Updated text, figures. -1435215739 Assigned ID: 351 ----- Attempting Submission 1 (Fri Apr 08 01:56:21 GMT 2016) ----- HST Phase I Proposal 351 (draft_hst_ar_proposal.apt) successfully submitted. Receipt: # 351-1 ----- Attempting Submission 2 (Fri Apr 08 23:40:22 GMT 2016) ----- Correlations have played an important role in advancing our knowledge of astrophysics, from the Schmidt-Kennicutt law to the black hole-bulge mass relation. A surprisingly tight correlation between galaxy star formation rates (SFR) and stellar masses (M*) was discovered in 2007, and models of galaxy formation and evolution can be constrained by studying the evolution of this SFR-M* correlation and its intrinsic scatter. At present, such investigations are weakened by the need to assume a simple parametric form for the star formation history, typically constant or exponentially declining. We propose to use our new dense basis method to reconstruct star-formation histories (SFHs) through SED fitting using multi-band photometry of >10,000 galaxies in the 3D-HST and CANDELS catalogs. Armed with these reconstructed SFHs, we will then: 1. Better measure the SFR-M* correlation (aka star-forming sequence) in several redshift bins at 0.5<z<3. Properly accounting for SFH uncertainties will generate improved measurements of intrinsic scatter. 2. Determine trajectories by which galaxies observed at a given epoch arrived at their current coordinates in SFR-M* space, allowing us to evolve the correlation back in time to previous redshift bins. The fraction of star-forming galaxies found to remain on the correlation between epochs will determine the level of stochasiticity driving star formation in galaxies. 3. Measure the SFR-M* correlation and intrinsic scatter as a function of the timescale over which SFR is averaged; this can be compared with analytical models and simulation predictions to determine the origin of the correlation. Galaxies 0 Regular false false false SUPPORTED