← Home

Pythagoras Star Map v2 (≤50 ly) and v1.2 (≤20 ly): sources, methods and limitations

(Earlier text kept unchanged in SOURCES_v1.md and SOURCES_v1.1.md; all earlier outputs starmap_v1*.*, planets_v1*.csv are kept. SOURCES_v1.2.md is a copy of this file.)

Build date: 2026-10-17 (UTC). Two builds from the same code: - v1.2: primaries within 20 ly (6.13 pc): 98 systems, 61 confirmed planets (14 in the conservative HZ, 19 conservative or optimistic). Gas-cloud (LISM) lookup done for all 98 sight lines. - v2: primaries within 50 ly (15.33 pc) (STARMAP_DMAX_LY=50 STARMAP_VERSION=v2): 1267 systems, 229 confirmed planets in 123 host systems (30 in the conservative HZ, 46 conservative or optimistic). LISM lookup done for 193 sight lines (every planet host and every system in the top 181); the other 1074 systems (all without known planets) are unknown and flagged.

No value in the outputs was typed in by hand except the Redfield & Linsky calculator results (source 6), which were copied verbatim from the calculator's responses.

1. Data sources (credited by institution and country)

Planet-discovery instruments. These are credited in each system's sources column and in planets_v1.csv (facility_credit, reference): - ESO La Silla HARPS (and NIRPS) and ESO Paranal VLT/ESPRESSO, Chile (ESO is intergovernmental, headquartered in Germany): Proxima b and d, Barnard b (ESPRESSO), GJ 1061, Ross 128, Lacaille 9352 (GJ 887), Luyten's star, YZ Ceti, GJ 674, Wolf 1061, GJ 3323, HD 20794 and others. - CARMENES at Calar Alto, Spain (CAHA, operated by CSIC Spain and MPIA Germany): Teegarden's Star b and c, HD 180617 b. - HARPS-N on the TNG, La Palma, Spain (INAF, Italy): GJ 15 A c. - SOPHIE at the Observatoire de Haute-Provence, France (CNRS): GJ 411 b. - MAROON-X on Gemini North (NSF NOIRLab, international partnership): Barnard c, d and e. - Keck/HIRES and Lick, USA; the Anglo-Australian Telescope, Australia. - "Multiple Observatories" means the archive credits several, often HARPS together with Keck, CARMENES and others.

Requested but not used in v1: - Subaru/IRD (NAOJ, Japan): no confirmed planet within 20 ly in pscomppars lists IRD as its discovery facility, so there was nothing to credit. IRD follow-up RV data are not in the composite table. - LAMOST (NAOC, China): not queried. Every star in this volume already has SIMBAD spectral types and Gaia/Hipparcos astrometry, and LAMOST avoids stars this bright. Worth adding for the 50 ly build, for spectral types and activity indicators of faint M dwarfs. - CARMENES and ESO archives: used only through the discoveries credited above. Their raw RV time series were not downloaded.

2. Pipeline

pipeline/activity.py (new in v1.1) fetches catalogues 8a–8d (raw/activity/*.tsv + *_meta.json) and matches them by Gliese/GJ, HD, HIP or SIMBAD name against each star's SIMBAD identifiers. run_all.py runs it unless --no-fetch. (re-runnable; extend to 50 ly by changing one parameter)

pipeline/config.py DMAX_LY (default 20), ranking WEIGHTS, ship profile pipeline/fetch.py TAP queries above -> raw/*.csv + raw/*_meta.json (query, endpoint, UTC time, row count) pipeline/merge.py SIMBAD <-> Gaia DR3 (by Gaia DR3 id in SIMBAD identifiers); systems by union-find pipeline/score.py planets -> hosts, Kopparapu HZ, scores, trip times -> starmap_{VERSION}.csv, planets_{VERSION}.csv, build_meta_{VERSION}.json (VERSION = "v1.1" in config.py) pipeline/render_html.py self-contained three.js map -> starmap_{VERSION}.html pipeline/run_all.py runs everything + headless-Chrome PNG snapshot Builds: STARMAP_VERSION=v1.2 .venv/bin/python pipeline/run_all.py --no-fetch (20 ly) and STARMAP_DMAX_LY=50 STARMAP_VERSION=v2 .venv/bin/python pipeline/run_all.py --no-fetch (50 ly). Parallax cuts use a 3% margin; the NASA archive cut uses a 1.5× margin (NASA_MARGIN), because planets are placed by the host's Gaia/SIMBAD astrometry, not the archive's sy_dist. Raw TAP caches for 20 ly are in raw/, for 50 ly in raw/scope_50ly/; the activity catalogues (raw/activity/), RECONS and LISM files are shared. Sight lines not in raw/lism_redfield_linsky2008.tsv read "unknown".

