Yoav Rotman
@yoavrotman.bsky.social
PhD Candidate 🪐 @ ASU || UMD Astro '22 🐢 || studying exoplanet atmospheres and a big fan of TRAPPIST-1e || sometimes I play the trombone || he/him/הוא || find me on Twitter @Yoav_Rotman
This paper is now accepted in ApJ and the latest version is on arXiv as of last night, so go check it out! A huge thanks to the coauthors on here: @luiswel.bsky.social @astropeter.bsky.social @nixonmatthew.bsky.social
arxiv.org/abs/2503.21702
(13/13)
arxiv.org/abs/2503.21702
(13/13)
July 8, 2025 at 12:31 PM
This paper is now accepted in ApJ and the latest version is on arXiv as of last night, so go check it out! A huge thanks to the coauthors on here: @luiswel.bsky.social @astropeter.bsky.social @nixonmatthew.bsky.social
arxiv.org/abs/2503.21702
(13/13)
arxiv.org/abs/2503.21702
(13/13)
So does all this mean everything we've ever done is wrong? Absolutely not! GPs just present a new, statistically robust, more flexible modeling methodology for exoplanetary spectra that can provide insights into not just our data, but our models themselves!
(12/n)
(12/n)
July 8, 2025 at 12:31 PM
So does all this mean everything we've ever done is wrong? Absolutely not! GPs just present a new, statistically robust, more flexible modeling methodology for exoplanetary spectra that can provide insights into not just our data, but our models themselves!
(12/n)
(12/n)
In fact, we also see that the GP identifies a distinct global correlation in the NIRISS data, which is not unexpected from instrumental effects. At ~75 ppm, this is not a massive oversight, but something for future NIRISS enthusiasts to be aware of.
(11/n)
(11/n)
July 8, 2025 at 12:31 PM
In fact, we also see that the GP identifies a distinct global correlation in the NIRISS data, which is not unexpected from instrumental effects. At ~75 ppm, this is not a massive oversight, but something for future NIRISS enthusiasts to be aware of.
(11/n)
(11/n)
In fact, if a local kernel is placed, it falls on the CO2 feature at 2.7 microns, telling us that perhaps CO2 characterization with NIRISS is less robust than previously expected. That doesn't mean this is necessarily the case, but it's a good diagnostic already!
(10/n)
(10/n)
July 8, 2025 at 12:31 PM
In fact, if a local kernel is placed, it falls on the CO2 feature at 2.7 microns, telling us that perhaps CO2 characterization with NIRISS is less robust than previously expected. That doesn't mean this is necessarily the case, but it's a good diagnostic already!
(10/n)
(10/n)
Very possibly! We find that including GP kernels hint at wider H2O and CO2 abundance distributions than traditional retrievals. The latter even has a tail that contains the values from *both* previous estimates!
(9/n)
(9/n)
July 8, 2025 at 12:31 PM
Very possibly! We find that including GP kernels hint at wider H2O and CO2 abundance distributions than traditional retrievals. The latter even has a tail that contains the values from *both* previous estimates!
(9/n)
(9/n)
We then analyze the *real* NIRISS spectrum of WASP-96b! Previous analyses (Radica et al. 2023, Taylor et al. 2023) largely agree, but found discrepancies in their CO2 abundance. Can GPs help?
(8/n)
(8/n)
July 8, 2025 at 12:31 PM
We then analyze the *real* NIRISS spectrum of WASP-96b! Previous analyses (Radica et al. 2023, Taylor et al. 2023) largely agree, but found discrepancies in their CO2 abundance. Can GPs help?
(8/n)
(8/n)
Not only that, but the parameters of the GP ("hyperparameters") actually contain information! By analyzing the GP kernels, we can get an understanding for how much correlation there is in our data and *where* our model cannot fit it.
(6/n)
(6/n)
July 8, 2025 at 12:31 PM
Not only that, but the parameters of the GP ("hyperparameters") actually contain information! By analyzing the GP kernels, we can get an understanding for how much correlation there is in our data and *where* our model cannot fit it.
(6/n)
(6/n)
Yes! The traditional retrieval (black histograms) not only can't account for the injected feature, but it actually biases all our inferences. Different GP parameterizations (blue/pink/purple) give significantly less biased, more reliable answers.
(5/n)
(5/n)
July 8, 2025 at 12:31 PM
Yes! The traditional retrieval (black histograms) not only can't account for the injected feature, but it actually biases all our inferences. Different GP parameterizations (blue/pink/purple) give significantly less biased, more reliable answers.
(5/n)
(5/n)
We kick this off by creating a synthetic spectrum of WASP-96b, where we inject some underlying correlation and a "mystery" feature at 1.65 microns, that we *know* our model can't account for. Will a GP work better?
(4/n)
(4/n)
July 8, 2025 at 12:31 PM
We kick this off by creating a synthetic spectrum of WASP-96b, where we inject some underlying correlation and a "mystery" feature at 1.65 microns, that we *know* our model can't account for. Will a GP work better?
