Supercompact Ultraviolet Luminous Galaxies

• Supercompact UV luminous galaxies are compact star-forming systems defined by extreme UV surface brightness (≥10⁹ L☉ kpc⁻²) and intense burst activity.
• They act as local analogs to high-z Lyman-break galaxies, bridging detailed observations at low redshift with early universe properties.
• Multi-wavelength studies reveal clumpy dynamics, rapid evolution, varied dust attenuation, and signatures of unstable, feedback-driven star formation.

Supercompact ultraviolet luminous galaxies are very compact star-forming systems with extremely high ultraviolet luminosity and, more characteristically, very high ultraviolet surface brightness, typically $$I_{\rm FUV} \gtrsim 10^{9}\,L_\odot\,\mathrm{kpc}^{-2}$$. In the nearby universe they are widely used as low-redshift analogs of high-$z$ Lyman-break galaxies, and in the literature they overlap strongly with the high-surface-brightness subset of ultraviolet luminous galaxies, the Lyman Break Analogs, and the most extreme compact blue starbursts (Laag et al., 2010). Across local and intermediate-redshift samples, the same physical regime recurs: compact sizes, high specific star-formation rates, strong nebular emission, elevated ionization, and low to moderate metallicity; at higher redshift, JWST has identified comparably luminous and compact UV sources at cosmic dawn, extending the relevance of the class far beyond the local universe (Santos-Junior et al., 15 Jul 2025).

1. Definition and taxonomic placement

The canonical UVLG definition is based on integrated far-ultraviolet luminosity, $L_{\rm FUV} > 2\times10^{10}\,L_\odot$, while the compact and supercompact subclasses are defined by far-ultraviolet surface brightness: $$I_{\rm FUV} \gtrsim 10^{8}\,L_\odot\,\mathrm{kpc}^{-2}$$ for compact UVLGs and $$I_{\rm FUV} \gtrsim 10^{9}\,L_\odot\,\mathrm{kpc}^{-2}$$ for supercompact UVLGs (Laag et al., 2010). In the Hoopes et al. scheme, the high-surface-brightness UVLGs are the Lyman Break Analogs; the recent OSIRIS/Keck Pa$\alpha$ study explicitly identifies its sample as 18 Supercompact Ultraviolet Luminous Galaxies drawn from that LBA population (Santos-Junior et al., 15 Jul 2025).

Several adjacent categories overlap with this definition but are selected in different observables. Luminous Compact Blue Galaxies are defined optically by $M_B \le -18.5$, $(B-V)_0 \le 0.6$, and $\mu_e(B) \le 21\,\mathrm{mag\,arcsec^{-2}}$, with

$$\mu_e = M_B + 2.5\log_{10}\!\left(2\pi R_e^2\right) + 36.572,$$

and the LCBG literature explicitly notes overlap with UVLGs as defined by Heckman (Hunt et al., 2021). Luminous compact galaxies selected through strong nebular emission occupy a closely related regime: the 803 LCGs of Izotov et al. are compact, metal-poor, high-sSFR systems intermediate between nearby blue compact dwarfs and high-$z$0 LBGs (Parnovsky et al., 2012). At lower masses, the VUDS extreme emission-line galaxies are not explicitly labeled UVLGs, but they are characterized by $z$1, very high $z$2 equivalent widths, low metallicity, and $z$3–$z$4, making them low-mass analogs of the same compact starburst regime (Amorín et al., 2014).

This taxonomic structure indicates that “supercompact UVLG” is both a strict UV surface-brightness definition and a physically broader regime encompassing compact, UV-intense, high-excitation starbursts selected through UV, emission-line, or blue-continuum criteria.

