Abstract This article describes new methods for estimating survival distributions based on nonparametric curve estimators. One approach improves the estimation of long-term survival rates. Simulation studies using Weibull and lognormal data show that even in the case found to be least favorable, the new method has less than one-seventh the prediction error of all conventional life-table (LT) or Kaplan–Meier (KM) estimators, even when the LT and KM techniques are optimized for the purpose of long-term survival estimation. In addition to conventional survival applications, one can also estimate the probability of being disease-free at different ages and following different exposures to possibly harmful environmental contaminants. This approach is particularly useful in situations where the effects of a confounding, nuisance, or effect-modifying variable cannot be confidently modeled in a parametric form. The new techniques are based on a closed-form nonparametric maximum likelihood curve estimator expressed in terms of separate curve estimates obtained from samples of randomly censored and uncensored times to failure—that is, subsurvival populations.
Author(s): Freeman, William Robert | Advisor(s): Siana, Brian | Abstract: We present results from the MOSFIRE Deep Evolution Field (MOSDEF) survey on broad emission from the nebular emission lines \Ha, \NII, \OIII, \Hb, and \SII. After removing known AGN, the sample consists of 127 galaxies with $1.37 l z l 2.61$ and 84 galaxies with $2.95 l z l 3.80$. We study broad flux by decomposing the emission lines using narrow and broad components for individual galaxies and stacks. For the z $\sim2$ sample, the broad flux accounts for 20-50\% of the flux in nebular emission lines and in the z $\sim3.3$ sample the broad component comprises 30-60\% of the flux. We calculate \SII/\Ha, \NII/\Ha, and \OIII/\Hb line ratios for the narrow components and broad components and compare these to a single Gaussian fit. When placed on the BPT diagram (\OIII/\Hb vs. \NII/\Ha) the broad components are shifted towards the higher \OIII/\Hb and \NII/\Ha ratios. This is likely the results of partial ionization from shocks or low luminosity AGN. The narrow component line ratios are closer to local measurements than other studies at z $\sim2$ but are still slightly offset. Assuming the broad component is an outflow we estimate the mass loading factor ($\eta=$mass outflow rate/SFR) as a function of mass and find generally good agreement with previous studies. We find our galaxies are only compatible with simulations (which predict $\eta$ should decrease as a function of mass) only if a large fraction of the outflows are below 300 km s$^{-1}$ for galaxies below $10^{10}$ stellar mass. Outflows and shocks may significantly impact measurements from outflows from the \Ha line, change line ratios, and impact single line measurements such as calculating star formation rate from \Ha. We find that partially shocked flux from outflows could account for some of the offset seen in the BPT diagram from z $\sim0$ to z $\sim2$.
We present results on the dust attenuation curve of z ∼ 2 galaxies using early observations from the MOSFIRE Deep Evolution Field survey. Our sample consists of 224 star-forming galaxies with zspec = 1.36–2.59 and high signal-to-noise ratio measurements of Hα and Hβ obtained with Keck/MOSFIRE. We construct composite spectral energy distributions (SEDs) of galaxies in bins of Balmer decrement to measure the attenuation curve. We find a curve that is similar to the SMC extinction curve at λ ≳ 2500 Å. At shorter wavelengths, the shape is identical to that of the Calzetti et al. relation, but with a lower normalization. Hence, the new attenuation curve results in star formation rates (SFRs) that are lower, and stellar masses that are dex lower, than those obtained with the Calzetti relation. We find that the difference in the total attenuation of the ionized gas and stellar continuum correlates strongly with SFR, such that for dust-corrected SFRs ≳ 20 M⊙ yr−1, assuming a Chabrier initial mass function, the nebular emission lines suffer an increasing degree of obscuration relative to the continuum. A simple model that can account for these trends is one in which the UV through optical stellar continuum is dominated by a population of less-reddened stars, while the nebular line and bolometric luminosities become increasingly dominated by dustier stellar populations for galaxies with large SFRs, as a result of the increased dust enrichment that accompanies such galaxies. Consequently, UV- and SED-based SFRs may underestimate the total SFR at even modest levels of ≈20 M⊙ yr−1.