3. Cross-match and system merge

• Astrometry priority. Gaia DR3 when the SIMBAD object carries a Gaia DR3 id. Otherwise the parallax SIMBAD adopts, with its bibcode recorded: Hipparcos (ESA) for bright stars, plus the literature values listed in source 4b.
• Gaia match. All 100 Gaia DR3 sources with parallax > 158.33 mas matched a SIMBAD object, so there are no Gaia-only sources (gaia_unmatched_sources.csv is empty).
• Planet hosts. Matched by gaia_dr3_id, falling back to the nearest SIMBAD object within 10′ (needed for GJ 411 and GJ 273).
• System grouping merges objects that are:
• linked in the SIMBAD hierarchy (membership ≥ 50% or unspecified), or
• within 0.1 pc projected with distances consistent to 3σ (minimum 0.05 pc), or
• within 0.02 pc projected with Δd < 0.2 pc. This catches tight pairs whose parallaxes disagree because of orbital motion or poor astrometry, e.g. Wolf 424 AB, eps Ind B, SCR 1845 B.
• Component counts. RECONS counts override where they match. Otherwise, SIMBAD system-level entries such as "alf Cen" or "70 Oph" are not counted as bodies, and a lone unresolved **/SB* entry counts as two.
• System distance is the primary's (brightest member) distance. Trip times use the destination component: the host of the best-scoring planet, else the primary. For alpha Centauri that is Proxima at 4.246 ly; the trip to A/B is given separately (trip_*_to_primary).
• Galactic XYZ (light-years, heliocentric) come from the astropy ICRS→Galactic transform: X toward the Galactic Centre, Y toward l = 90°, Z toward the North Galactic Pole.
• Distance error = d × σ_plx / plx.

4. Ranking (first pass; weights are user-set and live in config.py)

score_equal_weight = w_d·score_distance + w_h·score_habitability + w_r·score_route, with w_d = w_h = w_r = 1/3 (user decision 2026-10-17: balanced mix).

No-planet gate (new in v1.1, user decision 2026-10-17): score_total = score_equal_weight for systems with ≥1 confirmed planet; score_total = NO_PLANET_FACTOR × score_equal_weight with NO_PLANET_FACTOR = 0.5 (in pipeline/config.py) for systems with no confirmed planet. Reason: a nearby system with a clean route but no known planet (e.g. Luhman 16, v1 rank 8, score 0.53) should not outrank systems with known HZ planets. With 0.5, the best no-planet system (Luhman 16, 0.26) ranks 29th, below every planet host. CSV columns: score_equal_weight, no_planet_gate_applied, score_total. Set the factor to 1.0 to get the v1 ranking back.

Distance: 1 − d_dest / DMAX_LY, linear (20-ly build: Proxima 0.79, 20 ly → 0; 50-ly build: Proxima 0.92, 50 ly → 0). Because the scale depends on DMAX, v1.x and v2 scores are not comparable; in v2 distance differences matter less, so habitability and route weigh relatively more.

