New preprint. Machine learning can predict chemical composition from simple photos of dried drop stains, but does it still work for complex mixtures like synthetic urine? We collected 70,000 images and tried it ...
doi.org/10.26434/che...
APL ML Abstract screenshot: Image-based identification of solutes from dried drop stains offers a low-cost, scalable alternative to traditional chemical analysis, enabled by artificial intelligence and annotated image libraries. While recent advances in robotic droplet imaging have facilitated the collection of such datasets, experimental acquisition remains a key bottleneck. Here, we explore whether synthetic image generation can supplement or replace real data in the context of salt classification. Using Stable Diffusion with low-rank adaptation, we generate 133 000 synthetic images from as few as N = 1–128 experimental images per salt across seven common inorganic solutes.
Our recent APL Machine Learning abstract as a slow K-pop song 🎶 🇰🇷🎵 Enjoy!!
suno.com/s/mUwia4C6uu...
Scientific Reports might benefit from some LLM reviewers ... no current LLM would have approved this paper.
I see that makes sense. Thanks!
Also, if NIH believes in the merit of their rankings, shouldn't the curve be more sigmoidal? Looks like review percentiles only explain about half of the funding decision variance.
how does this line up with a steady NIH budget? Applications increased? Bigger grants?
... ergo, older scientists should get longer grants ... everything else is cruel :)
you can see the demon doing a lot of information processing 😅 (and work) ... that's where the "paradox" is resolved
New preprint posted on ChemRxiv: "Charge-Selective Adsorption of Organic Molecules in Microfluidic Mg(OH)2 Precipitate Membranes" chemrxiv.org/doi/full/10....
It's a bit perplexing, but I agree. I believe the biggest trap is giving out lecture notes. Coming to class & taking handwritten notes is the 1st step toward learning the material. Downloading the pdf's for possible future consumption is not, but it does feel good (equiv. to us downloading papers).
New paper on arXiv arxiv.org/abs/2602.07183. We show that the aging of precipitate membranes and chemical gardens controls pattern formation by favoring growth at freshly formed regions. This time dependence follows stretched exponentials that transition to a scale-free power law.
great idea!
Marangoni convection meets chemical waves in the BZ reaction, where small pillars cause flower-like patterns. New paper from a collaboration with Azam Gholami, Detlef Lohse, and others.
advanced.onlinelibrary.wiley.com/doi/full/10....
PDB 2NZ2
Today in our #RareDiseaseThursday spotlight: Argininosuccinate synthetase deficiency (ASSD), a urea cycle disorder caused by pathogenic variants in ASS1, impairing the conversion of citrulline+aspartate to argininosuccinate. This leads to long-term neurological damage.
Some thoughts on grading ... if you want to see how ThermoTutor is doing it, check out thermo-tutor.qwyzzee.com
#HigherEd #Chemistry #STEM #college #thermodynamics #pchem
My photo is in Discover magazine! Looks great 😀
Congrats Wen! 🎉🎉➡️FSU chemist earns $2.1 million NIH grant to investigate molecular causes of disease: news.fsu.edu/news/science...
Over the past summer, I built an AI-powered learning platform for thermodynamics students. It’s called ThermoTutor. Here’s a short manifesto on what it does and why I think it’s useful:
thermo-tutor.qwyzzee.com
And Einstein's dissertation? Basically physical chemistry! 😊
“A New Determination of Molecular Dimensions” estimated molecular radii and Avogadro’s number from sugar diffusion and viscosity. From macroscopic measurements to molecular insight ... classic pchem move.
Quick introduction to ThermoTutor. Get personalized thermodynamics tutoring with instant feedback and follow-up conversations. This 3-minute demo shows how ThermoTutor works, from problem selection to grading to asking follow-up questions. #chemistry #studyhelp #pchem
PDB: 8GS8
Succinate dehydrogenase (SDH) uniquely powers both the TCA cycle & electron transport chain. SDH mutations cause SDH deficiency, disrupting energy metabolism & leading to developmental delays. #RareDiseaseThursday
KARS1 encodes lysyl-tRNA synthetase. Biallelic KARS mutations cause KARS syndrome, a severe disorder that affects multiple organs, particularly the nervous system. #RareDiseaseThursday
Our new paper describes a synthetic "cell" that moves over superhydrophobic surfaces. It consists of an H2O2 drop and a chemical garden tube that spins like a propeller or orbits around a large internal O2 bubble; bursting causes repeating cm-scale motion. @softmatter.rsc.org doi.org/10.1039/D5SM...
great photo!
The abstract submission deadline for our symposium is TODAY! Submit your abstract at sermacs-swrm2025.abstractcentral.com/submission before it’s too late; I hope to see you all in sunny Orlando this October!
Congratulations to Julia and Adwaith for completing the 2025 Young Scholar Program and research symposium. It was a joy having you both in lab and we are very proud of you!
I always enjoy GRCs. And if you criticize GRCs for being mainly in the US ... well, it's a US org and they do have several international sites (Italy, Switzerland, HK, Spain ... UK in the past). I think the science is awesome and, at ~$1300 for 5 days incl. fees, food, room, it's a great deal.
I agree it's complex and there are many layers to it. But then look at Nat Comm which published >10,000 papers last year, probably absorbing >$70m alone. Contrast this to the 2nd most cited paper ever that wasn't peer-reviewed at all ... The current system needs a fundamental overhaul, not tweaking
