grid_cells': Value('int64'), 'minority_proton_fraction': Value('float64'), 'minority_ring_energy_eV': Value('float64'), 'phase_code_modulus': Value('int64'), 'physical_claim_scope': Value('string'), 'pump_mode_numbers': List(Value('int64')), 'purpose': Value('string'), 'smoke_target_steps': Value('int64'), 'spatial_model': Value('string')}, 'claim_boundary': {'autonomy_implies_correctness': Value('bool'), 'experimental_validation_claimed': Value('bool'), 'fusion_breakthrough_claimed': Value('bool'), 'mathematics_asserted_correct': Value('bool'), 'physics_asserted_correct': Value('bool'), 'record_is_process_evidence': Value('bool'), 'software_failure_treated_as_physical_evidence': Value('bool')}, 'current_position': {'constant_density_control_valid': Value('bool'), 'custom_assay_blocker': Value('string'), 'custom_assay_produced_physical_observables': Value('bool'), 'field_probe_is_sufficient_cause_of_failure': Value('bool'), 'lower_level_ladder_executed': Value('bool'), 'minority_ring_is_sufficient_cause_of_failure': Value('bool'), 'next_scientific_claim_available': Value('bool'), 'upstream_software_runtime_demonstrated': Value('bool')}, 'direction_changes': List({'from': Value('string'), 'to': Value('string'), 'trigger': Value('string')}), 'generated_utc': Val
...
56': Value('string')}, 'bounded_interpretation': Value('string'), 'branch_id': Value('string'), 'classification': Value('string'), 'input_validation_failure_detected': Value('bool'), 'intervention': Value('string'), 'maximum_step_marker': Value('int64'), 'post_state': {'byte_count': Value('int64'), 'present': Value('bool'), 'sha256': Value('string')}, 'reached_step_2': Value('bool'), 'run_log_byte_count': Value('int64'), 'run_log_sha256': Value('string'), 'stack_smashing_detected': Value('bool'), 'used_inputs': {'byte_count': Value('int64'), 'present': Value('bool'), 'sha256': Value('string')}}), 'schema': Value('string'), 'source_artifacts': List({'artifact_id': Value('string'), 'byte_count': Value('int64'), 'present': Value('bool'), 'sha256': Value('string')}), 'source_pins': {'amrex_archive_sha256': Value('string'), 'amrex_commit': Value('string'), 'warpx_1d_executable_sha256': Value('string'), 'warpx_archive_sha256': Value('string'), 'warpx_commit': Value('string')}, 'upstream_runtime_parity': {'run_log': {'byte_count': Value('int64'), 'classification': Value('string'), 'input_validation_failure_detected': Value('bool'), 'maximum_step_marker': Value('int64'), 'present': Value('bool'), 'reached_step_1': Value('bool'), 'reached_step_2': Value('bool'), 'sha256': Value('string'), 'stack_smashing_detected': Value('bool'), 'successful_completion_marker_detected': Value('bool')}, 'used_inputs': {'byte_count': Value('int64'), 'present': Value('bool'), 'sha256': Value('string')}}}
because column names don't match
Traceback: Traceback (most recent call last):
File "/src/services/worker/src/worker/utils.py", line 147, in get_rows_or_raise
return get_rows(
dataset=dataset,
...<4 lines>...
column_names=column_names,
)
File "/src/libs/libcommon/src/libcommon/utils.py", line 272, in decorator
return func(*args, **kwargs)
File "/src/services/worker/src/worker/utils.py", line 127, in get_rows
rows_plus_one = list(itertools.islice(safe_iter(ds, dataset=dataset), rows_max_number + 1))
File "/src/services/worker/src/worker/utils.py", line 483, in safe_iter
yield from ds.decode(False) if ds.features else ds
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2840, in __iter__
for key, example in ex_iterable:
^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2373, in __iter__
for key, pa_table in self._iter_arrow():
~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2398, in _iter_arrow
for key, pa_table in self.ex_iterable._iter_arrow():
~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 536, in _iter_arrow
for key, pa_table in iterator:
^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 419, in _iter_arrow
for key, pa_table in self.generate_tables_fn(**gen_kwags):
~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 343, in _generate_tables
self._cast_table(pa_table, json_field_paths=json_field_paths),
~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 132, in _cast_table
pa_table = table_cast(pa_table, self.info.features.arrow_schema)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2378, in table_cast
return cast_table_to_schema(table, schema)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2306, in cast_table_to_schema
raise CastError(
...<3 lines>...
