In this work, we examine spectral energy distribution (SED) fitting techniques commonly used to measure the dust temperatures of galaxies. Using toy models, we show that current far infrared (FIR) fitting techniques do not accurately recover the temperature of the dust within a galaxy. Instead, current fitting techniques recover temperatures which are higher than the actual mean of the dust temperature due to fact that hotter blackbodies are brighter than cooler blackbodies at all wavelengths. We demonstrate a method for FIR SED fitting which uses Markov Chain Monte Carlo (MCMC) methods to recover a distribution of dust temperatures given a galaxy's SED. We show that the mean dust temperature of the distribution recovered using MCMC methods on generated data is significantly different from the temperature recovered from conventional curve fitting techniques. We then demonstrate the power of our our fitting method on synthetic SEDs generated using cosmological galaxy formation simulations coupled with 3D dust radiative transfer models. Using this simulated data, we also establish a temperature distribution shape to use for this MCMC method based on the temperature of the CMB at a given redshift. Code to implement MCMC curve fitting will be made publicly available.
On August 21, 2017, millions of people across North America turned their (protected) eyes to the Sun to witness a total solar eclipse. At the same time in Jackson Hole, Wyoming, LightSound, an Arduino device developed at Harvard University, streamed the event online for the blind and visually impaired around the world. In a process called "sonification," the device uses a light sensor that takes the measured intensity of light (in the IR band, visual band, or both) and converts it to a pitch so that the listener can experience the real-time darkening during an eclipse. In preparation for the 2019 and 2020 eclipses in Chile and Argentina, LightSound has been redesigned with a MIDI synthesizer to allow the user to choose a variety of sound outputs and to offer a more rugged and telescope-adaptable interface. The documentation and code for LightSound, which costs about $70 to build, will be freely available online so that others may build their own (and modify the code, if they wish).
Dark energy, a force that counteracts gravity and accelerates the expansion of the universe on the grandest of scales, continues to confound the astrophysicists. In order to constrain its behavior in the universe, cosmologists are undertaking large surveys to understand its effects. The Dark Energy Survey (DES) is surveying an eighth of the sky for galaxies with redshift 0.2 to 1.5 to observe weak lensing effects and analyze cosmic shear. DES measures galaxy photometry in the four distinct g r i z bands, and hope to constrain their redshift from these photometric measurements. The error associated with these photo-z's contribute significantly to the DES cosmological parameters' error from weak lensing. Invoking Self Organized Maps (SOM), we establish a novel method of estimating the N(z) associated with the tomographic bins created to organize the SOM cells which capture galaxies with discreet color properties, based on the photometric measurements. By using a better constrained photometric survey for a smaller subset of the sky, we can leverage the few known redshifts to inform our color measurements across the SOM, and estimate N(z) for the full survey sample. This method has proven to constrain the mean difference between estimated and true N(z) for a simulated survey within 4%.
In this project, we analyze the spatial structure of HII regions in the interstellar medium of galaxy formation simulations from the FIRE (Feedback in Realistic Environments) project. We focus on a simulation of a Milky Way-like galaxy at high resolution evolved with the time-dependent chemistry solver CHIMES. We compare the properties of HII regions around young, luminous stars to the spherical HII regions predicted by the Strömgren model. To do this, we used Firefly, an interactive web-based visualization application we developed for exploring particle-based data. In addition to zooming in, rotating around, and visually manipulating the particles, we have also implemented in Firefly the ability to apply a unique colormap to each particle type based on a certain attribute. Using this new feature reveals sharp boundaries between ionized and neutral hydrogen regions, similar to what is predicted by the Strömgren approximation. A more quantitative analysis, however, reveals that the Strömgren radius is systematically smaller than the true "ionized radius" measured from the simulation data. We find that this is because the Strömgren model assumes isolated stars, while in the more realistic simulations, young stars are clustered. The assumption of isolated stars underestimates the true ionizing luminosity of an HII region, causing the Strömgren radius to be generally smaller than the true ionized radius. In the future, we plan to use Firefly to further analyze how molecular clouds (traced by H2, CO, etc.) relate to HII regions in the simulations.