J/A+A/705/A232 miniJPAS survey quasar selection. V. (Perez-Rafols+, 2026)
The miniJPAS survey quasar selection. V. Combined algorithm.
Perez-Rafols I., Abramo L.R., Martinez-Solaeche G., Rodrigues V.N.N.,
Pieri M.M., Burjales-del-Amo M., Escola-Gallinat M., Ferre-Abad M.,
Isern-Vizoso M., Alcaniz J., Benitez N., Bonoli S., Carneiro S., Cenarro J.,
Cristobal-Hornillos D., Dupke R., Ederoclite A., Gonzalez Delgado R.M.,
Gurung-Lopez S., Hernan-Caballero A., Hernandez-Monteagudo C.,
Lopez-Sanjuan C., Marin-Franch A., Marra V., Mendes de Oliveira C.,
Moles M., Sodre L.Jr, Taylor K., Varela J., Vazquez Ramio H.
<Astron. Astrophys. 705, A232 (2026)>
=2026A&A...705A.232P 2026A&A...705A.232P (SIMBAD/NED BibCode)
ADC_Keywords: Surveys ; QSOs ; Photometry ; Redshifts ; Optical
Keywords: methods: data analysis - techniques: photometric - quasars: general -
cosmology: observations
Abstract:
Quasar catalogues from narrow-band photometric data are used in a
variety of applications, including targeting for spectroscopic
follow-up, measurements of supermassive black hole masses, or Baryon
Acoustic Oscillations. Here, we present the final quasar catalogue,
including redshift estimates, from the miniJPAS Data Release
constructed using several flavours of machine-learning algorithms.
In this work, we use a machine learning algorithm to classify quasars,
optimally combining the output of 8 individual algorithms. We assess
the relative importance of the different classifiers. We include
results from 3 different redshift estimators to also provide improved
photometric redshifts. We compare our final catalogue against both
simulated data and real spectroscopic data. Our main comparison metric
is the f1 score, which balances the catalogue purity and completeness.
We evaluate the performance of the combined algorithm using synthetic
data. In this scenario, the combined algorithm outperforms the rest of
the codes, reaching f1=0.88 and f1=0.79 for high- and low-z quasars
(with z≥2.1 and z<2.1, respectively) down to magnitude r=23.6. We
further evaluate its performance against real spectroscopic data,
finding different performances (some of the codes show better
performance, some worse, and the combined algorithm does not
outperform the rest). We conclude that our simulated data is not
realistic enough and that a new version of the mocks would improve the
performance. Our redshift estimates on mocks suggest a typical
uncertainty of σNMAD=0.11, which, according to our results
with real data, could be significantly smaller (as low as
σNMAD=0.02). We note that the data sample is still not large
enough for a full statistical consideration.
Description:
Catalogues of quasar candidates for the miniJPAS based on the
combination of multiple machine learning algorithms. The *point_like*
catalogue contains only point-like sources and should be more
reliable. The other file includes all objects, both point-like and
extended sources. See Appendix D of the paper for a detailed
description of the columns.
Convolutional Neural Network (CNN1, CNN1NE and CCN2), Random Forest (RF),
LightGBM (LGBM) classifiers from Rodrigues et al. (2023MNRAS.520.3494R 2023MNRAS.520.3494R).
Artificial Neural Network (ANN1 and ANN2) from Martinez-Solaeche et al.
(2023A&A...673A.103M 2023A&A...673A.103M).
Random Forest (SQUEzE) classifier from Perez-Rafols et al.
(2023A&A...678A.144P 2023A&A...678A.144P).
Template-based fitting (QPz and LePhare) form Queiroz, private comm.
