Owl Posting Can AI improve the current state of molecular simulation? (Corin & Ari Wagen, Ep #1)
4 snips
Dec 3, 2024 In this engaging discussion, Corin Wagen, co-founder of Rowan, a breakthrough molecular simulation company, shares insights alongside his brother Ari. They delve into neural network potentials and their potential to revolutionize drug design and materials science. The conversation tackles the dichotomy between classical and quantum simulations, along with the importance of improved accuracy in computational chemistry. Corin and Ari also explore the challenges of benchmarking simulations and the role of AI in enhancing molecular dynamics, making a complex field more accessible.
AI Snips
Chapters
Transcript
Episode notes
Emergent Crystallization
- Tim's crystallization work observed nucleation events using NNPs, even in potentially unrealistic conditions.
- This suggests NNPs can learn fundamental relationships about chemistry and phase transitions.
Simulation vs. Reality
- Molecular dynamics simulations, even the best ones, show a gap between simulated and real-world behavior.
- For example, common water simulations solidify at room temperature, highlighting inaccuracies.
DFT: Accuracy Levels
- Density Functional Theory (DFT), used to train NNPs, is considered a high-accuracy method in life sciences.
- However, in high-accuracy simulation, DFT is considered a basic method, showing the complexity of electronic structure.