Jae-Hyun Min
Research Subjects
Latest Publications
- Open Access
- 4 November 2026
- Published by MDPI AG in Preprints.org
MIsCORE-RIUD: Paper VII: Residual Classification, Minimum Common Support, and the Observation Boundary
Gravitational-wave detection pipelines are highly successful at identifying modeled and coherent signals, but the evidential meaning of post-model structure is a different question. This paper does not dispute established LIGO detections, nor does it claim that LIGO formally labels every residual sample as noise. It asks a narrower methodological question that became unavoidable during MIsCORE-RIUD: when H1 and L1 do not express the same merger-local residual structure in an identical lower-order form, does failure of a stronger representation-level consistency criterion justify rejecting the remaining structure before its higher-order organization is tested? RIUD answered operationally by requiring minimum concurrent support rather than global detector identity. A local candidate could not be carried by one detector alone: h1_z=min(H1 family), l1_z=min(L1 family), and network_z=min(h1_z,l1_z). Only candidates surviving both detector support conditions could enter the frozen NODE set, and only connected multiscale NODE organization could be promoted to PATH. Paper VII retrospectively audits that rule rather than inventing a new one. Four frozen evidence layers are decisive. First, Paper II gives 13/13 detector-resolved map compatibility and a fixed-path raw-strain population statistic of 1.171221 with 0/100,000 empirical-null exceedances, while a stricter eventwise coherence contract yields 0/13 FDR05 survivors. Second, a 96-event, 57,696-row detector-representation audit is directionally non-equivalent: median out-of-fold gain is -0.2044 for H1->L1 and +0.6496645697 for L1->H1; the final 9,999 whole-event null gives 0/9,999 exceedances (plus-one p=0.0001). Third, the NODE-conditioned median log10(L1/H1) high-band balance reverses from +0.4373 at 10 ms to -0.7916 at 80 ms. Fourth, controlled relational replay is invariant to detector swap, constant gain, and constant phase rotation, but sensitive to relative timing and time-varying phase; lower-order NODE occupancy can change while the higher-order PATH persists. The resulting claim is bounded but consequential: failure of global representational identity is not, by itself, evidence that the residual structure is noise. Under the tested RIUD contracts, concurrent local support followed by higher-order structural survival is a defensible promotion rule. Paper VII therefore closes a specific methodological vulnerability in the earlier RIUD lineage: the cross-detector rule was not an ad hoc relaxation introduced to rescue a result, but a frozen minimum-support contract later tested against the observation system itself.
- 4 November 2026
- Published by MDPI AG in Preprints.org
Open AccessMIsCORE-RIUD: Paper VII: Residual Classification, Minimum Common Support, and the Observation Boundary
Gravitational-wave detection pipelines are highly successful at identifying modeled and coherent signals, but the evidential meaning of post-model structure is a different question. This paper does not dispute established LIGO detections, nor does it claim that LIGO formally labels every residual sample as noise. It asks a narrower methodological question that became unavoidable during MIsCORE-RIUD: when H1 and L1 do not express the same merger-local residual structure in an identical lower-order form, does failure of a stronger representation-level consistency criterion justify rejecting the remaining structure before its higher-order organization is tested? RIUD answered operationally by requiring minimum concurrent support rather than global detector identity. A local candidate could not be carried by one detector alone: h1_z=min(H1 family), l1_z=min(L1 family), and network_z=min(h1_z,l1_z). Only candidates surviving both detector support conditions could enter the frozen NODE set, and only connected multiscale NODE organization could be promoted to PATH. Paper VII retrospectively audits that rule rather than inventing a new one. Four frozen evidence layers are decisive. First, Paper II gives 13/13 detector-resolved map compatibility and a fixed-path raw-strain population statistic of 1.171221 with 0/100,000 empirical-null exceedances, while a stricter eventwise coherence contract yields 0/13 FDR05 survivors. Second, a 96-event, 57,696-row detector-representation audit is directionally non-equivalent: median out-of-fold gain is -0.2044 for H1->L1 and +0.6496645697 for L1->H1; the final 9,999 whole-event null gives 0/9,999 exceedances (plus-one p=0.0001). Third, the NODE-conditioned median log10(L1/H1) high-band balance reverses from +0.4373 at 10 ms to -0.7916 at 80 ms. Fourth, controlled relational replay is invariant to detector swap, constant gain, and constant phase rotation, but sensitive to relative timing and time-varying phase; lower-order NODE occupancy can change while the higher-order PATH persists. The resulting claim is bounded but consequential: failure of global representational identity is not, by itself, evidence that the residual structure is noise. Under the tested RIUD contracts, concurrent local support followed by higher-order structural survival is a defensible promotion rule. Paper VII therefore closes a specific methodological vulnerability in the earlier RIUD lineage: the cross-detector rule was not an ad hoc relaxation introduced to rescue a result, but a frozen minimum-support contract later tested against the observation system itself.
