We present numerical simulations of the growth and saturation of the Kelvin-Helmholtz instability in a compressible fluid layer with and without a weak magnetic field. In the absence of a magnetic field, the instability generates a single eddy which flattens the velocity profile, stabilizing it against further perturbations. Adding a weak magnetic field - weak in the sense that it has almost no effect on the linear instability - leads to a complex flow morphology driven by MHD forces and to enhanced broadening of the layer, due to Maxwell stresses. We corroborate earlier studies which showed that magnetic fields destroy the large scale eddy structure through periodic cycles of windup and resistive decay, but we show that the rate of decay decreases with decreasing plasma resistivity, at least within the range of resistivity accessible to our simulations. Magnetization increases the efficiency of momentum transport, and the transport increases with decreasing resistivity.

We present a study of radiative transfer in dusty, clumpy star-forming regions. A series of self-consistent, 3-D, continuum radiative transfer models are constructed for a grid of models parameterized by central luminosity, filling factor, clump radius, and face-averaged optical depth. The temperature distribution within the clouds is studied as a function of this parameterization. Among our results, we find that: (a) the effective optical depth is smaller in clumpy regions than in equivalent homogeneous regions; (b) penetration of radiation is drive by the fraction of open sky (FOS) -- which measures the fraction of solid angle which is devoid of clumps; (c) FOS increases as clump radius increases and filling factor decreases; (d) for FOS > 0.6-0.8 the sky is sufficiently open that the temperature is relatively insensitive to FOS; (e) the physical process by which radiation penetrates is streaming between clumps; (f) filling factor dominates the temperature distribution for large optical depths, and at small clump radii for small optical depths; (g) at lower optical depths, the temperature distribution is most sensitive to filling factors of 1-10 per cent, in accordance with many observations; (h) direct shadowing can be important approximately one clump radius behind a clump.

A series of self-consistent, three-dimensional continuum radiative transfer models are constructed of the pre-protostellar core L1544, with the results compared with existing SCUBA and ISO data. The source is well-fit by a prolate spheroid, having an ellipsoidal power-law density distribution of index m ~ 2 (1.75 < m < 2.25) in to at least r ~ 1600AU. For r<1600 AU, the data are consistent with either an extension of the power law to smaller radii, or a flattened (Bonner-Ebert like) density distribtion. We can further constrain the optical depth along the short axis at 1300um to be ~ 5e-3, the central luminosity to be L < 1e-3 solar luminosities, the long axis diameter D ~ 0.1 pc, the axis ratio to be q ~ 2, and the external ISRF to be similar to that defined by Mathis, Mezger, & Panagia (1983) to within 50 per cent. The outer diameter and axis ratio may each be somewhat larger due to potential on-source chopping in the observations, and the projection of the long axis onto the plane of the sky. While these results are similar to those inferred directly from observations or spherical modeling due to the source transparency at submillimeter wavelengths, we infer a smaller size, lower mass, and higher optical depth / column density, exposed to a stronger external radiation field than previously assumed. Finally, we find that both the spectral energy distribution (SED) and surface brightness distribution are necessary to constrain the source properties in this way.

We present an evaluative study of some current methods utilized in the analysis of infrared (IR) observatinons of star-forming regions. A series of self-consistent radiative transfer models are constructed, with the outputs analysed using these methods to infer source properties such as dust temperature, mass, opacity function, and density distribution. Any discrepancies between the inferred and model quantities can be attributed to the analysis methods. The range of validity of most methods is smaller than expected, due to two effects: (1) limited applicability of the Rayleigh-Jeans limit except to very long wavelengths, and (2) significant errors in the isothermal approximation, even when temperature variations are less than 2 K over 90% of the region. Still, an accurate mean dust temperature can be found using a modified Wien's law. This temperature can yield dust masses to within 10-25% -- much better than masses inferred from the integrated luminosity. Using long wavelengths (greater than 1000 - 2000 microns), the opacity index can be determined from the far-IR spectrum to within 20%. Fitting the spectrum yields better results. The desnity distribution can be somewhat constrained by fitting the surface brightness, for well-resolved sources. Better results are found by fitting the flux spectrum with detailed models.