Jake Sullivan (@SodeAndy) header
Jake Sullivan (@SodeAndy) profile picture

https://xitter.gq/t.co/WiGwNCIrWw

Joined July 2019
That notebook idea finally became my Halloween 3D creation! Made with @V2FUN_official for Halloween 8K Universe. Vote for my entry via the link in bio! #v2fun #3DModelGenerator #Halloween8KUniverse
681
Cute, not creepy: my little ghost Vtoo is now in 3D! Made with @V2FUN_official for Halloween 8K Universe. Vote for my entry via the link in bio! #v2fun #3DModelGenerator #Halloween8KUniverse
93
If you find BO for materials discovery interesting, take a look at our GitHub: github.com/AndySAnker/ScattB… Here, you can find examples of how to use the benchmark and follow the scoreboard.
1
159
Based on a set of synthesis parameters, ScattBO ‘synthesises’ a structure, calculates the scattering pattern and compares this to the scattering pattern of the target structure.
1
90
Thanks to the organisers of this amazing Bayesian optimisation hackathon. I learned a lot! ☺️ Our team developed a simulation of a self-driving laboratory for materials discovery serving to benchmark the efficiency of various BO algorithms 🤖 @gabe__graves @AppiahAmpofoSa1
2
1
17
1,241
Thrilled to announce that for the coming two years, I'll be located at Oxford, collaborating with the @VLDGroupOx, thanks to @novonordiskfond's Postdoctoral Programme. Today, I spent some time in the Radcliffe Camera, an iconic place I've longed to explore. #PostdocAdventures
2
1
37
1,616
Had an amazing first day at @DTUEnergy! Received a warm welcome with a pile of chocolate, which I couldn't resist before picture time! 🍫✨ Chocolates brought back by Dall-E 🪄 Looking forward to an exciting journey ahead with @TVegge, #CAPeX. #NewChapter #SweetStart #CAPeXCap
1
39
3,190
Friday I had an incredible day defending my PhD. I am extremely grateful to everyone who supported me in person or online. Especially the committee @taylordsparks @GemmaCSolomon and Andrew Goodwin 🙏 What an amazing time I have had in @jensengroup_cph with @kirsten_mj 🕺💃
4
2
51
2,064
1) Handling data with contributions from multiple chemical components. 2) Handling data arising from structures not present in the training database. 3) Accounting for experimental data that contain signals not included in the simulated data. 7/n
1
1
199
We observe three main applications of supervised ML for the analysis of scattering and spectroscopy data: 1) Identifying a structure model from a dataset. 2) Predicting data from a structure model. 3) Bypassing the structure refinement step to directly obtain structural info🦸
1
1
140
Conventional structure refinement can provide invaluable structural information on e.g., lattice parameters, and atomic positions. 👊 However, extensive database and literature searches are often needed to identify the structure model to use in structure refinement. 📚😟 3/n
1
155
Our Perspective on 'Machine Learning for Analysis of Experimental Scattering and Spectroscopy Data in Materials Chemistry' is now on @ChemRxiv 👇 doi.org/10.26434/chemrxiv-20… @kirsten_mj @raghavian @keeeto2000 A thread 🧵1/8
1
8
2
39
12,960
Finally, we show that POMFinder can predict the structural model from both X-ray, neutron and electron PDFs and Small-Angle X-ray and neutron scattering and combinations of these. We show that combining the dataset improves the accuracy.
94
We use SHAP analysis to understand why POMFinder predicts as it does. Here it seems like its predictions correlate with the scattering power of the elements.
1
96