遇见数据集

Rotation Field of the Cosmic Microwave Background — Multipole Structure & Model Selection (v1.5)

收藏
Zenodo2026-06-22 更新2026-05-26 收录
官方服务:

资源简介:

Summary Version v1.5 performs model selection and multipole structure analysis on the same locked α(n̂) rotation field used in v1.43 and v1.44.No α reconstruction is performed and no calibration or preprocessing changes occur.The purpose of v1.5 is to determine what spatial structure best explains the rotation field. All models use: • the same locked α(n̂) map (alpha_lowL_combined_v141.fits), • the same apodized polarization mask (NSIDE=512, f_sky≈0.467), • the same low-ℓ band (2–10). The anisotropy detection from v1.43 (4.38σ empirical exceedance over 165,000 null simulations) remains unchanged. Models tested: Model FormA — Dipole α(n̂) = a₀ + d · n̂B — Dipole + traceless quadrupole α(n̂) = a₀ + d · n̂ + Q : (n̂n̂)C — Harmonic in multipole space α(ℓ) = A sin(kℓ + φ) + C + G/ℓ Results • Dipole-only is decisively rejected. Model selection strongly prefers a dipole + quadrupole structure: ΔAIC ≈ +2.16×10⁵, ΔBIC ≈ +2.16×10⁵. • The rotation spectrum α(ℓ) contains a harmonic periodicity. Note on Δℓ: Earlier harmonic fits (v1.2) favored Δℓ ≈ 360, using a different fitting methodology and model-selection framework than v1.5. In v1.5, the harmonic frequency scan was initially restricted to ±20% around a starting estimate for model-selection stability (AIC/BIC), and the bootstrap distribution reached the upper edge of that range at Δℓ ≈ 454.7. A direct head-to-head comparison of Δℓ ≈ 363.8 and Δℓ ≈ 454.7, refit on the identical α(ℓ) series, confirms that the longer period yields lower residual sum of squares and improved AIC/BIC. Δℓ ≈ 454.7 is therefore the better-supported value between the two — not merely an artifact of the scan boundary — though the scan was not widened beyond that ±20% range, so a still-better solution outside the tested window cannot be ruled out. This differs from the Δℓ ≈ 360 periodicity reported in v1.2, reflecting the different fitting methodology and model-selection framework applied in v1.5 rather than a change in the underlying data. Best-fit periodicity: Δℓ = 454.7 68% confidence interval (bootstrap): 433.07 – 454.72 (the median and upper bound of the bootstrap distribution coincide at the edge of the tested k-scan range.) • Δℓ is stable under: – different bin widths (Δℓ_bin = 12, 15, 20) – mask apodization changes (+0.5°, +1.5°) – posterior predictive checks (best-fit model reproduces the amplitude envelope) • Cross-validated prediction error decreases from Model A → Model B → Model C. v1.5 does not modify the detection established in v1.43/v1.44; it characterizes the structure of α(n̂). Archive contents(All files are timestamped; no file is ever overwritten.) • dipole_fit_modelA.json/png• quadrupole_fit_modelB.json• harmonic_modelC_fit_*.json/png• harmonic_modelC_series_*.csv• v1_5_metrics_table_*.json/txt (AIC/BIC/logL/CV-RMSE/residual ratios; the Model C row reflects the original Δℓ ≈ 363.8 fit — see numbers table below for the confirmed Δℓ ≈ 454.7 comparison)• v1_5_numbers_table_*.txt (human-readable summary, including the head-to-head Δℓ verification)• v1_5_README_paragraph_*.txt (text for manuscripts / papers)• v1_5_bundle_manifest_*.json (SHA-256 for every file) Reproducibility • Same α(n̂) FITS map and same mask from v1.43/v1.44• No α computation — this version analyzes structure only Conservative statement The CMB rotation field is not well described by a dipole-only pattern.The data require a dipole + quadrupole component and show harmonic periodicity