Habitability: the best planet's score, HZ × size × confirmation × host type × activity. Systems with no known planet score 0. That means no data, not proof there are no planets. - HZ, using the planet's insolation S (archive pl_insol, else L/a²), compared with the Kopparapu 2013 limits at the host Teff: - conservative (moist greenhouse to maximum greenhouse) = 1.0 - optimistic (recent Venus to early Mars) = 0.75 - "near" (within 25% in flux outside the optimistic limits) = 0.35 - outside = 0 - Size (from mass or M sin i; radii in pscomppars for RV planets are mass-radius estimates, not measurements): ≤2 M⊕ 1.0; 2–5 M⊕ 0.8; 5–10 M⊕ 0.4; >10 M⊕ 0.05. - Confirmation: NASA pl_controv_flag = 1 gives × 0.5. - Host type (heuristic for tidal locking and pre-main-sequence water loss): G/K 1.0; F 0.9; M0–M2.5 0.85; M3–M4.5 0.75; M5+ 0.65; A/white dwarf/brown dwarf 0.4. - Activity (flare hazard), weights: low 1.0; unknown 0.85; moderate 0.8; high 0.6. Levels come from fetched catalogues first (rows 8a–8d, code in pipeline/activity.py): - TESS flare rate R (flares/day, Medina+2020): high if R ≥ 0.1, moderate if 0.01 ≤ R < 0.1, low if R < 0.01. - log(Lx/Lbol) (Stelzer+2013, Wright+2011, NEXXUS + archive L*): high if ≥ −3.5 (near saturation); moderate −5.0 to −3.5; low < −5.0 (Sun ≈ −6 to −7). An upper limit counts only if it is below −5.0 (then low). - If several indicators exist, the most severe one sets the level; all values and citations are listed in host_activity_note and host_activity_source. - Fallback only when no catalogue value exists: SIMBAD object types, labelled high (coarse) (Er*/Fl*) or moderate (coarse) (BY*/RS*/Ro*). Otherwise unknown. "Unknown" means no data, not a quiet star. - These thresholds are heuristic bins chosen for this map, not values from the papers.

Route: from the Redfield & Linsky 2008 sight-line result. Start at 1.0, then: - −0.30 if no warm cloud lies on the sight line. The route would then be mainly hot Local Bubble gas, so the magsail would likely see far less plasma than the n_i = 0.1 cm⁻³ the design assumes. - −0.10 for each cloud traversed beyond the first. Each boundary changes density and flow vector, which affects magsail drag and cross-track forces (design doc §3.2). - −0.03 for each cloud passing within 20° (edge uncertainty), capped at −0.15.

Sight lines not queried get score_route = blank / unknown; only for them the composite uses the median of the queried sight lines (0.88) and sets route_score_imputed = True. v1.2: 0 of 98 imputed (v1.1: 54). v2: 1074 of 1267 imputed, all without known planets and all ranked ≥182, so no planet host or top-181 system uses the fallback.

Distance-driven dust and proton fluence is deliberately left out of the route score so distance isn't counted twice.

5. Trip times (Pythagoras Concept Design v2.4 §4.7, 20 kt ship, propellant carried)

• Profile: laser boost at 0.1 g (2.924 yr Earth / 2.894 yr crew, 0.367 ly) → coast at 0.23c (γ = 1.02755) → trailing-magsail brake to 0.02c (22.517 yr, 2.452 ly) → flip → D-³He fusion brake at 0.01 g (1.938 yr, 0.019 ly).
• Fixed part: the non-cruise phases take 27.378 yr Earth / 27.165 yr crew over 2.839 ly.
• Formula: t_Earth = 27.378 + (D − 2.839)/0.23 and t_crew = 27.165 + (D − 2.839)/(0.23 γ). The crew time of the magsail phase (22.333 yr) is derived from the doc's crew total.
• Check: at the doc's D = 4.24 ly this gives 33.47 / 33.09 yr, matching the doc's 33.5 / 33.1 yr.
• Alpha Centauri mismatch: the doc's 33.5 yr uses D = 4.24 ly, i.e. Proxima Centauri. Gaia DR3 puts Proxima at 4.2465 ly (33.50 yr), while alpha Cen A/B are at 4.344 ly (Akeson et al. 2021 parallax) → 33.92 yr Earth / 33.53 yr crew.
• Coast-only times (D/0.23c, no acceleration phases) are also in the CSV for comparison.
• Assumptions: the magsail phase uses the same ISM density everywhere (the doc's n_i = 0.1 cm⁻³). The +4.9 kt neutron face (doc: 35.9 yr) is excluded.

