Might be basic stuff for many, but it took me some effort to tune the hyperparameters to get the NN loss down by 2 orders of magnitude. Can you guess which hyperparameter I was mainly adjusting?

I always wondered how I would teach a class on a topic I don't know, if I ever manage to be a Prof. Well now I know the trade secret. Kudos to all the Professors who do this everyothe day and come to the class like they know everything. Just wow!

#MatterIssue31: @toniobuonassisi & collabs show a framework for the general #InverseDesign of inorganic crystals! #ML First authors: Dannyzekunren & S. Isaac P. Tian w/Xiaonan Wang Hip Lab Shijing Sun @felipeoviedop Harry Liang & more See it here: bit.ly/3KGme5D

When you have a movie buff friend back in Germany, this is what they are want you to take back with you from the US. 74 individual dvds. I don’t even know how many movies are on them.

New simulation paper on the beneficial or detrimental effects of doping on photovoltaic performance in halide perovskite solar cells. Basita Das Forschungszentrum Jülich | @[email protected] FZ Jülich ENERGIE UND KLIMA / Energy & Climate onlinelibrary.wiley.com/doi/full/10.10…

A new Consensus Report from the National Academies National Academies: Automated Research Workflows Are Speeding Pace of Scientific Discovery. It has a few pages devoted to workflows powering simulation-based inference & reinterpretation (RECAST) in physics nationalacademies.org/news/2022/05/a…

Happy month of Pride ! 🏳️‍🌈🏳️‍⚧️🌈

Pleased to announce our last speakers before the summer break!☀️😎 We will have one talk on #OPV by Mohammed Azzouzi from Jenny Nelson Group and one about #perovskite and #ML by Basita Das from Thomas Kirchartz's group. More info at: ngse.info/phd-postdoc-se…

First PhD party after Covid years. Had a great evening! FZ Jülich ENERGIE UND KLIMA / Energy & Climate Forschungszentrum Jülich | @[email protected] #PhD #fun #iek5

Tomorrow I talk about using Bayesian Inference for determining what’s limiting the performance of your devices. I will share insights from my learning curve over last one year and how to get started with it.

Great thread. Imagine an integrated system of (1) algos predicting materials with target properties, (2) a robot creating the predicted materials, (3) characterization tools measuring the ground truth and feeding the data back to (1) for better prediction. All in a single click!

View from my flight to India. It was an amazing sight. Next time when I would need a color scale, this is the color palette I will use.

Guten Morgen aus Jülich. 🍁🌫

An AI model that has read 48M academic paper…it’s mind boggling to even think what new possibilities it might open. From knowledge transfers from one field to another to finding new interdisciplinary fields. Question is how do we access what it has learnt and interpret it!

Some of our most-cited studies from 2022 are electric — like this one on a #solar cell’s capacity to hold #electricity. The analysis finds that resistive contact layers in #perovskite solar cells strongly affect the cells’ total capacitance. Read more: go.aps.org/3PGKdEP

This is an awesome contibution Sterling G. Baird

New paper by Basita Das and myself on Transforming characterization data into information in the case of perovskite solar cells doi.org/10.1088/2515-7… via JPhys Energy Forschungszentrum Jülich | @[email protected] FZ Jülich ENERGIE UND KLIMA / Energy & Climate

🔺The #DEPERO Online Conference on 'Device Physics Characterization and Interpretation in Perovskite and Organic Materials' will feature an astounding line-up of Invited Speakers! 📆3rd-5th October 2023 👉Check it out and submit your oral abstract: nanoge.org/DEPERO/home

Nobel Prize winner Katalin Karikó was 'demoted 4 times' at her old job. How she persisted: 'You have to focus on what's next' cnb.cx/3tqKkxm