)
datasets.table.CastError: Couldn't cast
pump_h8: struct<success: bool, wr: double, gi: double, residual: double, nfev: int64, m: int64, p: int64, nz: (... 31 chars omitted)
child 0, success: bool
child 1, wr: double
child 2, gi: double
child 3, residual: double
child 4, nfev: int64
child 5, m: int64
child 6, p: int64
child 7, nz: int64
child 8, kx: double
child 9, kz: double
pump_h9: struct<success: bool, wr: double, gi: double, residual: double, nfev: int64, m: int64, p: int64, nz: (... 31 chars omitted)
child 0, success: bool
child 1, wr: double
child 2, gi: double
child 3, residual: double
child 4, nfev: int64
child 5, m: int64
child 6, p: int64
child 7, nz: int64
child 8, kx: double
child 9, kz: double
pump_h10: struct<success: bool, wr: double, gi: double, residual: double, nfev: int64, m: int64, p: int64, nz: (... 31 chars omitted)
child 0, success: bool
child 1, wr: double
child 2, gi: double
child 3, residual: double
child 4, nfev: int64
child 5, m: int64
child 6, p: int64
child 7, nz: int64
child 8, kx: double
child 9, kz: double
daughter_d1: struct<x: double, minimum_singular_value: double, optimal_gain: double, helical_gain: double, work_e (... 145 chars omitted)
child 0, x: double
child 1, minimum_singular_value: double
child 2, optimal_gain: double
child 3, helical_gain: double
child 4, work_e: double
child 5, work_d: double
child 6, work_t: double
child 7, work_alpha: double
child 8, d_positive_fraction: double
child 9, m:
...
hars omitted)
child 0, background_electron_density_m_minus_3: double
child 1, background_magnetic_field_tesla: double
child 2, deuteron_temperature_eV: double
child 3, electron_temperature_eV: double
child 4, grid_cells: int64
child 5, minority_proton_fraction: double
child 6, minority_ring_energy_eV: double
child 7, phase_code_modulus: int64
child 8, physical_claim_scope: string
child 9, pump_mode_numbers: list<item: int64>
child 0, item: int64
child 10, purpose: string
child 11, smoke_target_steps: int64
child 12, spatial_model: string
current_position: struct<constant_density_control_valid: bool, custom_assay_blocker: string, custom_assay_produced_phy (... 246 chars omitted)
child 0, constant_density_control_valid: bool
child 1, custom_assay_blocker: string
child 2, custom_assay_produced_physical_observables: bool
child 3, field_probe_is_sufficient_cause_of_failure: bool
child 4, lower_level_ladder_executed: bool
child 5, minority_ring_is_sufficient_cause_of_failure: bool
child 6, next_scientific_claim_available: bool
child 7, upstream_software_runtime_demonstrated: bool
provisional_assay_deck: struct<byte_count: int64, present: bool, sha256: string>
child 0, byte_count: int64
child 1, present: bool
child 2, sha256: string
prepared_but_not_run: list<item: struct<branch_id: string, purpose: string>>
child 0, item: struct<branch_id: string, purpose: string>
child 0, branch_id: string
child 1, purpose: string
to
{'assay_scope': {'background_electron_density_m_minus_3': Value('float64'), 'background_magnetic_field_tesla': Value('float64'), 'deuteron_temperature_eV': Value('float64'), 'electron_temperature_eV': Value('float64'), 'grid_cells': Value('int64'), 'minority_proton_fraction': Value('float64'), 'minority_ring_energy_eV': Value('float64'), 'phase_code_modulus': Value('int64'), 'physical_claim_scope': Value('string'), 'pump_mode_numbers': List(Value('int64')), 'purpose': Value('string'), 'smoke_target_steps': Value('int64'), 'spatial_model': Value('string')}, 'claim_boundary': {'autonomy_implies_correctness': Value('bool'), 'experimental_validation_claimed': Value('bool'), 'fusion_breakthrough_claimed': Value('bool'), 'mathematics_asserted_correct': Value('bool'), 'physics_asserted_correct': Value('bool'), 'record_is_process_evidence': Value('bool'), 'software_failure_treated_as_physical_evidence': Value('bool')}, 'current_position': {'constant_density_control_valid': Value('bool'), 'custom_assay_blocker': Value('string'), 'custom_assay_produced_physical_observables': Value('bool'), 'field_probe_is_sufficient_cause_of_failure': Value('bool'), 'lower_level_ladder_executed': Value('bool'), 'minority_ring_is_sufficient_cause_of_failure': Value('bool'), 'next_scientific_claim_available': Value('bool'), 'upstream_software_runtime_demonstrated': Value('bool')}, 'direction_changes': List({'from': Value('string'), 'to': Value('string'), 'trigger': Value('string')}), 'generated_utc': Val
...