2. Physical properties in representative low- and intermediate-redshift samples

The nearby and intermediate-redshift literature resolves supercompact UVLGs into a consistent set of global observables: high star-formation rates, small effective radii, high specific star-formation rates, and low to moderate metallicity. The OSIRIS/Keck Pa$z$5 sample of 18 SC-UVLGs at $z$6–0.25 has stellar masses $z$7–10.9, global star-formation rates of $z$8–$z$9, and effective radii $L_{\rm FUV} > 2\times10^{10}\,L_\odot$0–2.8 kpc, with many systems $L_{\rm FUV} > 2\times10^{10}\,L_\odot$1 kpc (Santos-Junior et al., 15 Jul 2025). These galaxies were selected specifically because they share many properties with star-forming galaxies at cosmic noon.

The 803 luminous compact galaxies analyzed in the GALEX+SDSS statistical study provide a large comparison set. They lie at $L_{\rm FUV} > 2\times10^{10}\,L_\odot$2–0.63, satisfy $L_{\rm FUV} > 2\times10^{10}\,L_\odot$3 and $L_{\rm FUV} > 2\times10^{10}\,L_\odot$4, and have oxygen abundances $L_{\rm FUV} > 2\times10^{10}\,L_\odot$5–8.4 with a median near 8.1 (Parnovsky et al., 2012). Their star-formation rates span $L_{\rm FUV} > 2\times10^{10}\,L_\odot$6–$L_{\rm FUV} > 2\times10^{10}\,L_\odot$7 from H$L_{\rm FUV} > 2\times10^{10}\,L_\odot$8, $L_{\rm FUV} > 2\times10^{10}\,L_\odot$9–$$I_{\rm FUV} \gtrsim 10^{8}\,L_\odot\,\mathrm{kpc}^{-2}$$0 from FUV, and $$I_{\rm FUV} \gtrsim 10^{8}\,L_\odot\,\mathrm{kpc}^{-2}$$1–$$I_{\rm FUV} \gtrsim 10^{8}\,L_\odot\,\mathrm{kpc}^{-2}$$2 from NUV, with $$I_{\rm FUV} \gtrsim 10^{8}\,L_\odot\,\mathrm{kpc}^{-2}$$3–$$I_{\rm FUV} \gtrsim 10^{8}\,L_\odot\,\mathrm{kpc}^{-2}$$4 (Parnovsky et al., 2012). The same study shows that for burst ages $$I_{\rm FUV} \gtrsim 10^{8}\,L_\odot\,\mathrm{kpc}^{-2}$$5 Myr, H$$I_{\rm FUV} \gtrsim 10^{8}\,L_\odot\,\mathrm{kpc}^{-2}$$6, FUV, and NUV luminosities per unit young stellar mass decline approximately exponentially, with half-times of about 1.1, 1.6, and 2.1 Myr, respectively (Parnovsky et al., 2012). That temporal behavior is central to the interpretation of supercompact UVLGs as very young bursts rather than long-lived equilibrium disks.

A complementary view comes from the optically selected Multi-wavelength Extreme Starburst Sample at $$I_{\rm FUV} \gtrsim 10^{8}\,L_\odot\,\mathrm{kpc}^{-2}$$7. The 138 MESS galaxies have DR7 total star-formation rates of 11–$$I_{\rm FUV} \gtrsim 10^{8}\,L_\odot\,\mathrm{kpc}^{-2}$$8 with a median of $$I_{\rm FUV} \gtrsim 10^{8}\,L_\odot\,\mathrm{kpc}^{-2}$$9, a median $$I_{\rm FUV} \gtrsim 10^{9}\,L_\odot\,\mathrm{kpc}^{-2}$$0, and Balmer-derived median $$I_{\rm FUV} \gtrsim 10^{9}\,L_\odot\,\mathrm{kpc}^{-2}$$1 (Laag et al., 2010). Among the 70 MESS objects detected in GALEX FUV, 20 satisfy the UVLG luminosity threshold, about 17 would qualify as compact UVLGs, and a handful plausibly fall into the supercompact UVLG regime, although the paper stresses that SDSS-based radii are too uncertain for a robust supercompact fraction (Laag et al., 2010).