- Open Access
- 27 October 2026
- Published by MDPI AG in Preprints.org
MIsCORE-RIUD: Paper VI—Reduction Asymmetry and Progression-Dependent Operator Reorganization in Black-Hole Merger Dynamics
The prior RIUD Paper VI tested whether an already-frozen merger-local state/history system and transition-rate law could be compressed into simpler frequency, timing, geometry, or source-native descriptions [3]. This manuscript performs a frozen secondary analysis of those results and separates four levels that are often conflated: descriptive organization, held-out predictive compression, source-side replacement, and progression-dependent composition of the effective transition rule. Among 98 path-bearing DEV99 events, frequency structure is strongly organized (K=2 census 90/8; FREQ_MAG 34/64; FREQ_SHAPE 63/35), yet the 91-event pre-transition out-of-fold test gives direction accuracy 0.3956 and net-rate Spearman rho=0.0416, with null p=0.683 and 0.264; frozen geometry adds no incremental support. A verified nested direct-GR representation gives DeltaLL=+0.2243 versus +5.3370 for the frozen RIUD reference, while richer GR-history/source-state branches do not reconstruct the complete memory-bearing rule. Conversely, source-native SXS PATH continuity survives center-step changes of 5, 10, and 15 ms in 10/10 cases and 20 ms in 9/10, and transient-to-QNM mismatch decreases across later source-native windows. The central secondary result is progression-dependent composition within the unchanged R40 rule: the relative absolute-contribution share of physical S versus history/run/age changes from 58.78/41.22% in PRE to 18.91/81.09% in POST, a 39.87 percentage-point shift, with six-cell max-T p=0.0076 and no phase-specific refit. A 95-event whole-waveform time-warp control is negative (p=1.0), and a 20-variable, 9,999-permutation source screen identifies CHI_A_NORMALIZED as the strongest corrected association (MaxT p=0.0321) without refitting R40. The combined result is a bounded reduction asymmetry: higher-order structural organization is more stable than the tested low-dimensional predictors or source-side replacements used to explain it, while the composition of the same effective rule reorganizes across progression. No GR violation, native-second R40 clock, universal S-QNM law, or unique microscopic derivation is claimed.