56': Value('string')}, 'bounded_interpretation': Value('string'), 'branch_id': Value('string'), 'classification': Value('string'), 'input_validation_failure_detected': Value('bool'), 'intervention': Value('string'), 'maximum_step_marker': Value('int64'), 'post_state': {'byte_count': Value('int64'), 'present': Value('bool'), 'sha256': Value('string')}, 'reached_step_2': Value('bool'), 'run_log_byte_count': Value('int64'), 'run_log_sha256': Value('string'), 'stack_smashing_detected': Value('bool'), 'used_inputs': {'byte_count': Value('int64'), 'present': Value('bool'), 'sha256': Value('string')}}), 'schema': Value('string'), 'source_artifacts': List({'artifact_id': Value('string'), 'byte_count': Value('int64'), 'present': Value('bool'), 'sha256': Value('string')}), 'source_pins': {'amrex_archive_sha256': Value('string'), 'amrex_commit': Value('string'), 'warpx_1d_executable_sha256': Value('string'), 'warpx_archive_sha256': Value('string'), 'warpx_commit': Value('string')}, 'upstream_runtime_parity': {'run_log': {'byte_count': Value('int64'), 'classification': Value('string'), 'input_validation_failure_detected': Value('bool'), 'maximum_step_marker': Value('int64'), 'present': Value('bool'), 'reached_step_1': Value('bool'), 'reached_step_2': Value('bool'), 'sha256': Value('string'), 'stack_smashing_detected': Value('bool'), 'successful_completion_marker_detected': Value('bool')}, 'used_inputs': {'byte_count': Value('int64'), 'present': Value('bool'), 'sha256': Value('string')}}}
because column names don't matchNeed help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
Ouroboros autonomous fusion research attempt
Process, pivots, falsifiers, and the current boundary
Claim boundary — read this first. This repository documents a research process. It does not claim that the mathematics is correct, that the plasma model is physically correct, that any fusion mechanism was validated, that net energy is plausible, or that a breakthrough occurred. A software failure is not physical evidence. Autonomy is not correctness.
This is a public record of an autonomous research attempt driven by Ouroboros. Ouroboros selected and re-ranked hypotheses, generated falsifiers, and changed routes when evidence contradicted the active explanation. OpenAI Codex served as its coding and operations agent. The human owner set the objective and retained stop authority; on 2026-08-31 the owner explicitly stopped the run, and the publication audit verified that no matching Windows or WSL fusion worker remained active.
The point of the release is the loop:
- represent a problem structurally;
- attach provisional meaning only where needed;
- derive a test that could fail;
- retain negative, invalid, and incomplete outcomes;
- redirect from the observed failure class;
- stop claims at the boundary of the evidence.
The original white paper remains available as Markdown, PDF, and DOCX. Its machine-readable record is data/results.json. This update adds the post-publication autonomous runtime campaign in data/autonomous_research_update_2026-08-31.json.
What changed after the first release
The earlier work started from a low-energy anomaly, held that route when a normalization or instrumentation seam became more plausible than new physics, and then built and attacked a separate ion-cyclotron resonance candidate. Corrections to the mass model and gain labels, negative controls, and the lack of decisive public phase-resolved data prevented that candidate from being promoted as a physical result.
Ouroboros then redirected from analysis of existing papers and public data to a source-pinned computational instrument. The instrument was intentionally narrow: a provisional 1D3V route-observability assay, not a reactor model and not a proof of nonlinear coupling.