(4/n)
(4/n)
We use a GP that looks for a) global correlations throughout the spectrum that may come from instrumental effects and b) localized correlations that manifest where the model can't explain the data!
(3/n)
(3/n)
July 8, 2025 at 12:31 PM
We use a GP that looks for a) global correlations throughout the spectrum that may come from instrumental effects and b) localized correlations that manifest where the model can't explain the data!
(3/n)
(3/n)
Paper day! A 🧵
JWST transmission spectra are showing more and more signs of new exciting physics, which is awesome! But if our models don't take that into account, it can propagate as correlated noise and bias our inferences. What do we do?
arxiv.org/abs/2503.21702
(1/n)
JWST transmission spectra are showing more and more signs of new exciting physics, which is awesome! But if our models don't take that into account, it can propagate as correlated noise and bias our inferences. What do we do?
arxiv.org/abs/2503.21702
(1/n)
July 8, 2025 at 12:31 PM
Paper day! A 🧵
JWST transmission spectra are showing more and more signs of new exciting physics, which is awesome! But if our models don't take that into account, it can propagate as correlated noise and bias our inferences. What do we do?
arxiv.org/abs/2503.21702
(1/n)
JWST transmission spectra are showing more and more signs of new exciting physics, which is awesome! But if our models don't take that into account, it can propagate as correlated noise and bias our inferences. What do we do?
arxiv.org/abs/2503.21702
(1/n)
Conclusions/TLDR:
- Traditional retrievals fail in the presence of data unknowns, and we might not even realize it!
- Using a GP can help mitigate these biasing effects and give accurate answers!
- GPs can also tell you about the unknowns themselves!
(14/15)
- Traditional retrievals fail in the presence of data unknowns, and we might not even realize it!
- Using a GP can help mitigate these biasing effects and give accurate answers!
- GPs can also tell you about the unknowns themselves!
(14/15)
March 28, 2025 at 2:41 PM
Conclusions/TLDR:
- Traditional retrievals fail in the presence of data unknowns, and we might not even realize it!
- Using a GP can help mitigate these biasing effects and give accurate answers!
- GPs can also tell you about the unknowns themselves!
(14/15)
- Traditional retrievals fail in the presence of data unknowns, and we might not even realize it!
- Using a GP can help mitigate these biasing effects and give accurate answers!
- GPs can also tell you about the unknowns themselves!
(14/15)
Paper editing is uh..... not going well
February 20, 2025 at 10:28 PM
Paper editing is uh..... not going well
"Guys being dudes" isn't always about sports and beer and cars, sometimes it's about me and the homies spending 15 minutes discussing this free-body diagram of a slice of cheese in an academic paper
January 3, 2025 at 11:46 PM
"Guys being dudes" isn't always about sports and beer and cars, sometimes it's about me and the homies spending 15 minutes discussing this free-body diagram of a slice of cheese in an academic paper
Congratulations to one of the funniest bits I've done in a while on working to perfection
November 19, 2024 at 12:39 AM
Congratulations to one of the funniest bits I've done in a while on working to perfection
Hey look, I gave a talk at AAS!
(Shameless plug: if anyone is interested in chatting further or missed my talk about using covariance matrices in exoplanet retrievals, my email is yrotman@asu.edu - reach out!) 🪐🧪🔭
(Shameless plug: if anyone is interested in chatting further or missed my talk about using covariance matrices in exoplanet retrievals, my email is yrotman@asu.edu - reach out!) 🪐🧪🔭
January 11, 2024 at 9:36 PM
Hey look, I gave a talk at AAS!
(Shameless plug: if anyone is interested in chatting further or missed my talk about using covariance matrices in exoplanet retrievals, my email is yrotman@asu.edu - reach out!) 🪐🧪🔭
(Shameless plug: if anyone is interested in chatting further or missed my talk about using covariance matrices in exoplanet retrievals, my email is yrotman@asu.edu - reach out!) 🪐🧪🔭
My priority score on the ASU cluster was so low I couldn't get anything running so I made this instead
August 26, 2023 at 4:30 AM
My priority score on the ASU cluster was so low I couldn't get anything running so I made this instead
Hold on now when did python get an attitude
July 25, 2023 at 7:14 PM
Hold on now when did python get an attitude
Ah yes, this is good rebranding technique
July 24, 2023 at 2:49 PM
Ah yes, this is good rebranding technique
Finally solved an issue in my code I've been trying to figure out for a while
July 12, 2023 at 12:18 AM
Finally solved an issue in my code I've been trying to figure out for a while
Captain's log, day #(?) of hunting this code error:
July 3, 2023 at 9:55 PM
Captain's log, day #(?) of hunting this code error:
Unfortunately, I too have fallen victim to the Grad Student to Rock Climbing pipeline
July 2, 2023 at 7:21 PM
Unfortunately, I too have fallen victim to the Grad Student to Rock Climbing pipeline