The Universe at high-redshift is quite different from the local Universe that we are familiar. At the epoch of z-2, a large fraction of massive galaxies are still very active, and their morphologies are not the same as that of galaxies in the local Universe. To check the current theoretical model for galaxy formation and evolution, studies per-formed at large ranges of redshift is needed. In the past decades, many novel techniques have been applied to the multi-band observations for high-redshift galaxies, and many results make us understand better for the high-redshift galaxies. Compared to the local galaxies, the study on high-redshift galaxies (especially at z-2) is very important. Firstly, the cosmic star formation rate density (SFRD) and the number density of QSOs peaked at z-2; secondly, the types and morphologies of galaxies changed dramatically at z-2; thirdly, the massive galaxies at z-2 are suposed to be the progenitors of the local elliptical galaxies, studies on these high-redshift galaxies provide constraints on the theoretical model. Thus our study focus on the redshift at z-2. In this thesis, our main works are as follows:(1) Physical properties of distant red galaxies in the COS-MOS/UltraVISTA field. (2) Physical properties of compact star-forming galaxies in the COSMOS/UltraVISTA field. (3) The size of passive galaxies as regulated by their formation redshift.We present a study on physical properties for a large distant red galaxy (DRG) sam-ple, using the K-selected multi-band photometry catalog of the COSMOS/UltraVISTA field and the CANDELS NIR data. Our sample includes 4485 DRGs with (J-K)AB> 1.16 and KAB<23.4 mag, and 132 DRGs have HST/WFC3 morphological measure-ments. The results of nonparametric measurements of DRG morphology are consistent with our rest-frame UVJ color classification:quiescent DRGs are generally compact while star-forming DRGs tend to have extended structures. We find the star formation rate (SFR) and the stellar mass of star-forming DRGs present tight "main sequence" relations in all redshift bins. The infrared (IR) derived SFR dominate the total SFR of DRGs which occupy the high-mass range, implying that the J - K color criterion effectively selects massive and dusty galaxies. DRGs with higher M* generally have redder (U - V)rest colors, and the (U - V)rest colors of DRGs become bluer at higher redshifts, suggesting high-mass galaxies have higher internal dust extinctions or older stellar ages and they evolve with time. We find that DRGs have different overlaps with EROs, BzKs, IEROs and high-z ULIRGs indicating DRGs is not a special population and they can also be selected by other color criteria.We present a study on the physical properties of compact star-forming galaxies (cSFGs) with M*≥1010M(?) and 2≤ z≤ 3 in the COSMOS and GOODS-S fields, to explore whether they are a transitional type of galaxies between extended star-forming galaxies (eSFGs) and compact quiescent galaxies (cQGs) on the evolutionary path. The cSFGs distribute at nearly the same locus on the main sequence as eSFGs and dominate the high mass end. On the rest-frame U - V vs. V-J and U - B vs. MB diagrams, cSFGs mainly distribute at the middle of eSFGs and cQGs in all colors, but are more inclined to "red sequence" than "green valley" galaxies. We find cSFGs have similar distributions with cQGs on the nonparametric morphology diagrams. About one third of cSFGs show signatures of post mergers, and nearly none of them can be recognized as disks. The general distributions of cSFGs on stellar population and structural pa-rameters are very similar to cQGs, implying the cSFGs are the direct progenitors of cQGs. In the meanwhile, cSFGs show the highest fraction of AGNs compared to eS-FGs and cQGs, supporting the expectation that the transformation from eSFG to cSFG by gas-rich dissipative processes will intensify both the star formation and the black hole growth. The results of all the analysis described above suggest that cSFG is likely to be a transitional type of galaxy between eSFG towards cQG at z> 2.Finally, we present a study on the size evolution with the formation redshift based on a passive BzK selected sample, to explore whether the observed size growth of pas-sive galaxies with passing time is caused by the later comers of newly quenched galaxies with larger sizes at lower redshifts. We select a sample of 2738 passive galaxies in the COSMOS/UltraVISTA field with photometric redshift z-1.3-3 and average stellar masses M*~1010.7M(?). We determine the stellar ages by SED fitting method with multiple metallicity based on the photometry catalog of Ultra VISTA, then we use this age to derive the formation redshift of galaxies. Based on the ACS I-band and CFHT Ks-band images, we do stacking and measure their average sizes for five bins which are divided on the (1+Zform)-1 axis. We find that the sizes of passive galaxies approxi-mately scale with formation redshift as re ∠(1+Zform)-1 at lower redshifts (zform-2.3), but when it towards to higher redshifts (Zform-4 - 9) the trend of sizes tend to be flat and close to about lkpc. This result indicates that at lower redshifts the newly quenched galaxies with larger sizes, rather than the minor mergers, are mainly responsible for the apparent size growth of the passive galaxies. |