At lower masses, the VUDS EELGs establish that the compact starburst regime extends well below the classical UVLG mass scale. Their stellar masses are $$I_{\rm FUV} \gtrsim 10^{9}\,L_\odot\,\mathrm{kpc}^{-2}$$2, circularized half-light radii are $$I_{\rm FUV} \gtrsim 10^{9}\,L_\odot\,\mathrm{kpc}^{-2}$$3–0.8 kpc, and metallicities are $$I_{\rm FUV} \gtrsim 10^{9}\,L_\odot\,\mathrm{kpc}^{-2}$$4, with four galaxies below $$I_{\rm FUV} \gtrsim 10^{9}\,L_\odot\,\mathrm{kpc}^{-2}$$5 (Amorín et al., 2014). This suggests that supercompact UV intensity is not confined to one mass scale; rather, it is a mode of star formation that appears in both dwarf and $$I_{\rm FUV} \gtrsim 10^{9}\,L_\odot\,\mathrm{kpc}^{-2}$$6-scale systems.

3. Dust attenuation, infrared power, and the meaning of UV brightness

A central result of the mid-infrared literature is that UV brightness does not imply negligible dust attenuation. The Spitzer/IRS analysis of 571 starbursts compares PAH luminosity to rest-UV emission and finds that the median intrinsic-to-observed UV luminosity ratio is $$I_{\rm FUV} \gtrsim 10^{9}\,L_\odot\,\mathrm{kpc}^{-2}$$7 for infrared-selected starbursts and $$I_{\rm FUV} \gtrsim 10^{9}\,L_\odot\,\mathrm{kpc}^{-2}$$8 for ultraviolet-selected starbursts, including luminous GALEX sources and Markarian galaxies (Houck et al., 2010). The same study concludes that extinction corrections commonly applied to UV-selected galaxies are underestimated by factors of 2–3, and that the starbursts appearing most luminous in the ultraviolet are those with the least obscuration (Houck et al., 2010).

This has a direct bearing on supercompact UVLGs. In the IRS framework, a UV-selected compact starburst with observed $$I_{\rm FUV} \gtrsim 10^{9}\,L_\odot\,\mathrm{kpc}^{-2}$$9 is more appropriately interpreted as having $\alpha$0 on average, with a corresponding increase in inferred star-formation rate (Houck et al., 2010). The implication is not that supercompact UVLGs are dust-free, but that they occupy the low-obscuration wing of a broader starburst population.

The MESS sample places that conclusion in a local multiwavelength context. Its median $\alpha$1 is higher than the median $\alpha$2 reported for the Hoopes UVLG control sample, but lower than typical values for 1 Jy ULIRGs and the IRAS Bright Galaxy Sample (Laag et al., 2010). UV-luminous MESS galaxies tend to have lower $\alpha$3 and $\alpha$4 than the more obscured objects in the same sample, reinforcing the picture that strong observed UV output signals comparatively transparent geometry rather than low total dust mass (Laag et al., 2010).

Accordingly, supercompact UVLGs are best understood as systems in which intense star formation remains visible in the ultraviolet despite substantial reprocessing. Their observed UV continuum traces a real starburst, but not necessarily the full bolometric power.

4. Clumps, turbulence, and internal dynamical state

High-resolution integral-field spectroscopy shows that supercompact UVLGs are not monolithic UV sources. In the OSIRIS/Keck Pa$\alpha$5 survey, 18 SC-UVLGs contain 84 star-forming clumps with typical sizes of a few hundred parsecs; 38 are resolved, and their radii typically fall in the 200–500 pc range (Santos-Junior et al., 15 Jul 2025). The clumps exhibit low velocity shear, with mean $\alpha$6 of roughly 11–13 km s$\alpha$7, and high velocity dispersion, with mean $\alpha$8–76 km s$\alpha$9 (Santos-Junior et al., 15 Jul 2025). Their dynamical masses are estimated with

$M_B \le -18.5$0

and are typically higher than gas masses inferred from the Schmidt–Kennicutt relation,

$M_B \le -18.5$1

a result interpreted as evidence that the clumps contain both stellar mass and strong non-virial motions driven by feedback (Santos-Junior et al., 15 Jul 2025).

