1908.00543
Deep learning dark matter map reconstructions from DES SV weak lensing data
Jeffrey, Lanusse, Lahav, Starck
We present the first reconstruction of dark matter maps from weak lensing observational data using deep learning. We train a convolution neural network (CNN) with a Unet based architecture on over $3.6\times10^5$ simulated data realisations with non-Gaussian shape noise and with cosmological parameters varying over a broad prior distribution. We interpret our newly created DES SV map as an approximation of the posterior mean $P(\kappa | \gamma)$ of the convergence given observed shear. Our DeepMass method is substantially more accurate than existing mass-mapping methods. With a validation set of 8000 simulated DES SV data realisations, compared to Wiener filtering with a fixed power spectrum, the DeepMass method improved the mean-square-error (MSE) by 11 per cent. With N-body simulated MICE mock data, we show that Wiener filtering with the optimal known power spectrum still gives a worse MSE than our generalised method with no input cosmological parameters; we show that the improvement is driven by the non-linear structures in the convergence. With higher galaxy density in future weak lensing data unveiling more non-linear scales, it is likely that deep learning will be a leading approach for mass mapping with Euclid and LSST.
Deep learning dark matter map reconstructions from DES SV weak lensing data
Jeffrey, Lanusse, Lahav, Starck
We present the first reconstruction of dark matter maps from weak lensing observational data using deep learning. We train a convolution neural network (CNN) with a Unet based architecture on over $3.6\times10^5$ simulated data realisations with non-Gaussian shape noise and with cosmological parameters varying over a broad prior distribution. We interpret our newly created DES SV map as an approximation of the posterior mean $P(\kappa | \gamma)$ of the convergence given observed shear. Our DeepMass method is substantially more accurate than existing mass-mapping methods. With a validation set of 8000 simulated DES SV data realisations, compared to Wiener filtering with a fixed power spectrum, the DeepMass method improved the mean-square-error (MSE) by 11 per cent. With N-body simulated MICE mock data, we show that Wiener filtering with the optimal known power spectrum still gives a worse MSE than our generalised method with no input cosmological parameters; we show that the improvement is driven by the non-linear structures in the convergence. With higher galaxy density in future weak lensing data unveiling more non-linear scales, it is likely that deep learning will be a leading approach for mass mapping with Euclid and LSST.
1908.00589
Astro2020 APC White Paper: Finding s and recommendations from the AAS Committee on the status of women in astronomy: awards eliminating harassment in Astronomy
Zellner, et al
The Committee on the Status of Women in Astronomy (CSWA) is calling on federal science funding agencies, in their role as the largest sources of funding for astronomy in the United States, to take actions that will end harassment, particularly sexual harassment, in astronomical workplaces. Funding agencies can and should lead the charge to end harassment in astronomy by the 2030 Astrophysics Decadal Survey. Anecdotal and quantitative evidence, gathered both by the CSWA and other groups, shows that harassment is prevalent and damaging for women and minority astronomers and those in related fields. Actions recommended herein will increase the rate of reporting of harassment to agencies and improve their ability to investigate and take action against harassers. We also recommend that agencies participate in harassment prevention by creating and implementing the best anti-harassment education possible. Key recommendations are: - Federal agencies should improve their ethics policies by making harassment a form of scientific misconduct. - Federal agencies should mandate that institutions report to them when a funded Principal Investigator (PI) or co-Principal Investigator (co-PI) is found to be a perpetrator of harassment. - Federal funding agencies should provide online guides to help scientists identify harassment and connect them to the right resources for making confidential or official reports. - Federal agencies should create and ensure the implementation of anti-harassment trainings by making them a requirement of receiving grant funding.
1908.00597
Astro2020 APC White Paper: Findings and recommendations from the AAS Committee on the status of women in Astronomy: advancing the career development of women in Astronomy
Zellner, et al
The Committee on the Status of Women in Astronomy (CSWA) is calling on federal science funding agencies, in their role as the largest sources of funding for astronomy in the United States, to take actions that will advance career development and improve workplace conditions for women and minorities in astronomy. Funding agencies can and should lead the charge to produce excellent diversity and inclusion outcomes in our field by the 2030 Astrophysics Decadal Survey. Anecdotal and quantitative evidence, gathered both by the CSWA and other groups, shows that many structural barriers to success remain in our community. We acknowledge the success of programs like NSF-INCLUDES and NSF-ADVANCE and endorse their continued work. We also recommend further action to remove barriers to success for women and minority astronomers. Key recommendations are: - Federal agencies should push academic institutions to reconsider their support systems for scientists by requiring a high standard of pay and benefits. - Federal agencies should fund research and programs that explore and implement strategies for improving the experiences of scientists. - Federal agencies should require Principal Investigators (PIs) to provide plans for creating inclusive work environments and mentoring young scientists.
1908.00771
MAGPHYS+photo-z: constraining the physical properties of galaxies with unknown redshifts
Battisti, et al
We present an enhanced version of the multiwavelength spectral modeling code MAGPHYS that allows the estimation of galaxy photometric redshift and physical properties (e.g., stellar mass, star formation rate, dust attenuation) simultaneously, together with robust characterization of their uncertainties. The self-consistent modeling over ultraviolet to radio wavelengths in MAGPHYS+photo-z is unique compared to standard photometric redshift codes. The broader wavelength consideration is particularly useful for breaking certain degeneracies in color vs. redshift for dusty galaxies with limited observer-frame ultraviolet and optical data (or upper limits). We demonstrate the success of the code in estimating redshifts and physical properties for over 4,000 infrared-detected galaxies at 0.4<z<6.0 in the COSMOS field with robust spectroscopic redshifts. We achieve high photo-z precision ($\sigma_{\Delta z/(1+z_{spec})}\lesssim0.04$), high accuracy (i.e., minimal offset biases; median$(\Delta z/(1+z_{spec}))\lesssim0.02$), and low catastrophic failure rates ($\eta\simeq4\%$) over all redshifts. Interestingly, we find that a weak 2175A absorption feature in the attenuation curve models is required to remove a subtle systematic photo-z offset ($z_{phot}-z_{spec}\simeq-0.03$) that occurs when this feature is not included. As expected, the accuracy of derived physical properties in MAGPHYS+photo-z decreases strongly as redshift uncertainty increases. The all-in-one treatment of uncertainties afforded with this code is beneficial for accurately interpreting physical properties of galaxies in large photometric datasets. Finally, we emphasize that MAGPHYS+photo-z is not intended to replace existing photo-z codes, but rather offer flexibility to robustly interpret physical properties when spectroscopic redshifts are unavailable. The MAGPHYS+photo-z code is publicly available online.
No comments:
Post a Comment