- 27 October 2026
- Published by MDPI AG in Preprints.org
Open AccessMIsCORE-RIUD: Paper VI—Reduction Asymmetry and Progression-Dependent Operator Reorganization in Black-Hole Merger Dynamics
The prior RIUD Paper VI tested whether an already-frozen merger-local state/history system and transition-rate law could be compressed into simpler frequency, timing, geometry, or source-native descriptions [3]. This manuscript performs a frozen secondary analysis of those results and separates four levels that are often conflated: descriptive organization, held-out predictive compression, source-side replacement, and progression-dependent composition of the effective transition rule. Among 98 path-bearing DEV99 events, frequency structure is strongly organized (K=2 census 90/8; FREQ_MAG 34/64; FREQ_SHAPE 63/35), yet the 91-event pre-transition out-of-fold test gives direction accuracy 0.3956 and net-rate Spearman rho=0.0416, with null p=0.683 and 0.264; frozen geometry adds no incremental support. A verified nested direct-GR representation gives DeltaLL=+0.2243 versus +5.3370 for the frozen RIUD reference, while richer GR-history/source-state branches do not reconstruct the complete memory-bearing rule. Conversely, source-native SXS PATH continuity survives center-step changes of 5, 10, and 15 ms in 10/10 cases and 20 ms in 9/10, and transient-to-QNM mismatch decreases across later source-native windows. The central secondary result is progression-dependent composition within the unchanged R40 rule: the relative absolute-contribution share of physical S versus history/run/age changes from 58.78/41.22% in PRE to 18.91/81.09% in POST, a 39.87 percentage-point shift, with six-cell max-T p=0.0076 and no phase-specific refit. A 95-event whole-waveform time-warp control is negative (p=1.0), and a 20-variable, 9,999-permutation source screen identifies CHI_A_NORMALIZED as the strongest corrected association (MaxT p=0.0321) without refitting R40. The combined result is a bounded reduction asymmetry: higher-order structural organization is more stable than the tested low-dimensional predictors or source-side replacements used to explain it, while the composition of the same effective rule reorganizes across progression. No GR violation, native-second R40 clock, universal S-QNM law, or unique microscopic derivation is claimed.
- Open Access
- 21 October 2026
- Published by MDPI AG in Preprints.org
MIsCORE-RIUD: Survival Without Representational Identity in Black-Hole Merger Dynamics
RIUD Papers V–VII tested different consequences of an already-frozen merger-local population system: survival outside the residual frontend, reduction to simpler predictive or source-side variables, and interchangeability of detector representations. The present manuscript performs a frozen secondary analysis of those results and asks a new question: are representation survival, representational identity, and predictive/mechanistic reduction empirically equivalent properties? They are evaluated as separate logical propositions without pooling their statistics or rebuilding the lower-order object.Paper V provides scoped survival outside the residual frontend: in the availability-defined RAW17 branch, ONPATH>OFFPATH holds in 15/17 raw H1/L1 pairs, NULL200 rarity in 16/17, and both criteria in 15/17. Its strongest tested known-physics surrogate recovers 82.6449% of the frozen likelihood improvement, but nested cross-validation does not yield a stable replacement.Paper VI then attacks reduction directly. In 91-event pre-transition out-of-fold prediction, direction accuracy is 0.3956 and net-rate Spearman rho=0.0416, with null p=0.683 and 0.264; geometry adds no incremental support. A verified direct-GR representation gives DeltaLL=+0.2243 against +5.3370 for the frozen reference, while source-native SXS PATH continuity survives center steps of 5, 10, and 15 ms in 10/10 cases and 20 ms in 9/10.Paper VII attacks identity. Event-blocked detector prediction gives median gain -0.2044 for H1-to-L1 and +0.6497 for L1-to-H1; the latter has 0/9,999 whole-event null exceedances, corresponding to a plus-one p=0.0001. Relative high-band balance reverses from median log10(L1/H1)=+0.4373 at 10 ms to -0.7916 at 80 ms, while higher-order PATH persists across a substantial lower-order support sweep.The combined frozen evidence therefore supports survival without representational identity and survival without stable reduction under the tested contracts. This is a relational invariance claim, not a universal waveform identity, a universal K-sector equation, a native SXS replay of R40, or a claim that lower-order representation is irrelevant.