6. Known limitations and uncertainties

• Completeness: the volume is limited to objects with a SIMBAD parallax > 158 mas. Undiscovered cold brown dwarfs and any object without a SIMBAD parallax are missing.
• Questionable members: four brown dwarfs/binaries have parallax errors > 5% (HD 101847BC: ±20%; 2MASS J0254+0223: ±23%), so whether they belong inside 20 ly is uncertain.
• Planets are almost all radial-velocity detections: masses are minimum masses, radii are estimated, and nothing is known about atmospheres or habitability.
• Disputed or retracted planets:
• Barnard's star: the 2018 super-Earth (Ribas et al.) was later refuted. The current Barnard b–e (2024–2025, ESPRESSO and MAROON-X) are all too hot for the HZ.
• Proxima d: first reported with ESPRESSO (Faria et al. 2022) and listed as confirmed; the archive's current reference is Suárez Mascareño et al. 2025 (La Silla). It is a very low-amplitude signal.
• Proxima c: a SIMBAD candidate only, not in the NASA confirmed table.
• alpha Cen Ab: a JWST candidate, unconfirmed.
• Flagged controversial by the NASA archive: tau Ceti f/g/h, GJ 15 A b, Kapteyn c, the GJ 229 planets, HD 20794 e.
• SIMBAD Pl? objects not in the NASA table are listed in candidate_planets_unconfirmed and never scored.
• LISM: the Redfield & Linsky calculator gives projected cloud morphology only, not distances to cloud boundaries. A traversed cloud may lie beyond a target, and the model is built from absorption toward stars within ~15 pc. The Sun is at the LIC/G-cloud transition.
• Dust and radiation: no direction-dependent interstellar dust data at the scale of these routes was found. Dust hazard is unknown beyond the design doc's LIC-average estimates (§2.7: 25.8 MeV ISM protons at 0.23c; dust impacts dominate erosion).
• Activity: catalogue values exist for 27 of 28 planet hosts (Teegarden's Star has none in these catalogues → unknown). The 14 remaining (coarse) flags are all on systems with no known planet, so they do not change any score. Bins are heuristic; X-ray data are mostly ROSAT-era (1990s) and activity varies with cycles; Medina flare rates count flares above a fixed energy, so they are not scaled to each star's luminosity (Proxima's rate 0.089/day falls just under the "high" bin, and its NEXXUS log(Lx/Lbol) −3.51 sits right at the boundary). ROSAT survey sources can blend close pairs (e.g. GJ 15 A/B share one survey count rate).
• RECONS: the 2012 list was used only to check multiplicity. Its parallaxes were not used (Gaia DR3 supersedes them).
• v2 activity coverage: catalogue values exist for 67 of 123 planet hosts; 5 more have SIMBAD-only (coarse) flags and 51 are unknown (weight 0.85). The catalogues are volume-limited (NEXXUS 6 pc, Stelzer 10 pc, Medina 15 pc M dwarfs, Wright 2011 a selection), so beyond 20 ly "unknown" is common, e.g. GJ 667 C, Wolf 1069, GJ 3378. Over all 1267 systems: 840 unknown.
• v2 completeness: limited to SIMBAD parallaxes > 63.33 mas. The 15 pc census is less complete for cold brown dwarfs and faint companions than the 20 ly one; 44 systems have parallax errors > 5%.
• Archive distances: the NASA archive's sy_dist can be wrong (Gl 725 B: 6.84 pc in the archive vs ≈3.5 pc in Gaia DR3), which is why v1.2 widened the archive cut. A planet whose archive distance is more than 1.5× its true distance would still be missed.
• LISM beyond 15 pc: the Redfield & Linsky model was built from sight lines mostly within ~15 pc, so for 30–50 ly targets the cloud list is an extrapolation of projected morphology.
• Map readability (v2): only the top 10 are labelled by default; the HTML has a distance slider and planet-host / HZ filters plus label toggles. The PNG is a fixed snapshot at 50 ly with all systems shown, so it is crowded.
• Map overlap: labels near the Sun overlap in the PNG; the interactive map resolves this.

