Optically-compact star-forming galaxies (SFGs) have been proposed as immediate progenitors of quiescent galaxies, although their origin and nature are debated. Were they formed in slow secular processes or in rapid merger-driven starbursts? Addressing this question would provide fundamental insight into how quenching occurs. We explore the location of the general population of galaxies with respect to fundamental star-forming and structural relations, identify compact SFGs based on their stellar core densities, and study three diagnostics of the burstiness of star formation: 1) Star formation efficiency, 2) interstellar medium (ISM), and 3) radio emission. The overall distribution of galaxies in the fundamental relations points towards a smooth transition towards quiescence while galaxies grow their stellar cores, although some galaxies suddenly increase their specific star-formation rate when they become compact. From their star formation efficiencies compact and extended SFGs appear similar. In relation to the ISM diagnostic, by studying the CO excitation, the density of the neutral gas, and the strength of the ultraviolet radiation field, compact SFGs resemble galaxies located in the upper envelope of the SFGs main sequence, although yet based on a small sample size. Regarding the radio emission diagnostic we find that galaxies become increasingly compact as the starburst ages, implying that at least some compact SFGs are old starbursts. We suggest that compact SFGs could be starburts winding down and eventually crossing the main sequence towards quiescence.

In this data paper we present and characterise the multi-component radio sources identified in the VLA-COSMOS Large Project at 3 GHz (0.75 arcsec resolution, 2.3 {\mu}Jy/beam rms), i.e. the radio sources which are composed of two or more radio blobs.The classification of objects into multi-components was done by visual inspection of 351 of the brightest and most extended blobs from a sample of 10,899 blobs identified by the automatic code blobcat. For that purpose we used multi-wavelength information of the field, such as the 1.4-GHz VLA-COSMOS data and the UltraVISTA stacked mosaic available for COSMOS. We have identified 67 multi-component radio sources at 3 GHz: 58 sources with AGN powered radio emission and 9 star-forming galaxies. We report 8 new detections that were not observed by the VLA-COSMOS Large Project at 1.4 GHz, due to the slightly larger area coverage at 3 GHz. The increased spatial resolution of 0.75 arcsec has allowed us to resolve (and isolate) multiple emission peaks of 28 extended radio sources not identified in the 1.4-GHz VLA-COSMOS map. We report the multi-frequency flux densities (324 MHz, 325 MHz, 1.4 GHz & 3 GHz), star-formation-rates, and stellar masses of these objects. Multi-component objects at 3-GHz VLA-COSMOS inhabit mainly massive galaxies (>10^10.5 Msun). The majority of the multi-component AGN lie below the main-sequence of star-forming galaxies (SFGs), in the green valley and the quiescent region. We provide detailed description of the objects: amongst the AGN there are 2 head-tail, 10 core-lobe, 9 wide-angle-tail (WAT), 8 double-double or Z-/X-shaped, 3 bent-tail radio sources, and 26 symmetric sources, while amongst the SFGs we find the only star-forming ring seen in radio emission in COSMOS. We report a large number (32/58) of disturbed/bent multi-component AGN, 18 of which do not lie within X-ray groups in COSMOS (0.08 < z < 1.53). [abridged]

We construct the average radio spectral energy distribution (SED) of highly star-forming galaxies (HSFGs) up to z~4. Infrared and radio luminosities are bound by a tight correlation that is defined by the so-called q parameter. This infrared-radio correlation provides the basis for the use of radio luminosity as a star-formation tracer. Recent stacking and survival analysis studies find q to be decreasing with increasing redshift. It was pointed out that a possible cause of the redshift trend could be the computation of rest-frame radio luminosity via a single power-law assumption of the star-forming galaxies' (SFGs) SED.To test this, we constrained the shape of the radio SED of a sample of HSFGs. To achieve a broad rest-frame frequency range, we combined previously published VLA observations of the COSMOS field at 1.4 GHz and 3 GHz with unpublished GMRT observations at 325 MHz and 610 MHz by employing survival analysis to account for non-detections in the GMRT maps. We selected a sample of HSFGs in a broad redshift range (0.3<z<4,SFR>100M0/yr) and constructed the average radio SED. By fitting a broken power-law, we find that the spectral index changes from $\alpha_1=0.42\pm0.06$ below a rest-frame frequency of 4.3 GHz to $\alpha_2=0.94\pm0.06$ above 4.3 GHz. Our results are in line with previous low-redshift studies of HSFGs (SFR>10M0/yr) that show the SED of HSFGs to differ from the SED found for normal SFGs (SFR<10M0/yr). The difference is mainly in a steeper spectrum around 10 GHz, which could indicate a smaller fraction of thermal free-free emission. Finally, we also discuss the impact of applying this broken power-law SED in place of a simple power-law in K-corrections of HSFGs and a typical radio SED for normal SFGs drawn from the literature. We find that the shape of the radio SED is unlikely to be the root cause of the q-z trend in SFGs.

The infrared-radio correlation (IRRC) offers a way to assess star formation from radio emission. Multiple studies found the IRRC to decrease with increasing redshift. This may in part be due to the lack of knowledge about the possible radio spectral energy distributions (SEDs) of star-forming galaxies. We constrain the radio SED of a complete sample of highly star-forming galaxies ($SFR>100\,\mathrm{M_{\odot}}/\,\mathrm{yr}$) based on the VLA-COSMOS $1.4\,\mathrm{GHz}$ Joint and $3\,\mathrm{GHz}$ Large Project catalogs. We reduce archival GMRT $325\,\mathrm{MHz}$ and $610\,\mathrm{MHz}$ observations, broadening the rest-frame frequency range to $0.3-15\,\mathrm{GHz}$. Employing survival analysis and fitting a double power law SED, we find that the slope steepens from a spectral index of $\alpha_1=0.51\pm 0.04$ below $4.5\,\mathrm{GHz}$ to $\alpha_2=0.98\pm0.07$ above $4.5\,\mathrm{GHz}.$ Our results suggest that the use of a K-correction assuming a single power-law radio SED for star forming galaxies is likely not the root cause of the IRRC trend.