We present ionized gas kinematics for 681 galaxies at $z\sim 1.4-3.8$ from the MOSFIRE Deep Evolution Field survey, measured using models which account for random galaxy-slit misalignments together with structural parameters derived from CANDELS Hubble Space Telescope (HST) imaging. Kinematics and sizes are used to derive dynamical masses. Baryonic masses are estimated from stellar masses and inferred gas masses from dust-corrected star formation rates (SFRs) and the Kennicutt-Schmidt relation. We measure resolved rotation for 105 galaxies. For the remaining 576 galaxies we use models based on HST imaging structural parameters together with integrated velocity dispersions and baryonic masses to statistically constrain the median ratio of intrinsic ordered to disordered motion, $V/\sigma_{V,0}$. We find that $V/\sigma_{V,0}$ increases with increasing stellar mass and decreasing specific SFR (sSFR). These trends may reflect marginal disk stability, where systems with higher gas fractions have thicker disks. For galaxies with detected rotation we assess trends between their kinematics and mass, sSFR, and baryon surface density ($\Sigma_{\mathrm{bar},e}$). Intrinsic dispersion correlates most with $\Sigma_{\mathrm{bar},e}$ and velocity correlates most with mass. By comparing dynamical and baryonic masses, we find that galaxies at $z\sim 1.4-3.8$ are baryon dominated within their effective radii ($R_E$), with Mdyn/Mbaryon increasing over time. The inferred baryon fractions within $R_E$, $f_{\mathrm{bar}}$, decrease over time, even at fixed mass, size, or surface density. At fixed redshift, $f_{\mathrm{bar}}$ does not appear to vary with stellar mass but increases with decreasing $R_E$ and increasing $\Sigma_{\mathrm{bar},e}$. For galaxies at $z\geq2$, the median inferred baryon fractions generally exceed 100%. We discuss possible explanations and future avenues to resolve this tension.
We present results on the z~2.3 mass-metallicity relation (MZR) using early observations from the MOSFIRE Deep Evolution Field (MOSDEF) survey. We use an initial sample of 87 star-forming galaxies with spectroscopic coverage of H\beta, [OIII]\lambda 5007, H\alpha, and [NII]\lambda 6584 rest-frame optical emission lines, and estimate the gas-phase oxygen abundance based on the N2 and O3N2 strong-line indicators. We find a positive correlation between stellar mass and metallicity among individual z~2.3 galaxies using both the N2 and O3N2 indicators. We also measure the emission-line ratios and corresponding oxygen abundances for composite spectra in bins of stellar mass. Among composite spectra, we find a monotonic increase in metallicity with increasing stellar mass, offset ~0.15-0.3 dex below the local MZR. When the sample is divided at the median star-formation rate (SFR), we do not observe significant SFR dependence of the z~2.3 MZR among either individual galaxies or composite spectra. We furthermore find that z~2.3 galaxies have metallicities ~0.1 dex lower at a given stellar mass and SFR than is observed locally. This offset suggests that high-redshift galaxies do not fall on the local "fundamental metallicity relation" among stellar mass, metallicity, and SFR, and may provide evidence of a phase of galaxy growth in which the gas reservoir is built up due to inflow rates that are higher than star-formation and outflow rates. However, robust conclusions regarding the gas-phase oxygen abundances of high-redshift galaxies await a systematic reappraisal of the application of locally calibrated metallicity indicators at high redshift.