- 21 October 2026
- Published by MDPI AG in Preprints.org
Open AccessMIsCORE-RIUD: Survival Without Representational Identity in Black-Hole Merger Dynamics
RIUD Papers V–VII tested different consequences of an already-frozen merger-local population system: survival outside the residual frontend, reduction to simpler predictive or source-side variables, and interchangeability of detector representations. The present manuscript performs a frozen secondary analysis of those results and asks a new question: are representation survival, representational identity, and predictive/mechanistic reduction empirically equivalent properties? They are evaluated as separate logical propositions without pooling their statistics or rebuilding the lower-order object.Paper V provides scoped survival outside the residual frontend: in the availability-defined RAW17 branch, ONPATH>OFFPATH holds in 15/17 raw H1/L1 pairs, NULL200 rarity in 16/17, and both criteria in 15/17. Its strongest tested known-physics surrogate recovers 82.6449% of the frozen likelihood improvement, but nested cross-validation does not yield a stable replacement.Paper VI then attacks reduction directly. In 91-event pre-transition out-of-fold prediction, direction accuracy is 0.3956 and net-rate Spearman rho=0.0416, with null p=0.683 and 0.264; geometry adds no incremental support. A verified direct-GR representation gives DeltaLL=+0.2243 against +5.3370 for the frozen reference, while source-native SXS PATH continuity survives center steps of 5, 10, and 15 ms in 10/10 cases and 20 ms in 9/10.Paper VII attacks identity. Event-blocked detector prediction gives median gain -0.2044 for H1-to-L1 and +0.6497 for L1-to-H1; the latter has 0/9,999 whole-event null exceedances, corresponding to a plus-one p=0.0001. Relative high-band balance reverses from median log10(L1/H1)=+0.4373 at 10 ms to -0.7916 at 80 ms, while higher-order PATH persists across a substantial lower-order support sweep.The combined frozen evidence therefore supports survival without representational identity and survival without stable reduction under the tested contracts. This is a relational invariance claim, not a universal waveform identity, a universal K-sector equation, a native SXS replay of R40, or a claim that lower-order representation is irrelevant.
- Open Access
- 13 October 2026
- Published by MDPI AG in Preprints.org
MIsCORE-RIUD: Paper IV — Population Dynamics and a Frozen Physical Clock
This study tests whether frozen connected multiscale paths across a 99-event development cohort form a higher-order population dynamical object. Structural paths occur in 98/99 events, but rarity is graded rather than binary: 58 path-bearing events satisfy the frozen rare-path criterion and 40 do not, while node-count complexity correlates with the frozen rarity coordinate (Spearman ρ = 0.6619802493). A search-aware discrete-family model fails (7,565/9,999 null exceedances; p = 0.7566), motivating a move from event labels to event-internal geometry.Intrinsic registration yields a 97-event atlas with 1,077 nodes and 980 transitions. PRE and POST dynamics are strongly asymmetric: PRE contains 730 nodes and 633 transitions across 90 events with macro R² = 0.131234 and minimum-fold R² = 0.106708, whereas POST contains 347 nodes and 250 transitions across 92 events with macro R² = 0.033268 and minimum-fold R² = -0.104449. Rate balance, re-entry, multistate flow, run index, state age, history, and frozen forward replay establish a recurrent memory-conditioned process.Broad physical-covariate screening produces large apparent fits, but leverage and search-aware controls reject them as population laws. Three-phase organization survives removal of physical terms: the physics-free F0 score is 0.5952381984 versus 0.5970683812 for the best physical candidate. Only after this structural freeze does a state-first physical search survive full selection correction, with 5/9,999 primary null exceedances (p = 0.0006) and 0/9,999 fixed-score exceedances (p = 0.0001).The frozen score then modifies transition rate with γ = 0.3162354634393393, so a one-unit increase in S multiplies the conditional hazard by exp(γ) = 1.371953262430. The frozen clock gives 3/999 primary exceedances under the plus-one rule (p = 0.004). All four developmental views retain positive likelihood gain, and Holdout5 retains positive no-refit likelihood gain in W1–W4; W1–W3 reproduce the frozen ordering direction, while W4 is underpowered at four scorable events.A transition-local depletion interval has generic maximum-gap strength (p = 0.99) but specific frozen transition-local placement at the tested resolution (p = 0.01). The surviving object is therefore an effective, history-bearing population state system with intrinsic registration and a source-dependent traversal-rate modifier.