in multipole space. This release is part of an iterative research series in which analysis methods, masks, and calibration procedures were progressively refined. Later releases establish the physical calibration of the α(n) field. PUBLICATION RECORD PREDECESSOR PUBLICATION (Separate Record) Sep 20, 2025 (v1.0) — Harmonic Phase Alignments in Planck 2018 CMB — DOI:10.5281/zenodo.17167268 MAIN RESEARCH SERIES Concept DOI:10.5281/zenodo.17317397 Oct 10, 2025 (v1.0) — Scale-Dependent Anisotropic Birefringence: Initial Detection — DOI:10.5281/zenodo.17317398 Oct 20, 2025 (v1.1) — Scale-Dependent Anisotropic Birefringence: Validation Dataset — DOI:10.5281/zenodo.17396428 Oct 21, 2025 (v1.2) — Two-Harmonic Extension — DOI:10.5281/zenodo.17410764 Oct 28, 2025 (v1.3) — Two-Harmonic Dipole Verification — DOI:10.5281/zenodo.17468988 Nov 1, 2025 (v1.4) — MASTER-Calibrated Dipole — DOI:10.5281/zenodo.17500791 Nov 1, 2025 (v1.41) — Extended MASTER Calibration and Robustness — DOI:10.5281/zenodo.17508908 Nov 7, 2025 (v1.42) — Dependence-Aware Joint Validation — DOI:10.5281/zenodo.17553829 Nov 8, 2025 (v1.43) — Phase Model Validation — DOI:10.5281/zenodo.17561313 Nov 8, 2025 (v1.44) — Axis + Frequency + Half-Mission Validation — DOI:10.5281/zenodo.17561768 Nov 9, 2025 (v1.5) — Multipole Structure and Model Selection — DOI:10.5281/zenodo.17562965 Nov 9, 2025 (v1.6) — Phenomenology and Physical Interpretation — DOI:10.5281/zenodo.17566197 Nov 9, 2025 (v1.7) — Prediction and Experiment Overlays — DOI:10.5281/zenodo.17566870 Nov 9, 2025 (v1.8) — Model Rejection and Δℓ Persistence — DOI:10.5281/zenodo.17567241 Nov 10, 2025 (v2.0) — Intrinsic Periodicity in ℓ-Space — DOI:10.5281/zenodo.17574048 Nov 10, 2025 (v2.1) — Physical Origin of Δℓ Modulation — DOI:10.5281/zenodo.17577086 Nov 11, 2025 (v2.2) — Universe-Model Evaluation — DOI:10.5281/zenodo.17585419 Nov 12, 2025 (v2.3) — Domain Geometry and Topological Inference — DOI:10.5281/zenodo.17594157 Nov 13, 2025 (v2.4) — Real-Space Correlation of the Birefringence Field — DOI:10.5281/zenodo.17597537 Nov 13, 2025 (v2.5) — Spectral Surgery on the Δℓ ≈ 109 Harmonic — DOI:10.5281/zenodo.17604982 Nov 14, 2025 (v2.6) — Angular Locality of the Δℓ = 109 Standing Wave — DOI:10.5281/zenodo.17613348 Nov 15, 2025 (v2.7) — Sky-Local Origin of the Δℓ ≈ 109 Standing Wave — DOI:10.5281/zenodo.17620029 Nov 15, 2025 (v2.8) — Domain Topology of the Δℓ ≈ 109 Standing Wave — DOI:10.5281/zenodo.17620605 Nov 16, 2025 (v2.9) — Dual-Domain Coherence and Boundary Geometry — DOI:10.5281/zenodo.17621871 Nov 17, 2025 (v2.10) — Boundary Sequence Structure on the Dual-Domain Loop — DOI:10.5281/zenodo.17635811 Nov 19, 2025 (v2.11) — Boundary Standing-Wave and Phase-Structure Analysis — DOI:10.5281/zenodo.17648033 Nov 21, 2025 (v2.12) — Boundary Universality and Standing-Wave Fingerprints — DOI:10.5281/zenodo.17676377 Nov 23, 2025 (v2.13) — Interior Propagation and Boundary-Driven Structure — DOI:10.5281/zenodo.17693540 Jun 18, 2026 (v2.29) — Rotation Field of the Cosmic Microwave Background — Interior Propagation Audit & Harmonic Normalization — DOI:10.5281/zenodo.20753037 Jun 19, 2026 (v2.30) — Calibrated Interior Propagation Validation — DOI: 10.5281/zenodo.20755330 Jun 20, 2026 (v2.31) — Rotation Field of the Cosmic Microwave Background — Physical Origin of Boundary-to-Interior Propagation — DOI: 10.5281/zenodo.20777435 Jun 21, 2026 (v2.32) — Rotation Field of the Cosmic Microwave Background — Antipodal Specificity, Boundary Ordering, and Cross-Scale Spectral Organization — DOI: 10.5281/zenodo.20787307 22 Blue - The Heartbeat of the Universe

