In case you're working on NSF COA
27.10.2025 12:49
👍 4
🔁 2
💬 0
📌 0
In case you're working on NSF COA
Really thrilled to share this new publication on the AMORE 2.0 algorithm 💗
This goal of this EPA-funded project was to develop an automated way to reduce the complexity of atmospheric chemical mechanisms using the principles of graph theory
academic.oup.com/pnasnexus/ar...
As an aside, I wouldn't necessarily use the ESS Open Archive for publishing preprints in the future. It took over two weeks for the preprint to show up on the website after we submitted i.t
We have a new preprint out! essopenarchive.org/doi/full/10....
In it, Xiaokai Yang, Lin Guo and I explore a method for making atmospheric chemistry surrogate models which are guaranteed to not have exploding errors. This is a step toward generally useful machine-learned models of the atmosphere.
Definitely relieved to learn that my work is not pursuant to their priorities.
🔔 New Master's internship opportunity 🔔
Come work with me this summer (from the comfort of your own home!) on global urban modeling at Pacific Northwest National Laboratory! Details & apply: careers.pnnl.gov/jobs/10599?l...
Due Monday, April 14!
More info in 🧵
[1/4]
This is not okay, Chuck.
I think the issue is that now is not the time to be debating this. The goal of this policy is not to increase the efficiency of universities, and engaging with the policy as if it were a good-faith effort is not helpful.
Image shows a drop down menu and search results page. It also repeats the text: Wayback Machine Collection Full Text Search Visit web.archive.org Scroll to 🔎Collection Search section Select "End Of Term (US Gov 2024)" from the menu Type in your query term and click search
How to Access the EOT Web Archives
Visit web.archive.org
Scroll to 🔎Collection Search section
Select "End Of Term (US Gov 2024)" from the menu
Type in your query term and click search
Or download bulk WARC files from eotarchive.org/data/
#EOT2024 #EOTArchive
The capitol hovering menacingly in the background is a nice touch
The paper is led by student Lin Guo. We have more impressive results in the pipeline, more coming soon! (In the academic sense of "soon".)
In this paper we use a machine learning method called e-sindy to simplify a chemical mechanism and quantify the uncertainty introduced by the simplification at the same time.
Atmospheric chemistry is computationally expensive to model, so we typically use simplified models when we want to study air quality across large amounts of time or space. The simplified models introduce uncertainty, but we typically aren't able to quantify that uncertainty.
We have a new paper just published about using machine learning to estimate uncertainty in atmospheric chemistry modeling! dx.doi.org/10.1029/2024...
1. Verify
2. ????
3. Profit!
Research scientist position on AI weather forecasting at UChicago's AI for Climate (AICE) initiative to work with me, Amir Jina, and Michael Kremer. Will work across our new Human-Centered Weather Forecasting Initiative + @dsi-uchicago.bsky.social + Climate Institute
In retrospect, I should have started preparing this material a long time ago...
🙋♂️
I did a bluesky thing and made a "list" of atmospheric chemists! Share with your friend and let me know if you want to be added! bsky.app/profile/did:...
We have a new preprint led by student Lin Guo! It is about uncertainty quantification in atmospheric chemistry modeling: arxiv.org/abs/2407.09757
I was quoted in this article! www.npr.org/2023/10/17/1...
We have a new preprint out today! arxiv.org/abs/2309.11035 We find that replacing the advection operator with a machine-learned surrogate may help the model run faster and/or be more accurate in certain situations.