- 13 October 2026
- Published by MDPI AG in Preprints.org
Open AccessMIsCORE-RIUD: Paper IV — Population Dynamics and a Frozen Physical Clock
This study tests whether frozen connected multiscale paths across a 99-event development cohort form a higher-order population dynamical object. Structural paths occur in 98/99 events, but rarity is graded rather than binary: 58 path-bearing events satisfy the frozen rare-path criterion and 40 do not, while node-count complexity correlates with the frozen rarity coordinate (Spearman ρ = 0.6619802493). A search-aware discrete-family model fails (7,565/9,999 null exceedances; p = 0.7566), motivating a move from event labels to event-internal geometry.Intrinsic registration yields a 97-event atlas with 1,077 nodes and 980 transitions. PRE and POST dynamics are strongly asymmetric: PRE contains 730 nodes and 633 transitions across 90 events with macro R² = 0.131234 and minimum-fold R² = 0.106708, whereas POST contains 347 nodes and 250 transitions across 92 events with macro R² = 0.033268 and minimum-fold R² = -0.104449. Rate balance, re-entry, multistate flow, run index, state age, history, and frozen forward replay establish a recurrent memory-conditioned process.Broad physical-covariate screening produces large apparent fits, but leverage and search-aware controls reject them as population laws. Three-phase organization survives removal of physical terms: the physics-free F0 score is 0.5952381984 versus 0.5970683812 for the best physical candidate. Only after this structural freeze does a state-first physical search survive full selection correction, with 5/9,999 primary null exceedances (p = 0.0006) and 0/9,999 fixed-score exceedances (p = 0.0001).The frozen score then modifies transition rate with γ = 0.3162354634393393, so a one-unit increase in S multiplies the conditional hazard by exp(γ) = 1.371953262430. The frozen clock gives 3/999 primary exceedances under the plus-one rule (p = 0.004). All four developmental views retain positive likelihood gain, and Holdout5 retains positive no-refit likelihood gain in W1–W4; W1–W3 reproduce the frozen ordering direction, while W4 is underpowered at four scorable events.A transition-local depletion interval has generic maximum-gap strength (p = 0.99) but specific frozen transition-local placement at the tested resolution (p = 0.01). The surviving object is therefore an effective, history-bearing population state system with intrinsic registration and a source-dependent traversal-rate modifier.
- Open Access
- 9 October 2026
- Published by MDPI AG in Preprints.org
MIsCORE-RIUD: Paper III — Control-Native Evidence for Partially Shared Dynamics in Recurrent Gravitational-Wave Post-Model Paths
Residual analyses of gravitational-wave observations commonly ask whether post-model data are compatible with detector noise or contain coherent excess power. Paper III asks a different question: once recurrent merger-local morphology has been converted into frozen connected multiscale paths, do those paths exhibit repeatable family-level dynamics that remain unusual when controls are allowed to construct their own maps and paths? The confirmatory panel contains 13 frozen events. Each event contributes 1,000 controls, and each control passes through preprocessing, feature construction, a free path search, trajectory formation, frozen family classification, and the hypothesis statistic. Three prelocked hypotheses are tested. In the canonical analysis, corrected empirical p-values are 0.005994 for multistep trajectories under raw/M5 (H1), 0.002997 for oscillatory trajectories under raw/M5 (H2), and 0.026973 for quasi-stationary trajectories under raw/M3 (H3). A separately regenerated internal computational replication uses 13,000 newly generated controls and 1,000 valid population draws with zero technical invalids; corrected p-values are 0.009990, 0.002997, and 0.045954, respectively. The two nulls are not pooled. H1 and H2 therefore support a shared dynamical representation across distinct morphology families, while H3 supports a separate and less stable regime. A separately frozen compact-law audit is negative after search correction (search-aware p=0.089910; fixed-law p=0.677323), preventing the family-level result from being promoted to one universal trajectory law. The supported conclusion is recurrent, partially shared path dynamics in the frozen panel. Physical origin, catalog-wide prevalence, exhaustive systematic exclusion, and Beyond-GR interpretation are not established.