### 数据集摘要 版本v1.5针对v1.43与v1.44中使用的同一锁定α(ñ)旋转场开展模型选择与多极结构分析。本次版本未执行α重建,亦未更改校准或预处理流程。v1.5的核心目标为确定何种空间结构可最优解释该旋转场。 所有模型均采用: • 同一锁定α(ñ)映射表(alpha_lowL_combined_v141.fits) • 同一切趾偏振掩模(NSIDE=512,f_sky≈0.467) • 同一低ℓ频段(2–10) v1.43中的各向异性检测结果(在165,000次零模拟中经验显著性超出4.38σ)保持不变。 #### 测试模型 模型A — 仅偶极子:α(ñ) = a₀ + d · ñ 模型B — 偶极子+无迹四极子:α(ñ) = a₀ + d · ñ + Q : (ññ) 模型C — 多极空间谐波模型:α(ℓ) = A sin(kℓ + φ) + C + G/ℓ #### 分析结果 • 仅偶极子模型被显著否决。模型选择结果强烈倾向于偶极子+四极子结构: ΔAIC ≈ +2.16×10⁵,ΔBIC ≈ +2.16×10⁵。 • 旋转频谱α(ℓ)存在谐波周期性。 #### 关于Δℓ的说明 此前无约束谐波拟合(v1.2)偏好Δℓ≈360。 在v1.5中,为提升模型选择稳定性(基于赤池信息准则(AIC)/贝叶斯信息准则(BIC)),我们对谐波频率扫描范围施加了限制。 自助法(bootstrap)分布饱和了该允许范围的上限,得到Δℓ≈454。 两种拟合结果均与α(ℓ)中的周期性结构相符,二者差异源于建模约束而非数据本身的变化。 最优拟合周期:Δℓ = 454.7 68%置信区间(自助法):433.07 – 454.72 – 454.72 (中位数与上限重合,因自助法分布被允许的k扫描范围截断) • Δℓ在以下条件下保持稳定: – 不同的分箱宽度(Δℓ_bin = 12、15、20) – 掩模切趾参数变化(+0.5°、+1.5°) – 后验预测检验(最优拟合模型可复现振幅包络) • 交叉验证预测误差随模型A→模型B→模型C依次降低。 v1.5未修改v1.43/v1.44中确立的检测结果,仅对α(ñ)的结构进行了表征。 #### 存档内容(所有文件均带有时间戳,无文件会被覆盖) • dipole_fit_modelA.json/png • quadrupole_fit_modelB.json • harmonic_modelC_fit_*.json/png • harmonic_modelC_series_*.csv • v1_5_metrics_table_*.json/txt(包含AIC/BIC/logL/CV-RMSE/残差比) • v1_5_robustness_summary_*.json(包含自助法、分箱、掩模切趾检验结果) • v1_5_numbers_table_*.txt(可读式人工总结) • v1_5_README_paragraph_*.txt(用于论文/期刊稿件的文本) • v1_5_bundle_manifest_*.json(所有文件的SHA-256哈希值) #### 可复现性 • 采用与v1.43/v1.44一致的α(ñ) FITS映射表与掩模 • 未执行α值计算——本版本仅开展结构分析 • 仅需单个Jupyter Notebook:run_v15_model_selection.ipynb即可复现全部结果 #### 保守性声明 宇宙微波背景(Cosmic Microwave Background, CMB)旋转场无法仅用偶极子模式进行良好描述。实验数据要求引入偶极子+四极子分量,并证明多极空间中存在谐波周期性。 #### 版本历史(22 Blue数据集系列) 单独发表 —— 2025年9月20日 • 《普朗克2018年CMB偏振数据中的谐波相位对齐证据》 识别出相干多谐波结构(Δℓ≈320)。 v1.0 —— 2025年10月10日 • 首次定量检测到尺度依赖的各向异性宇宙双折射 (8.5σ各向同性显著性 + 4–5σ方向性显著性)。 v1.1 —— 2025年10月20日 • 新增完整统计验证与可复现性材料 (TB/EB诊断、协方差检验)。 v1.2 —— 2025年10月21日 • 引入双谐波扩展模型(Δℓ≈180与360)。 v1.3 —— 2025年10月28日 • 完成验证与稳健性测试 (校准、模拟、注入-恢复测试)。 哈佛大学Dataverse镜像 —— 2025年10月29日 • 数据集镜像至哈佛大学Dataverse v2 DOI: 10.7910/DVN/PTDG20 v1.4 —— 2025年11月1日 • MASTER校准旋转场α(ℓ);验证了低ℓ频段传递函数。 v1.41 —— 2025年11月1日 • 扩展校准与稳健性评估 (SMICA/NILC、半任务数据拆分)。 v1.42 —— 2025年11月7日 • 依赖于观测的经验联合显著性检验(宇称+高ℓ谐波) 165,000次零实现;单次添加检验p≈6.06×10⁻⁶(≈4.38σ)。 v1.43 —— 2025年11月8日 • 相位模型验证——校准不变的宇称结果。 MASTER耦合将宇称再现精度控制在≤0.26%;零模拟显著性超出次数为0/165,000。 v1.44 —— 2025年11月8日 • 采用同一锁定α(ñ)场开展轴、频率与半任务验证。 各向异性旋转场在分量分离、频率选择与探测器拆分条件下均保持稳定。 v1.5 —— 2025年11月9日(本版本发布) • 采用同一锁定α(ñ)场开展多极结构分析与模型选择。 • 仅偶极子模型被显著否决(ΔAIC≈+2.16×10⁵,ΔBIC≈+2.16×10⁵)。 • α(ℓ)呈现谐波周期性:Δℓ=454.7(68%置信区间:433.07 – 454.72 – 454.72)。 • Δℓ在分箱、掩模变化与后验预测检验条件下保持稳定。 本数据集属于“22 Blue —— 宇宙的心跳”研究档案的一部分。

提供机构:
Zenodo
创建时间:
2025-11-09
二维码
社区交流群
二维码
科研交流群
商业服务