MIPOL 33 GHz circular-polarization measurements

This dataset mirrors all six scientific tables served on LAMBDA's Milano Polarimeter (MIPOL) data-product page. MIPOL measured the Stokes V parameter at 33 GHz. The tables themselves are the data source; the page links no separate ASCII, FITS, or other measurement file. The five captioned configurations are identified by their exact published captions. The remaining table has no caption or header and is identified by its two published rows and those two absences. Page position is not an identifier.

Tables 1 and 4–6 retain their published identifiers as fields. Tables 2 and 3 retain their native shapes. Their unnamed positions use archive transport identifiers because the former has a blank first source identifier and the latter has no source header.

How to use

Create an isolated fleet environment:

python -m venv .venv && .venv/bin/pip install datasets huggingface_hub pyarrow

from datasets import load_dataset
dataset = load_dataset("astro-legacy-archive/mipol", "stokes_v_summary", split="train[:2]", columns=["Pixel size", "<V>", "Unit"])
for row in dataset:
    print(row["Pixel size"], row["<V>"], row["Unit"])

2° x 2° unbinned data -(0.02±0.06) mK
8° x 8° -(0.02±0.08) mK

import json
from huggingface_hub import hf_hub_download
import pyarrow.parquet as pq
path = hf_hub_download("astro-legacy-archive/mipol", "cmb_circular_polarization_upper_limits.parquet", repo_type="dataset")
print(json.loads(pq.read_metadata(path).metadata[b"statistical_product"])["self_description"]["published_rows"][1])

['Pixel size', 'Π V CMB', 'Π V CMB', 'Π V CMB']

Dataset structure

Source-value columns have Arrow type string. Records and fields remain in source order and the native table shapes are retained. HTML cell text is projected by space-joining stripped text nodes after character references are decoded. The embedded statistical_product document records the complete published row matrix, corresponding verbatim cell-inner HTML slices, source table selector and observed index, naming state, field roles, meanings, unit evidence, source pin, rights, credit, and regeneration command.

radiometer_properties

Table 1 has eleven records and five columns: Type, Quantity, Circular Polariz., Linear Polariz., and Unit. The blank separator row in the source table is retained as a record.

table_2 and table_3

Table 2 has four rows and five fields. Its blank first source identifier is preserved in the complete source header and that unnamed field uses the archive transport identifier Column 1; the other four fields retain their published identifiers. Table 3 has two rows and two fields, Column 1 and Column 2, both declared as archive transport identifiers because its source header is absent. Its absent header and caption are recorded explicitly in self-description.

stokes_v_summary

Table 4 has four rows and seven columns.

column	unit	meaning
Pixel size	degree	Angular pixel size of the published map binning
Number of pixels	—	Number of pixels in the published map binning
<V>	mK	Mean of V, the circular-polarization Stokes parameter
σ V	mK	Standard deviation of V over the full map
σ V pix	mK	Average rms of V per pixel
σ &ltV> pix	mK	Average standard deviation of the mean V per pixel
Unit	—	Unit written for the four V summary quantities in the source row

The source supplies mK in every Unit cell and degree signs in every Pixel size cell. The first row is the 2° x 2° unbinned data result, followed by 8° x 8°, 12° x 12°, and 24° x 24°.

gaussianity_test

Table 5 has three rows and five columns.

column	unit	meaning
Pixel size	degree	Angular pixel size of the tested map binning
f	—	Degrees of freedom in the published Gaussianity test
μ=&ltV>     σ=σ V χ 2 P	—	Published chi-squared and Gaussian-model probability when μ= and σ=σ V
Best fit χ 2 P	—	Published chi-squared and Gaussian-model probability for the best-fit Gaussian model
Simulations χ 2 P	—	Published chi-squared and Gaussian-model probability for the simulations

The page defines f as degrees of freedom and P as the probability of the assumed Gaussian model. Each of the final three fields holds the source's composite chi-squared/probability cell. The final row's two -- cells remain literal strings.

cmb_circular_polarization_upper_limits

Table 6 has three stored rows and four columns.

column	unit	meaning
Method	degree	Top-level source heading whose second-level heading is Pixel size
Classical	—	95% confidence-level upper limit on the dimensionless CMB circular-polarization ratio Π_V^CMB = V/T_CMB from the classical analysis
Bayesian	—	95% confidence-level upper limit on the dimensionless CMB circular-polarization ratio Π_V^CMB = V/T_CMB from the Bayesian analysis
Spherical harmonics	—	95% confidence-level upper limit on that ratio from the spherical-harmonics analysis, with its angular-scale annotation

The source table has two header rows. Its first supplies the four field identifiers. Its second header level is retained in metadata: Pixel size, followed by Π V CMB repeated beneath each of the three distinct method headings. It is not a stored data record. The pinned paper defines Π_V = V/T_CMB and identifies the values as 95% confidence-level upper limits.