The same study computes clump-scale Toomre parameters,

$M_B \le -18.5$2

with $M_B \le -18.5$3, and finds that 17 of 31 resolved clumps have $M_B \le -18.5$4, while 29 of 31 have $M_B \le -18.5$5 relative to their host disks (Santos-Junior et al., 15 Jul 2025). Clumps in SC-UVLGs are therefore more unstable than the disks that contain them, consistent with in situ fragmentation in gas-rich, turbulent systems.

The WiggleZ AO-assisted H$M_B \le -18.5$6 kinematic study extends the same physical picture to $M_B \le -18.5$7. Among 13 luminous UV-selected star-forming galaxies, four show multiple 1–2 kpc H$M_B \le -18.5$8 clumps across 6–10 kpc, five show resolved compact emission with $M_B \le -18.5$9 kpc, and the remaining four show extended H$(B-V)_0 \le 0.6$0 emission (Wisnioski et al., 2011). Seven of the 13 are well fit by rotating disk models, yet the highest velocity dispersions, $(B-V)_0 \le 0.6$1, occur in the most compact sources (Wisnioski et al., 2011). The paper interprets these systems as gaseous unstable disks and links their kinematics to cold dense gas flows from the intergalactic medium (Wisnioski et al., 2011).

At still lower masses, HST F275W/F160W imaging reveals a related compact mode in field dwarfs at $(B-V)_0 \le 0.6$2. The UV-to-NIR size ratio follows

$(B-V)_0 \le 0.6$3

so low-mass galaxies with $(B-V)_0 \le 0.6$4 often have $(B-V)_0 \le 0.6$5, indicating outside-in growth with compact UV cores (Cheng et al., 2019). This broadens the structural interpretation of supercompact UV activity: compact UV emission may trace central compaction, clump coalescence, or bursty nuclear star formation, not a single morphological pathway.

5. Population statistics and evolution to $(B-V)_0 \le 0.6$6

The luminosity-function literature demonstrates that very UV-luminous compact starbursts are both cosmologically significant and rapidly evolving. Using 39,996 GALEX-selected galaxies with spectroscopic redshifts from WiggleZ, the NUV luminosity function was measured in 0.05 redshift intervals over $(B-V)_0 \le 0.6$7 (Jurek et al., 2013). At $(B-V)_0 \le 0.6$8, the median star-formation rate of the sample is in the upper 95th percentile of optically selected galaxies, and $(B-V)_0 \le 0.6$9 corresponds approximately to $\mu_e(B) \le 21\,\mathrm{mag\,arcsec^{-2}}$0 (Jurek et al., 2013). For the most luminous galaxies, $\mu_e(B) \le 21\,\mathrm{mag\,arcsec^{-2}}$1, the number density declines as $\mu_e(B) \le 21\,\mathrm{mag\,arcsec^{-2}}$2 from $\mu_e(B) \le 21\,\mathrm{mag\,arcsec^{-2}}$3 to $\mu_e(B) \le 21\,\mathrm{mag\,arcsec^{-2}}$4, and the star-formation-rate density of these very luminous galaxies evolves as $\mu_e(B) \le 21\,\mathrm{mag\,arcsec^{-2}}$5 (Jurek et al., 2013). They contribute only about 1 per cent of cosmic star formation over $\mu_e(B) \le 21\,\mathrm{mag\,arcsec^{-2}}$6–0.9, yet the bright end of the luminosity function at all redshifts greater than 0.55 is not well described by a pure Schechter form because of an excess at $\mu_e(B) \le 21\,\mathrm{mag\,arcsec^{-2}}$7 (Jurek et al., 2013).