- 9 October 2026
- Published by MDPI AG in Preprints.org
Open AccessMIsCORE-RIUD: Paper III — Control-Native Evidence for Partially Shared Dynamics in Recurrent Gravitational-Wave Post-Model Paths
Residual analyses of gravitational-wave observations commonly ask whether post-model data are compatible with detector noise or contain coherent excess power. Paper III asks a different question: once recurrent merger-local morphology has been converted into frozen connected multiscale paths, do those paths exhibit repeatable family-level dynamics that remain unusual when controls are allowed to construct their own maps and paths? The confirmatory panel contains 13 frozen events. Each event contributes 1,000 controls, and each control passes through preprocessing, feature construction, a free path search, trajectory formation, frozen family classification, and the hypothesis statistic. Three prelocked hypotheses are tested. In the canonical analysis, corrected empirical p-values are 0.005994 for multistep trajectories under raw/M5 (H1), 0.002997 for oscillatory trajectories under raw/M5 (H2), and 0.026973 for quasi-stationary trajectories under raw/M3 (H3). A separately regenerated internal computational replication uses 13,000 newly generated controls and 1,000 valid population draws with zero technical invalids; corrected p-values are 0.009990, 0.002997, and 0.045954, respectively. The two nulls are not pooled. H1 and H2 therefore support a shared dynamical representation across distinct morphology families, while H3 supports a separate and less stable regime. A separately frozen compact-law audit is negative after search correction (search-aware p=0.089910; fixed-law p=0.677323), preventing the family-level result from being promoted to one universal trajectory law. The supported conclusion is recurrent, partially shared path dynamics in the frozen panel. Physical origin, catalog-wide prevalence, exhaustive systematic exclusion, and Beyond-GR interpretation are not established.
- Open Access
- 23 September 2026
- Published by MDPI AG in Preprints.org
MIsCORE–RIUD: Paper II — Connected Multiscale Paths from Residual Morphology to Raw H1/L1 Strain
MIsCORE–RIUD Paper I established recurrent event-native merger-local morphology as an auditable observational object without assigning a unique physical origin. Paper II asks whether those local objects connect into ordered multiscale paths and whether the frozen higher-order object remains supported after projection into unsubtracted H1/L1 strain without reopening discovery freedom. A corrected 25-event atlas first showed that path construction was not automatic: 16/25 events contained at least one path, 9/25 contained no path, and 14/25 were multi-pattern. The canonical lock then fixed 13 events, 172 nodes, 26 H1/L1 raw-strain files, a merger-relative center window of −0.12 to +0.03 s, and the complete 10–60 ms width family, with no post-outcome event replacement, threshold retuning, or BEST_WIDTH selection. The frozen residual-derived coordinates were projected directly into 4-s, 4096-Hz H1/L1 strain in the locked 40–400 Hz band, with no raw-strain path reselection. The primary equal-weight statistic was 1.171221; 0/100,000 empirical-null draws reached the observation, giving p(+1)=9.9999×10⁻⁶, equivalent to approximately 4.265σ on a one-sided Gaussian tail only as a descriptive calibration of this finite empirical null. Every leave-one-event-out omission remained supported (maximum p=5.99994×10⁻⁵; ≈3.85σ one-sided), and support persisted without the 60-Hz notch and under a locked-coordinate ablation. A separate prelocked 62,400-job nuisance program constrained four coherent or independent artifact families while retaining weaker H1-only and L1-only transient alternatives. A selector-aware phase layer reran the exact selector on 2,600 real off-source controls; the observed equal-event real component was 0.3671572579932161, and 0/100,000 conditional phase-randomization draws exceeded it (p(+1)=9.9999×10⁻⁶; ≈4.265σ one-sided descriptive equivalent). The boundary remains explicit: population/map-level detector-resolved support is present, whereas the stricter eventwise coherence criterion yields zero FDR05 survivors and a separate 16-event physical-attribution audit yields zero global FDR05 survivors in both direct and visibility-controlled families. Paper II therefore establishes a connected multiscale observational path with fixed-coordinate raw-strain support and separate conditional phase directionality, while unique physical origin and exhaustive detector-artifact closure remain unresolved.