7. Changelog v1 → v1.1 (2026-10-17)

Six hand-written activity notes removed and replaced by catalogue values:

Other planet hosts also moved from SIMBAD-type or unknown flags to catalogue values. Examples: Luyten's star, GJ 251, eps Ind, e Eri, GJ 832 → low; GJ 682, GJ 674, GJ 687, GJ 725 A → moderate; GJ 411, GJ 229, GJ 752, GJ 83.1, GJ 3323 high (SIMBAD) → moderate; tau Ceti and Kapteyn's moderate → low; YZ Cet stays high (TESS 0.247/day). Overall counts: v1 high 31 / moderate 15 / low 4 / unknown 48. v1.1 high 7 / moderate 25 / low 20 / unknown 32, plus (coarse) SIMBAD-only flags: 10 high and 4 moderate.

No-planet gate added (section 4). Weights are unchanged.

8. Changelog v1.1 → v1.2 (2026-10-17, ≤20 ly)

• LISM: the 54 remaining sight lines within 20 ly were looked up with the Redfield & Linsky calculator (8 needed one retry, 1 needed two, all with l/b shifted 0.01 deg; no failures). The median route fallback is now used for no 20-ly system (route_score_imputed all False). Rows are in raw/lism_redfield_linsky2008.tsv after the # v1.2 line (pre-v1.2 copy: raw/lism_redfield_linsky2008_v1.tsv). All 54 are systems without known planets, so this moved only no-planet ranks (e.g. SCR 1845 41→38, Wolf 424 48→40).
• Missing planet fixed: Gl 725 B c (conservative HZ) was absent in v1/v1.1 because the archive lists its sy_dist as 6.84 pc (Gaia DR3 ≈ 3.5 pc), beyond the old 1.03× cut. The NASA cut now uses a 1.5× margin (NASA_MARGIN; old raw copy raw/nasa_pscomppars_v1-v1.1.csv). 60 → 61 planets; no other planet was missing.
• Top-8 change: HD 173739 / GJ 725 A enters at #2 (score 0.605, Gl 725 B c, moderate activity, LIC; 65.0 yr Earth / 63.8 yr crew). Teegarden's Star, Ross 128, Lacaille 9352, Luyten's star, GJ 1061 and GJ 682 each move down one place (#3–#8); GJ 251 drops from #8 to #9. alf Cen stays #1.
• sys_id column added; HTML gained the filter UI described above. Old raw/activity/J_*.tsv leftovers deleted.

9. v2 (2026-10-17, ≤50 ly)

• Same code, weights (1/3 each) and NO_PLANET_FACTOR = 0.5; same four activity catalogues; scope 50 ly.
• LISM prioritisation: sight lines were queried for all 123 planet hosts plus the top-40 provisional candidates (95 new queries beyond the 98 from v1.2). 21 of the 95 needed retries (16 once, 5 twice, each with l/b shifted 0.01 deg); none failed after retries. New rows follow the # v2 line in raw/lism_redfield_linsky2008.tsv (also raw/lism_v2_new.tsv).
• HTML: top 10 labelled; distance slider ("Show systems out to N ly"), "planet hosts only", "HZ planet hosts only", "label all planet hosts", "label all visible systems"; rings every 10 ly.

v3 findings log — sources and counts

v3 adds findings_v3.csv/json (55781 findings, long format), SYSTEM_PROFILES_v3.md, starmap_v3.{csv,html,png}. Older versions (v1, v1.1, v1.2, v2) are unchanged; the pre-v3 version of this file is kept as SOURCES_v2.md. Rule followed: every finding is generated by code from a file fetched in this build (raw/findings/…); nothing was typed in from memory. The only hand-written inputs are (a) the curated status list pipeline/findings/curated_status.json, whose every quote is machine-verified against a fetched abstract (rows whose quote is not found are rejected; 0 rejected), (b) the facility→operator/country map pipeline/findings/facilities.py (ASSUMPTION: operator/host country of the facility, not author affiliation), and (c) arXiv search terms (system names).