A parallel optical view is provided by the COSMOS LCBG luminosity function. Between $\mu_e(B) \le 21\,\mathrm{mag\,arcsec^{-2}}$8 and $\mu_e(B) \le 21\,\mathrm{mag\,arcsec^{-2}}$9, the characteristic luminosity $$\mu_e = M_B + 2.5\log_{10}\!\left(2\pi R_e^2\right) + 36.572,$$0 brightens by only $$\mu_e = M_B + 2.5\log_{10}\!\left(2\pi R_e^2\right) + 36.572,$$1 mag, but the number density increases by a factor of four (Hunt et al., 2021). LCBGs constitute roughly 18 per cent of galaxies more luminous than $$\mu_e = M_B + 2.5\log_{10}\!\left(2\pi R_e^2\right) + 36.572,$$2 at $$\mu_e = M_B + 2.5\log_{10}\!\left(2\pi R_e^2\right) + 36.572,$$3, and roughly 54 per cent at $$\mu_e = M_B + 2.5\log_{10}\!\left(2\pi R_e^2\right) + 36.572,$$4 (Hunt et al., 2021). Because that literature explicitly places higher-mass LCBGs in overlap with UVLGs, this suggests that supercompact UVLGs represent the high-UV, high-surface-brightness tail of a much more common compact blue star-forming population at earlier epochs (Hunt et al., 2021).

The broader implication is that the supercompact UV mode is not a dominant channel of cosmic star formation by volume density, but it becomes progressively more common toward $$\mu_e = M_B + 2.5\log_{10}\!\left(2\pi R_e^2\right) + 36.572,$$5. Its strong evolutionary signal tracks the broader decline of compact, intense star formation since that epoch.

6. Cosmic-dawn counterparts and current interpretive tensions

JWST has extended the relevance of the supercompact UV-luminous regime to $$\mu_e = M_B + 2.5\log_{10}\!\left(2\pi R_e^2\right) + 36.572,$$6. The UNCOVER survey measured a luminosity function over $$\mu_e = M_B + 2.5\log_{10}\!\left(2\pi R_e^2\right) + 36.572,$$7 at $$\mu_e = M_B + 2.5\log_{10}\!\left(2\pi R_e^2\right) + 36.572,$$8 and found that the number density of bright galaxies with $$\mu_e = M_B + 2.5\log_{10}\!\left(2\pi R_e^2\right) + 36.572,$$9 is 10–100 times larger than theoretical predictions and previous HST-based estimates (Chemerynska et al., 2023). A double power law fits the bright end better than a Schechter function, and the inferred star-formation-rate density is

$z$00

about 4–10 times higher than constant-efficiency galaxy formation models predict (Chemerynska et al., 2023). Although that work does not directly measure sizes, its completeness simulations adopt high-$z$01 size–luminosity relations with effective radii of order a few hundred parsecs to $z$02 kpc, placing these objects in a compact regime consistent with the “supercompact” designation (Chemerynska et al., 2023).

The JADES spectroscopic confirmation of two luminous galaxies at $z$03 and $z$04 provides a sharper structural benchmark (Carniani et al., 2024). JADES-GS-z14-0 has $z$05 and a resolved UV half-light radius of $z$06 pc, while JADES-GS-z14-1 has $z$07 and $z$08 pc (Carniani et al., 2024). Both show blue ultraviolet slopes, $z$09 and $z$10, no strong emission lines, and are interpreted as stellar-continuum-dominated systems; the paper explicitly concludes that the excess of luminous galaxies in the early universe cannot be entirely explained by accretion onto black holes (Carniani et al., 2024).

That conclusion coexists with an alternative interpretation proposed for the bright-end UV luminosity function at $z$11. In the supermassive primordial black hole scenario, objects with $z$12, $z$13, and sub-Eddington $z$14 can enhance the bright end of the UV luminosity function and naturally generate very compact UV emission from accretion-powered sources (Li et al., 3 Sep 2025). The present observational situation therefore admits two non-exclusive statements: some of the brightest compact UV sources at cosmic dawn are demonstrably stellar systems, and non-stellar compact UV emission remains an active explanatory framework for the bright-end excess.

This suggests that the concept of the supercompact ultraviolet luminous galaxy now spans two regimes. In the local and intermediate-redshift universe it denotes compact, high-surface-brightness starbursts used as resolved analogs of LBGs. At cosmic dawn it also names an observational phenotype—extreme UV luminosity emerging from sub-kpc scales—whose detailed power source can be stellar, accretion-driven, or composite depending on the object.