Filter
Exclude
Time range
-
Near
Day 43/50 #AtomGPTLab #50Apps Why does iron magnetize but aluminum doesn't? Stoner answered this in 1938: I·N(E_F) > 1. Now a browser app computes both sides. SlakoNet TB JARVIS-DFT. The first spin app of the campaign. 🧲 atomgpt.org/magnetic
2
119
Day 42/50 #AtomGPTLab #50Apps Not every material wants to be a crystal. Melt-quench MD compresses what nature does in microseconds. Heat past melting → quench rapidly → amorphous network. Zachariasen 1932 → ALIGNN-FF 2026. 🔥 atomgpt.org/amorphous
1
95
Piyush retweeted
Day 41/50 #AtomGPTLab #50Apps Day 40 told you WHETHER a reaction is favorable. Day 41 tells you HOW FAST. Eyring 1935 wrote the rate equation. Jónsson 1998 wrote the algorithm. Now it's a browser tab. 🚧 atomgpt.org/neb
1
1
2
368
Day 40/50 #AtomGPTLab #50Apps Balancing chemical equations is just linear algebra. Find the null space of the stoichiometry matrix. Now a browser tab plus JARVIS-DFT thermodynamics. Lavoisier 1789 → AtomGPT 2026. ⚛️ atomgpt.org/reaction_network
1
190
Day 39/50 #AtomGPTLab #50Apps Neutrons see hydrogen the way nothing else does. The 1994 Nobel-winning technique, now a browser tab. 7 real spectrometers modeled. JARVIS-DFT or ALIGNN-FF backend. 🌌 atomgpt.org/ins
3
264
Day 38/50 #AtomGPTLab #50Apps In 1928 Seishi Kikuchi saw lines in electron diffraction patterns nobody could explain. Today every SEM uses those lines to map crystal grains in steel, Al, additive-manufactured parts. Now it's a browser tab. 🔬 atomgpt.org/kikuchi
1
100
Day 36/50 #AtomGPTLab #50Apps Every materials database used to speak its own dialect. In 2016, OPTIMADE became the standard. Today AtomGPT.org joins the federation. 7 datasets, 1 filter syntax, 400K entries. 🌐 atomgpt.org/optimade_explore…
2
101
Day 35/50 #AtomGPTLab #50Apps Pb melts at 327°C. Sn melts at 232°C. Pb-Sn solder melts at 183°C; lower than either. That's CALPHAD. The 55-year-old methodology that designs every alloy you've ever touched, now free in your browser. 🔥 atomgpt.org/calphad
4
278
🚨 3 new papers from the #AtomGPTLab: 🔬 RamanGPT: crystal structures ⇄ Raman spectra arxiv.org/abs/2606.03764 🧪 Hybrid-LLM-GNN deterministics: iopscience.iop.org/article/1… ⚛️ XC functionals & Schottky barriers pubs.acs.org/doi/10.1021/acs… Open science 🔬
3
90
Day 34/50 #AtomGPTLab #50Apps Load any crystal (POSCAR/CIF/XYZ/PDB). Build supercells, surfaces, vacancies, strains. Run RDF, XRD, ALIGNN, SlaKoNet, all from one workbench. 🔬 atomgpt.org/structure_visual…
2
109
Day 33/50 #AtomGPTLab #50Apps A Wannier tight-binding Hamiltonian is a small matrix that contains every electronic property of a crystal. JARVIS-WTB: 1,800 of them, DFT-fitted, downloadable. Bandstructure DOS now. AHC, SHC, Chern numbers coming. ⚛️ atomgpt.org/wtb
3
205
Day 30/50 #AtomGPTLab #50Apps · 🎯 Midpoint The Shockley-Queisser limit (1961) says a single-junction solar cell maxes out at 33.7%. JARVIS-Solar computes both SQ AND SLME (the DFT-friendly upgrade) for 1000 materials. Browser, free, seconds. ☀️ atomgpt.org/solar
2
94
Day 28/50 #AtomGPTLab #50Apps Every battery cathode has a voltage curve. Computing it from DFT takes days. Now, paste a POSCAR, pick the ion (Li, Na, K, Mg, Ca, Zn), get the voltage profile theoretical capacity in minutes. ALIGNN-FF behind it. 🔋 atomgpt.org/battery
3
283
Day 26/50 #AtomGPTLab #50Apps A Shazam for Raman spectra: paste your peaks, find the matching material in seconds. Plus inverse (RamanGPT: spectrum → structure) and forward (ALIGNN: structure → spectrum) AI prediction. 🧪 atomgpt.org/raman
1
1
4
217
Day 25/50 #AtomGPTLab #50Apps The electric field gradient tensor Vᵢⱼ is what NMR, NQR, and Mössbauer spectra actually measure, local electronic anisotropy at every nucleus. EFG Explorer: per-site 3×3 tensors from JARVIS-DFT, in your browser. 🧲 atomgpt.org/efg
1
2
7
779
Day 24/50 #AtomGPTLab #50Apps How many of your AI-generated references are real? Probably not all of them. LLMs hallucinate plausible-looking citations constantly. Hallucination Detector cross-checks references against 200M papers . 🧪 atomgpt.org/hallucination_de…
2
173
Day 18/50 #AtomGPTLab #50Apps Perfect crystals are useless. Vacancies make them work- dopants, batteries, NV centers, catalysts. Vacancy Explorer: ~500 DFT point defects in 3D bulk 2D monolayers, formation energies, full decomposition. 🧪 atomgpt.org/vacancy
4
263
Day 12/50 #AtomGPTLab #50Apps 0 K convex hull phase diagrams from JARVIS-DFT, built on OptB88vdW (vdW-corrected), not PBE. Pick elements → 2D / ternary / 3D hull. Stable phases on hull, unstable above. 🧪 atomgpt.org/convexhull ▶️ youtube.com/watch?v=RLw9BBCl…
2
149
Day 11/50 #AtomGPTLab #50Apps AtomGPT.org now speaks protein. 3 tabs in one app: 🧬 ESMFold: sequence → structure 🧪 OpenFold3 (NVIDIA NIM): protein-DNA complex 🔍 PDB Explorer: 200k RCSB structures 🧪 atomgpt.org/protein ▶️ youtube.com/watch?v=kSA7nhfX…
1
3
202