Fetched inputs

Method

• Targets (raw/findings/targets_v3.csv): tier A = v2 top 20 ∪ every planet host ≤20 ly (33 systems); B = other ≤20 ly systems; C/D = 20–50 ly planet hosts / others. Coverage: tier A: 33/33 systems, 6356 findings, tier B: 65/70 systems, 5960 findings, tier C: 89/90 systems, 11960 findings, tier D: 924/1074 systems, 31505 findings.
• Matching: catalogue name columns vs the full SIMBAD alias index, then HIP, then position (≤15″, method recorded in match_method); NASA hosts by alias; Gaia by DR3 source_id; SIMBAD by oid.
• Literature: SIMBAD bibliography (papers indexing ≤50 objects for tier A, ≤10 for B/C) with abstracts, plus arXiv. A paper is kept only if its title/abstract names the system (or it indexes ≤3 objects; then no status is inferred). Kept 5345 paper-system rows (dropped 14153 that did not name the target, 939 unclassifiable). Category and status are keyword heuristics (status_basis column) — refute/rule-out/false-positive → refuted, dispute/challenge/doubt → disputed, non-detection/upper-limit → non-detection, confirm → confirmed, else reported. Statuses only fire when the triggering sentence names the target. Treat literature statuses as leads, not verdicts; the curated rows are the reviewed verdicts.
• Status vocabulary: confirmed / disputed / refuted plus three labelled additions: candidate (archive 'Published Candidate' or authors' own wording), non-detection (null result / upper limit), reported (a measurement or paper with no confirm/dispute signal).

Findings by data service

Findings by status

Findings by category

Findings by facility / observatory (where attributable; a finding can count for several)

Findings by operating agency / institution

Findings by country / region of the operating facility (split on '/', a finding can count for several)

9717 findings carry a facility country; 46064 do not (SIMBAD measurement compilations and papers whose abstract names no facility). | item | findings | |---|---| | Europe | 5425 | | USA | 3462 | | Spain | 2143 | | Germany | 2090 | | Chile | 674 | | Canada | 476 | | France | 117 | | Italy | 83 | | Japan | 56 | | Netherlands | 45 | | UK | 39 | | Australia | 32 | | International | 32 | | China | 23 | | USA (Puerto Rico) | 5 | | Russia | 3 | | South Africa | 3 | | East Asia | 3 |

Ranking re-check

The unchanged v2 scorer was rerun on the cached v2 inputs (S0 reproduces v2 exactly: True) and with v3 statuses (S1): planets that fetched abstracts call candidates or dispute get the 0.5 weight v2 already gives to NASA-controversial planets: GJ 15 A b, GJ 251 c, GJ 682 b, GJ 682 c, Gl 725 A b. No refuted planet is in the scored table (the refuted ones — Barnard b 233 d, van de Kamp's planets, GJ 667 C d, VB 10 b — had already been removed from the NASA table). Atmosphere/JWST results are logged but are not part of the score. Changes in the top 20: see raw/findings/rank_check/rank_comparison_v2_v3.csv.

Coverage gaps (not fetched / not found)

• Press releases (ESO/NASA/ESA/JAXA/CNSA): none were fetched; JAXA/CNSA show up only when an abstract names their missions (AKARI, CSST).
• Astrospheres (Wood et al. HST Lyα): no machine-readable VizieR table was found under the IDs tried; astrosphere findings come only from abstracts (72 rows).
• Technosignature searches: Breakthrough Listen/FAST result tables were not found on VizieR; target-list membership (Isaacson+2017) is logged as 'reported' and is not a non-detection; actual search results only from abstracts.
• DEBRIS (Sibthorpe+2018) and Lestrade+2012 catalogues not found under the IDs tried; debris disk results come from DUNES, Montesinos+2016, Chen+2014 and abstracts.
• Author affiliations not fetched; institution/country = facility operator (assumption above).
• SIMBAD measurement tables (mesFe_h etc.) carry no uncertainties or instruments.
• NASA hosts not matched to a v3 target: 3 v2 hosts (CWISEP J193518.59-154620.3, HD 63433, TOI-540; their SIMBAD objects are not in the target list); others are beyond 50 ly.
• Tier D (20–50 ly, no planets): 150 systems have no finding from the fetched sources.

v5 additions

See SOURCES_v5.md: Gaia DR3 Apsis/NSS/variability/rotation, TIC v8.2, 2MASS, AllWISE, GALEX, eRASS1, WDS, ExoFOP TOIs, Hunt & Reffert 2023 clusters, DSS2 cutouts, ESO/NOIRLab images.