File Summary:
--------------------------------------------------------------------------------
FileName Lrecl Records Explanations
--------------------------------------------------------------------------------
ReadMe 80 . This file
catall.dat 592 40805 Catalogue of quasar candidates for the miniJPAS,
all objects, both point-like and extended sources
catps.dat 592 6654 Catalogue of quasar candidates for the miniJPAS,
only point-like sources
--------------------------------------------------------------------------------
Byte-by-byte Description of file: catall.dat catps.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 4 I4 --- TileId ID of the Tile image of the object
(TILE_ID)
6- 10 I5 --- Number Number ID assigned by Sextractor to
object (NUMBER)
12 A1 --- ClassTrue [-] Correct classification (if available)
(CLASS_TRUE)
14- 20 F7.3 deg RAdeg Right Ascension (J2000) of the object
(RA)
22- 28 F7.4 deg DEdeg Declination (J2000) of the object (DEC)
30- 36 F7.4 mag rmag r-band APER-3 magnitude of the object
(RSDSS)
38- 45 F8.6 --- PtotStar Probability of the object being
point-like (TOTPROBSTAR)
47- 54 F8.4 --- e_PtotStar ?=-99 Probability of the object being
point-like error (ERTPROBSTAR)
56 I1 --- IsPointLike [0/1] 1 for point-like objects,
0 otherwise (ISPOINTLIKE)
58- 66 I9 --- SpecId Unique object identifier (SPECID)
68- 79 E12.6 --- ConfStarCNN1 Confidence of object being a star (CNN1)
(CONFSTARCNN1)
81- 92 E12.6 --- ConfGalCNN1 Confidence of object being a galaxy
(CNN1) (CONFGALCNN1)
94-105 E12.6 --- ConfLQSOCNN1 Confidence of object being a z<2.1 quasar
(CNN1) (CONFLQSOCNN1)
107-118 E12.6 --- ConfHQSOCNN1 Confidence of object being a z≥2.1
quasar (CNN1) (CONFHQSOCNN1)
120 I1 --- ClassCNN1 [0/3] Object class (CNN1) (CLASS_CNN1)
122-133 E12.6 --- ConfStarCNN1NE Confidence of object being a star
(CNN1NECONFSTARCNN1NE)
135-146 E12.6 --- ConfGalCNN1NE Confidence of object being a galaxy
(CNN1NECONFGALCNN1NE)
148-159 E12.6 --- ConfLQSOCNN1NE Confidence of object being a z<2.
1 quasar (CNN1NECONFLQSOCNN1NE)
161-172 E12.6 --- ConfHQSOCNN1NE Confidence of object being a z≥2.1
quasar (CNN1NECONFHQSOCNN1NE)
174 I1 --- ClassCNN1NE [0/3] Object class (CNN1NECLASS_CNN1NE)
176-187 E12.6 --- ConfStarCNN2 Confidence of object being a star (CNN2)
(CONFSTARCNN2)
189-200 E12.6 --- ConfGalCNN2 Confidence of object being a galaxy
(CNN2) (CONFGALCNN2)
202-213 E12.6 --- ConfLQSOCNN2 Confidence of object being a z<2.1 quasar
(CNN2) (CONFLQSOCNN2)
215-226 E12.6 --- ConfHQSOCNN2 Confidence of object being a z≥2.1
quasar (CNN2) (CONFHQSOCNN2)
228 I1 --- ClassCNN2 [0/3] Object class (CNN2) (CLASS_CNN2)
230-237 F8.6 --- ConfStarRF Confidence of object being a star (RF)
(CONFSTARRF)
239-246 F8.6 --- ConfGalRF Confidence of object being a galaxy (RF)
(CONFGALRF)
248-255 F8.6 --- ConfLQSORF Confidence of object being a z<2.1 quasar
(RF) (CONFLQSORF)
257-264 F8.6 --- ConfHQSORF Confidence of object being a z≥2.1
quasar (RF) (CONFHQSORF)
266 I1 --- ClassRF [0/3] Object class (RF) (CLASS_RF)
268-279 E12.6 --- ConfStarLGBM Confidence of object being a star (LGBM)
