Open software for Astronomical Data Analysis

@ NASA Goddard

Transcript

1. OPEN SOFTWARE FOR ASTRONOMICAL DATA ANALYSIS by Dan Foreman-Mackey
3. open software for astrophysics 0
4. credit: Adrian Price-Whelan / / data: SAO/NASA ADS
6. many fundamental software packages have a shockingly small number of maintainers.
7. 7 credit: Adrian Price-Whelan
8. * astronomical software can be very high impact * we should think about career trajectories & mechanisms for supporting this work
10. case study: gaussian processes 1
11. °0.6 °0.3 0.0 0.3 0.6 raw [ppt] 0 5 10 15 20 25 time [days] °0.30 °0.15 0.00 de-trended [ppt] N = 1000 reference: DFM+ (2017)
12. °0.6 °0.3 0.0 0.3 0.6 raw [ppt] 0 5 10 15 20 25 time [days] °0.30 °0.15 0.00 de-trended [ppt] N = 1000 reference: DFM+ (2017)
13. reference: Aigrain & DFM (2022)
14. reference: Aigrain & DFM (2022)
15. reference: Aigrain & DFM (2022) ignoring correlated noise accounting for correlated noise
16. reference: Aigrain & DFM (2022)
17. a Gaussian Process is a drop - in replacement for chi - squared
18. more details: Aigrain & Foreman-Mackey (2023) arXiv:2209.08940
20. 7 [1] model building [2] computational cost
21. reference: Luger, DFM, Hedges (2021)
22. [2] computational cost
23. 7 [1] bigger/better computers [2] exploit matrix structure [3] approximate linear algebra [4] etc.
24. 1 3 2
27. 1 3 2
28. °0.6 °0.3 0.0 0.3 0.6 raw [ppt] 0 5 10 15 20 25 time [days] °0.30 °0.15 0.00 de-trended [ppt] N = 1000 reference: DFM+ (2017)
29. reference: Gordon, Agol, DFM (2020) / tinygp.readthedocs.io
30. * a Gaussian Process is a drop - in replacement for chi squared * model building & computational cost are (solvable!) challenges * you should check out tinygp!
31. case study: probabilistic inference 2
32. have: physics = > data
33. want: data = > physics
34. 7 [1] physical models [2] legacy code
36. number of parameters patience required a few tenish not outrageously many reference: DFM (priv. comm.)
37. number of parameters patience required emcee a few tenish not outrageously many reference: DFM (priv. comm.)
38. number of parameters patience required emcee a few tenish not outrageously many how things should be reference: DFM (priv. comm.)
43. 3.0 3.5 4.0 4.5 5.0 Wavelength [micron] 2.05 2.10 2.15 2.20 2.25 2.30 Transit Depth [%] Alderson et al. 2023 Joint Fit (N = 50) reference: Soichiro Hattori, Ruth Angus, DFM, . . . (in prep) WASP-39b / NIRSpec
44. reference: Soichiro Hattori, Ruth Angus, DFM, . . . (in prep) showing 23 of the 404 parameters (8 per channel + 4 shared)
45. how?
46. d(physics = > data) / dphysics
47. automatic differentiation aka “backpropagation”
49. 7 [1] physical models [2] legacy code
50. 7 [1] domain - specif i c libraries [2] emulation
52. * gradient - based inference using autodiff can improve eff i ciency * there are practical challenges with these methods in astro * of interest: domain - specif i c libraries & emulation
53. aside: JAX 3
55. import numpy as np def linear_least_squares(x, y) : A = np.vander(x, 2) return np.linalg.lstsq(A, y)[0]
56. import jax.numpy as jnp def linear_least_squares(x, y) : A = jnp.vander(x, 2) return jnp.linalg.lstsq(A, y)[0]
58. open research practices 4
66. open software is foundational to astrophysics research there are opportunities at the interface of astro & applied f i elds there are ways you can participate & benef i t right away
67. 7 I want to chat about… [1] your data analysis problems [2] building astronomical software [3] writing documentation & tutorials
68. get in touch! dfm.io github.com/dfm