- 23 September 2026
- Published by MDPI AG in Preprints.org
Open AccessMIsCORE–RIUD: Paper II — Connected Multiscale Paths from Residual Morphology to Raw H1/L1 Strain
MIsCORE–RIUD Paper I established recurrent event-native merger-local morphology as an auditable observational object without assigning a unique physical origin. Paper II asks whether those local objects connect into ordered multiscale paths and whether the frozen higher-order object remains supported after projection into unsubtracted H1/L1 strain without reopening discovery freedom. A corrected 25-event atlas first showed that path construction was not automatic: 16/25 events contained at least one path, 9/25 contained no path, and 14/25 were multi-pattern. The canonical lock then fixed 13 events, 172 nodes, 26 H1/L1 raw-strain files, a merger-relative center window of −0.12 to +0.03 s, and the complete 10–60 ms width family, with no post-outcome event replacement, threshold retuning, or BEST_WIDTH selection. The frozen residual-derived coordinates were projected directly into 4-s, 4096-Hz H1/L1 strain in the locked 40–400 Hz band, with no raw-strain path reselection. The primary equal-weight statistic was 1.171221; 0/100,000 empirical-null draws reached the observation, giving p(+1)=9.9999×10⁻⁶, equivalent to approximately 4.265σ on a one-sided Gaussian tail only as a descriptive calibration of this finite empirical null. Every leave-one-event-out omission remained supported (maximum p=5.99994×10⁻⁵; ≈3.85σ one-sided), and support persisted without the 60-Hz notch and under a locked-coordinate ablation. A separate prelocked 62,400-job nuisance program constrained four coherent or independent artifact families while retaining weaker H1-only and L1-only transient alternatives. A selector-aware phase layer reran the exact selector on 2,600 real off-source controls; the observed equal-event real component was 0.3671572579932161, and 0/100,000 conditional phase-randomization draws exceeded it (p(+1)=9.9999×10⁻⁶; ≈4.265σ one-sided descriptive equivalent). The boundary remains explicit: population/map-level detector-resolved support is present, whereas the stricter eventwise coherence criterion yields zero FDR05 survivors and a separate 16-event physical-attribution audit yields zero global FDR05 survivors in both direct and visibility-controlled families. Paper II therefore establishes a connected multiscale observational path with fixed-coordinate raw-strain support and separate conditional phase directionality, while unique physical origin and exhaustive detector-artifact closure remain unresolved.
- Open Access
- 17 September 2026
- Published by MDPI AG in Preprints.org
MIsCORE–RIUD: Paper I — Recurrent Merger-Local Morphology in Gravitational-Wave Post-Model Data
Most post-model analyses begin with a named target: a waveform error, detector artifact, unmodeled transient, or parameterized departure from an established model. RIUD Paper I asks an earlier question: can recurrent merger-local structure be converted into a stable scientific object before its physical identity is known? The audited Paper-I development record comprises 612 preserved executions spanning successful, negative, unresolved, and technical branches. Development began with fixed physical phase-deformation models and reversed direction after those models failed to yield a stable population result. A residual-first reference in GW190521_074359 produced a two-stage merger-local morphology that survived synthetic-null, real off-source, waveform, posterior, and intrinsic-parameter-refit attacks, while a competing ringdown representation prevented the original basis from being treated as unique. The search was then generalized from fixed coordinates to an event-native family over local time, frequency, temporal width, inter-detector lag, and detector relation. In locked matched-background development tests, GW190521_074359, GW150914, and GW170104 each produced 0/1000 exceedances (p+1 = 0.000999); GW170814 entered as a non-blind catalog extension and also produced 0/1000 under its fixed-candidate validation. A later rigid temporal interpretation failed, separating the observational morphology from a stronger temporal story. The observation/scoring machinery itself was then audited. In a 227-row morphology/scorer audit, metadata-only leave-one-out classification reached 0.404, width-only classification reached 0.762, and frequency-width association gave Cramér’s V = 0.752, while direct mass dependence was