(CONFSTARLGBM)
281-292 E12.6 --- ConfGalLGBM Confidence of object being a galaxy
(LGBM) (CONFGALLGBM)
294-305 E12.6 --- ConfLQSOLGBM Confidence of object being a z<2.1 quasar
(LGBM) (CONFLQSOLGBM)
307-318 E12.6 --- ConfHQSOLGBM Confidence of object being a z≥2.1
quasar (LGBM) (CONFHQSOLGBM)
320 I1 --- ClassLGBM [0/3] Object class (LGBM) (CLASS_LGBM)
322-333 E12.6 --- ConfGalANN1 Confidence of object being a galaxy
(ANN1) (CONFGALANN1)
335-346 E12.6 --- ConfHQSOANN1 Confidence of object being a z≥2.1
quasar (ANN1) (CONFHQSOANN1)
348-359 E12.6 --- ConfLQSOANN1 Confidence of object being a z<2.1
quasar (ANN1) (CONFLQSOANN1)
361-372 E12.6 --- ConfStarANN1 Confidence of object being a star (ANN1)
(CONFSTARANN1)
374 I1 --- ClassANN1 [0/3] Object class (ANN1) (CLASS_ANN1)
376-387 E12.6 --- ConfGalANN2 Confidence of object being a galaxy
(ANN2) (CONFGALANN2)
389-400 E12.6 --- ConfHQSOANN2 Confidence of object being a z≥2.1
quasar (ANN2) (CONFHQSOANN2)
402-413 E12.6 --- ConfLQSOANN2 Confidence of object being a z<2.1
quasar (ANN2) (CONFLQSOANN2)
415-426 E12.6 --- ConfStarANN2 Confidence of object being a star (ANN2)
(CONFSTARANN2)
428 I1 --- ClassANN2 [0/3] Object class (ANN2) (CLASS_ANN2)
430-431 I2 --- ClassSQUEzE [-1/3] Object class (SQUEZE)
(CLASS_SQUEZE)
433-441 F9.6 --- ConfSQUEzE-0 Confidence of object being a z<2.1 quasar
(SQUEZE) (CONFSQUEZE0)
443-451 F9.6 --- zSQUEzE-0 SQUEZE redshift estimate (ZSQUEZE0)
453-461 F9.6 --- ConfSQUEzE-1 Confidence of object being a z<2.1 quasar
(SQUEZE) (CONFSQUEZE1)
463-471 F9.6 --- zSQUEzE-1 SQUEZE redshift estimate (ZSQUEZE1)
473-481 F9.6 --- ConfSQUEzE-2 Confidence of object being a z<2.1 quasar
(SQUEZE) (CONFSQUEZE2)
483-491 F9.6 --- zSQUEzE-2 SQUEZE redshift estimate (ZSQUEZE2)
493-501 F9.6 --- ConfSQUEzE-3 Confidence of object being a z<2.1 quasar
(SQUEZE) (CONFSQUEZE3)
503-511 F9.6 --- zSQUEzE-3 SQUEZE redshift estimate (ZSQUEZE3)
513-521 F9.6 --- ConfSQUEzE-4 Confidence of object being a z<2.1 quasar
(SQUEZE) (CONFSQUEZE4)
523-531 F9.6 --- zSQUEzE-4 SQUEZE redshift estimate (ZSQUEZE4)
533-537 F5.3 --- zphQPZ QPz redshift estimate (ZPHOT_QPZ)
539-545 F7.3 --- zphLePhare LePhare redshift estimate (ZPHOT_LEPHARE)
547-554 F8.6 --- ConfStar Final confidence of object being a star
(CONF_STAR)
556-563 F8.6 --- ConfGal Final confidence of object being a galaxy
(CONF_GAL)
565-572 F8.6 --- ConfLQSO Final confidence of object being a z<2.1
quasar (CONF_LQSO)
574-581 F8.6 --- ConfHQSO Final confidence of object being a z≥2.1
quasar (CONF_HQSO)
583 I1 --- Class Final object class (CLASS)
585-592 F8.6 --- z ?=- Final redshift estimate (Z)
--------------------------------------------------------------------------------
Acknowledgements:
Ignasi Perez-Rafols, ignasi.perez.rafols(at)upc.edu
References:
Queiroz et al., Paper I 2023MNRAS.520.3476Q 2023MNRAS.520.3476Q
Rodrigues et al., Paper II 2023MNRAS.520.3494R 2023MNRAS.520.3494R
Martinez-Solaeche et al., Paper III 2023A&A...673A.103M 2023A&A...673A.103M
Perez-Rafols et al., Paper IV 2023A&A...678A.144P 2023A&A...678A.144P
(End) Patricia Vannier [CDS] 14-Nov-2025