Decoding the source text

Every cell has a derived text projection, including ±, degree signs, Greek letters, literal 10 5 and 10 -9 exponent typography, --, empty cells, and the source string &ltV>. Composite strings such as 6.04    0.70, 5.0 10 -4, and 4.3 10 -4 <θ>=7.5° have not been split or numerically interpreted. Resolvers should use source field identifiers where present, archive transport identifiers where they are absent, the complete published_rows and published_cell_html matrices, and published_meanings in the embedded metadata.

Table 1's wholly blank separator row, table 2's blank first identifier, table 3's absent header and caption, and table 6's repeated second header are source properties, not replacement names or inferred fields.

Provenance and verification

The MIPOL product page was checked on 2026-08-31: 24,458 bytes, SHA-256 1a5cf72b15cd92ba865a3544dbfeb097c580962a4189c0e811165c9c6be8c212. The MIPOL experiment page, the source for the 33 GHz context, was checked on the same date: 16,775 bytes, SHA-256 0817fdd9f1418fd1aeeded8d71660e081dfb22bd3110395392dff7070ef4611f.

The defining Mainini et al. paper was also checked on 2026-08-31: 470,572 bytes, SHA-256 e05757cf87d7766f28249a03c215b3dba02b34838761e1b7edae182815746c7b. From the repository root, the complete build command is:

mkdir -p data/mipol
python tools/make_mipol.py --out data/mipol --source data/mipol --provenance-commit "$(git rev-parse HEAD)"

Verification covers all six source tables in the pinned HTML: four wholly named-field tables and two native-shape tables with transport identifiers for unnamed fields.

Considerations for use

These are summary statistics, Gaussianity-test results, and upper limits, not a pixel-level sky map or time-ordered data. Missing scalar units are recorded as documented source silence rather than inferred units. Consult the defining paper for the observation selection, simulations, analysis methods, and confidence interpretation before constructing a likelihood.

The product page states Data collected at night time (1 hour after sunset - 1 hour before sunrise). Mainini et al. section 4 states Only data collected at nighttime (i.e. between half an hour after sunset and half an hour before sunrise) have been analyzed. The disagreement is retained without selecting or normalizing either statement.

Licensing and credit

The pinned HEASARC data policy was checked on 2026-08-31: 23,756 bytes, SHA-256 0bea39b0bdccafdebf3cb3e32c3b5b81f310d606b14185b3ba377ab93cf2970a. The measurements and cited paper are products of the MIPOL Science Team and authors; LAMBDA/HEASARC hosts the tables. No separately identified data licence issued by the MIPOL authors was found. HEASARC's hosting policy states verbatim: “HEASARC materials are all available freely for your use.” If using HEASARC made a significant contribution, that policy requests:

This research has made use of data and/or software provided by the High Energy Astrophysics Science Archive Research Center (HEASARC), which is a service of the Astrophysics Science Division at NASA/GSFC.

The pinned LAMBDA contact page was checked on 2026-08-31: 13,730 bytes, SHA-256 1b3c2496e3036f4be3b39bcc23b4e0d7e97d8198ec27dddea5e1f6047c95d1d0. Its requested credit is:

We acknowledge the use of the Legacy Archive for Microwave Background Data Analysis (LAMBDA), part of the High Energy Astrophysics Science Archive Center (HEASARC). HEASARC/LAMBDA is a service of the Astrophysics Science Division at the NASA Goddard Space Flight Center.

Citation

Cite Mainini et al. (2013), Journal of Cosmology and Astroparticle Physics, 08, 033, arXiv:1307.6090, and acknowledge LAMBDA using the text above.

The pinned MIPOL publications page was checked on 2026-08-31: 13,280 bytes, SHA-256 c7f0c26ef6221d05b7e34dde6a35af6666f59f07e45c45e6480ee76a908f77e6. It lists only Mainini et al. (2013). The experiment page additionally prints five earlier references:

• Sironi G. et al. (1998): New Astronomy, 3, 1-13
• Sironi G. et al. (1998) ASP Conf. Ser. 141, 116
• Gervasi M. et al. (2000) SIF Conf. Proc. 68,165
• Spiga D. et al. (2002): New Astronomy, 7, 125-134
• Gervasi M. et al. (2002): SPIE, 4843, 336-347

Please also cite this dataset:

@misc{https://doi.org/10.57967/hf/10433,
  doi = {10.57967/hf/10433},
  url = {https://huggingface.co/datasets/astro-legacy-archive/mipol},
  author = {{Astro Legacy Archive}},
  title = {mipol},
  publisher = {Hugging Face},
  year = {2026}
}

The initial public release is Hub revision bfae66a0108af5ae05dec4089b74d1dcb0f2b84a.

Errata and requests

To report an erratum, propose a correction, or ask a question about this dataset, open a Discussion or Pull Request in its Community tab; posting requires a free Hugging Face account. Dataset requests and archive-wide questions may be posted in the Community tab of any dataset. Confirmed corrections are made in the source conversion tools and published by rebuilding the dataset.