We present results on the dust attenuation curve of z~2 galaxies using early observations from the MOSFIRE Deep Evolution Field (MOSDEF) survey. Our sample consists of 224 star-forming galaxies with nebular spectroscopic redshifts in the range z= 1.36-2.59 and high S/N measurements of, or upper limits on, the H-alpha and H-beta emission lines obtained with Keck/MOSFIRE. We construct composite SEDs of galaxies in bins of specific SFR and Balmer optical depth in order to directly constrain the dust attenuation curve from the UV through near-IR for typical star-forming galaxies at high redshift. Our results imply an attenuation curve that is very similar to the SMC extinction curve at wavelengths redward of 2500 Angstroms. At shorter wavelengths, the shape of the curve is identical to that of the Calzetti relation, but with a lower normalization (R_V). Hence, the new attenuation curve results in SFRs that are ~20% lower, and log stellar masses that are 0.16 dex lower, than those obtained with the Calzetti attenuation curve. Moreover, we find that the difference in the reddening---and the total attenuation---of the ionized gas and stellar continuum correlates strongly with SFR, such that for dust-corrected SFRs larger than 20 Msun/yr assuming a Chabrier IMF, the nebular emission lines suffer an increasing degree of obscuration relative to the continuum. A simple model that can account for these trends is one in which the UV through optical stellar continuum is dominated by a population of less reddened stars, while the nebular line and bolometric luminosities become increasingly dominated by dustier stellar populations for galaxies with large SFRs, as a result of the increased dust enrichment that accompanies such galaxies. Consequently, UV- and SED-based SFRs may underestimate the total SFR at even modest levels of ~20 Msun/yr. [Abridged]
We present results on the excitation properties of z ∼ 2.3 galaxies using early observations from the MOSFIRE Deep Evolution Field (MOSDEF) Survey. With its coverage of the full suite of strong rest-frame optical emission lines, MOSDEF provides an unprecedented view of the rest-frame optical spectra of a representative sample of distant star-forming galaxies. We investigate the locations of z ∼ 2.3 MOSDEF galaxies in multiple emission-line diagnostic diagrams. These include the [O iii]λ5007/Hβ vs. [N ii]/Hα and [O iii]λ5007/Hβ vs. [S ii]λλ6717, 6731/Hα "BPT" diagrams, as well as the O32 vs. R23 excitation diagram. We recover the well-known offset in the star-forming sequence of high-redshift galaxies in the [O iii]λ5007/Hβ vs. [N ii]/Hα BPT diagram relative to Sloan Digital Sky Survey star-forming galaxies. However, the shift for our rest-frame optically selected sample is less significant than for rest-frame-UV selected and emission-line selected galaxies at z ∼ 2. Furthermore, we find that the offset is mass-dependent, only appearing within the low-mass half of the z ∼ 2.3 MOSDEF sample, where galaxies are shifted toward higher [N ii]/Hα at fixed [O iii]/Hβ. Within the [O iii]λ5007/Hβ vs. [S ii]/Hα and O32 vs. R23 diagrams, we find that z ∼ 2.3 galaxies are distributed like local ones, and therefore attribute the shift in the [O iii]λ5007/Hβ vs. [N ii]/Hα BPT diagram to elevated N/O abundance ratios among lower-mass () high-redshift galaxies. The variation in N/O ratios calls into question the use at high redshift of oxygen abundance indicators based on nitrogen lines, but the apparent invariance with redshift of the excitation sequence in the O32 vs. R23 diagram paves the way for using the combination of O32 and R23 as an unbiased metallicity indicator over a wide range in redshift. This indicator will allow for an accurate characterization of the shape and normalization of the mass–metallicity relationship over more than 10 Gyr.
We present results on the dust attenuation curve of z~2 galaxies using early observations from the MOSFIRE Deep Evolution Field (MOSDEF) survey. Our sample consists of 224 star-forming galaxies with nebular spectroscopic redshifts in the range z= 1.36-2.59 and high S/N measurements of, or upper limits on, the H-alpha and H-beta emission lines obtained with Keck/MOSFIRE. We construct composite SEDs of galaxies in bins of specific SFR and Balmer optical depth in order to directly constrain the dust attenuation curve from the UV through near-IR for typical star-forming galaxies at high redshift. Our results imply an attenuation curve that is very similar to the SMC extinction curve at wavelengths redward of 2500 Angstroms. At shorter wavelengths, the shape of the curve is identical to that of the Calzetti relation, but with a lower normalization (R_V). Hence, the new attenuation curve results in SFRs that are ~20% lower, and log stellar masses that are 0.16 dex lower, than those obtained with the Calzetti attenuation curve. Moreover, we find that the difference in the reddening---and the total attenuation---of the ionized gas and stellar continuum correlates strongly with SFR, such that for dust-corrected SFRs larger than 20 Msun/yr assuming a Chabrier IMF, the nebular emission lines suffer an increasing degree of obscuration relative to the continuum. A simple model that can account for these trends is one in which the UV through optical stellar continuum is dominated by a population of less reddened stars, while the nebular line and bolometric luminosities become increasingly dominated by dustier stellar populations for galaxies with large SFRs, as a result of the increased dust enrichment that accompanies such galaxies. Consequently, UV- and SED-based SFRs may underestimate the total SFR at even modest levels of ~20 Msun/yr. [Abridged]