negligible. Physical-sign posterior subtraction removed or weakened a stronger interpretation in 79/150 samples, whereas wrong-sign controls removed 0/12. An attractive 310.4-Hz branch failed an independent raw-FFT placebo in 3/3 target events and was discarded. The mature event-native score was deployed across a frozen 374-event native-strain universe, explicitly as a deployment census rather than 374 detections; the frozen score distribution comprised 31 FAIL, 162 PASS, 150 STRONG, and 31 VERY STRONG events. A later audit retained 374/374 finite scores, 343 H1/L1-ready events, 327 QC-valid events, and 300 prospective-eligible events after excluding 27 outcome-seen cases; the selection score was not retuned. Subsequent physical-identity tests rejected or bounded simple ppE-like, single-wavepacket, rigid common-center, and universal fixed-GR interpretations. A 29-event held-out cross-detector audit supported detector-related coordinate localization, but a conservative matched-SNR waveform-mismatch bank remained statistically compatible with the observed residual distribution. The surviving result is therefore deliberately bounded: RIUD Paper I establishes a recurrent event-native merger-local morphology as an auditable observational object, while leaving its unique physical cause unresolved.
- 17 September 2026
- Published by MDPI AG in Preprints.org
Open AccessMIsCORE–RIUD: Paper I — Recurrent Merger-Local Morphology in Gravitational-Wave Post-Model Data
Most post-model analyses begin with a named target: a waveform error, detector artifact, unmodeled transient, or parameterized departure from an established model. RIUD Paper I asks an earlier question: can recurrent merger-local structure be converted into a stable scientific object before its physical identity is known? The audited Paper-I development record comprises 612 preserved executions spanning successful, negative, unresolved, and technical branches. Development began with fixed physical phase-deformation models and reversed direction after those models failed to yield a stable population result. A residual-first reference in GW190521_074359 produced a two-stage merger-local morphology that survived synthetic-null, real off-source, waveform, posterior, and intrinsic-parameter-refit attacks, while a competing ringdown representation prevented the original basis from being treated as unique. The search was then generalized from fixed coordinates to an event-native family over local time, frequency, temporal width, inter-detector lag, and detector relation. In locked matched-background development tests, GW190521_074359, GW150914, and GW170104 each produced 0/1000 exceedances (p+1 = 0.000999); GW170814 entered as a non-blind catalog extension and also produced 0/1000 under its fixed-candidate validation. A later rigid temporal interpretation failed, separating the observational morphology from a stronger temporal story. The observation/scoring machinery itself was then audited. In a 227-row morphology/scorer audit, metadata-only leave-one-out classification reached 0.404, width-only classification reached 0.762, and frequency-width association gave Cramér’s V = 0.752, while direct mass dependence was negligible. Physical-sign posterior subtraction removed or weakened a stronger interpretation in 79/150 samples, whereas wrong-sign controls removed 0/12. An attractive 310.4-Hz branch failed an independent raw-FFT placebo in 3/3 target events and was discarded. The mature event-native score was deployed across a frozen 374-event native-strain universe, explicitly as a deployment census rather than 374 detections; the frozen score distribution comprised 31 FAIL, 162 PASS, 150 STRONG, and 31 VERY STRONG events. A later audit retained 374/374 finite scores, 343 H1/L1-ready events, 327 QC-valid events, and 300 prospective-eligible events after excluding 27 outcome-seen cases; the selection score was not retuned. Subsequent physical-identity tests rejected or bounded simple ppE-like, single-wavepacket, rigid common-center, and universal fixed-GR interpretations. A 29-event held-out cross-detector audit supported detector-related coordinate localization, but a conservative matched-SNR waveform-mismatch bank remained statistically compatible with the observed residual distribution. The surviving result is therefore deliberately bounded: RIUD Paper I establishes a recurrent event-native merger-local morphology as an auditable observational object, while leaving its unique physical cause unresolved.