New computational data
1. Source-pinned runtime parity
WarpX commit 95e97221f4d422c2c17f40319ffa6c6de2574418 and AMReX commit f2cbba4afc9eb3cc0a68b311112def524ada8cd3 were built into a 1D executable with SHA-256 565ba3381f8415e55b72bae4ddb50198b217d425ba543e9bfc1dd84f51bf3f4d. The upstream bounded example reached step 100 without the later failure signature. Its log SHA-256 is afbfc722ee6a4946d8c66db353b511f5c8f8febad7039c0224a8c24e85db9cad.
This establishes only that the pinned build can run the upstream bounded example. It does not validate the custom deck or any physics claim.
2. Provisional custom assay
The custom deck used 512 cells, an electron density of 1.3e19 m^-3, 1 keV electron and deuteron temperatures, a 1.99 T transverse field, a 1e-3 minority-proton fraction, 150 keV nominal ring energy, pump mode numbers 8/9/10, and a phase code modulo 7. The smoke target was two steps. These are test inputs, not accepted physical parameters.
| Branch | Intervention | Observed class | Highest step | Step 2? | Log SHA-256 |
|---|---|---|---|---|---|
| Baseline | FieldProbe + minority ring | step-1 runtime memory failure | 1 | no | 6c653e3d0123… |
| No FieldProbe | remove FieldProbe | step-1 runtime memory failure | 1 | no | 8ec4625f096f… |
| No ring | remove FieldProbe and ring | step-1 runtime memory failure | 1 | no | 6c653e3d0123… |
| Constant density | replace parsed profiles after removing probe/ring | invalid input control | — | no | 89346c4d192f… |
| Gaussian ring | replace ring parser; remove FieldProbe | step-1 runtime memory failure | 1 | no | 6c653e3d0123… |
Four branches initialized and entered step 1, then emitted the same runtime memory-failure signature before step 2. The constant-density branch is different: it failed input validation because explicit densities were missing. It is retained as an invalid control, not counted as evidence against a physical or software hypothesis.
No custom branch produced the intended physical observables. Therefore there is no nonlinear-coupling result to interpret.
How Ouroboros changed direction
| Evidence or failure | Direction change |
|---|---|
| Low-energy anomaly became more consistent with a normalization/instrument seam | Hold the anomaly; search for an independently falsifiable plasma route |
| Corrected mass/gain bookkeeping and public-data closure failure | Publish a provisional process record instead of promoting the resonance candidate |
| Decisive phase-resolved data were absent publicly | Build a source-pinned runtime and a deliberately narrow instrument |
| The first Windows-to-WSL traversal failed before launch | Move hash, launch, and post-state checks into WSL-native paths |
| Baseline entered step 1 and failed | Remove FieldProbe |
| No-FieldProbe branch failed the same way | Test ring removal and ring-parser substitution |
| No-ring and Gaussian-ring branches failed the same way | Move below feature-level explanations toward time-step, species, field, and allocation controls |
| Constant-density control lacked required input | Withdraw that causal claim and redesign the control |
| Human owner stopped the campaign | Verify no matching workers remain; preserve the next branches as unrun |
The repeated step-1 failure makes FieldProbe alone and the minority ring alone insufficient explanations for this specific software failure. It does not validate the remaining mathematics or physics.
Where the process stands
- The source-pinned WarpX build ran its upstream bounded example through step 100.
- The custom 1D3V instrument is blocked at its first time advance by a runtime memory failure.
- No custom physical observable was produced.
- The constant-density control was invalid and its earlier interpretation was withdrawn.
- Five lower-level controls were prepared but not run: a valid constant-density branch, smaller time step, proton-mass substitution, zero background field, and smaller grid.
- The campaign is stopped. This page records the checkpoint; it does not imply the scientific question is solved.
Audit boundary
This repository remains an audit-only process record, not an end-to-end reproduction package. The new JSON contains exact log hashes, byte counts, branch classifications, source pins, and the unexecuted queue. Raw local logs and private control-plane artifacts are intentionally not published. Third-party sources remain under their upstream terms.
License
Original prose and derived tables: CC BY 4.0. Original code already present under code/: MIT. Third-party materials are referenced, not redistributed.
- Downloads last month
- 47