5 ms·

Here's a link to the actual article in PNAS: https://www.pnas.org/content/early/2019/02/14/1818555116 https://www.pnas.org/content/early/2019/02/14/1818555116

• wenc 8y ago
  Ok, so here's where they use NN models: "Although ab initio calculations such as those involving many-body corrections can provide accurate energy-band results, the scope of such calculations is somewhat limited to about 1,000 strain points because of high computational cost. On the other hand, by discretizing ε with a regular grid comprising 20 nodes separated at each 1% strain interval over the strain range of −10 to +10%, the computational model would entail about 108 band structures, up to five orders of magnitude higher computational requirement than what can be reasonably achieved presently. To overcome these difficulties, we present here a general method that combines machine learning (ML) and ab initio calculations to identify pathways to ESE. This method invokes artificial neural networks (NNs) to predict, to a reasonable degree of accuracy, material properties as functions of the various input strain combinations on the basis of only a limited amount of data." Further down: "We aim to describe the electronic bandgap and band structure as functions of strain by training ML models on first-principles density-functional theory (DFT) data. This approach leads to reasonably accurate training with much fewer computed data than fine-grid ab initio calculations and a fast evaluation time." -- If I'm reading this correctly, it sounds like they already have a high-fidelity 1st-principles model that is computationally intractable to solve at scale, so they are using ML techniques to create an surrogate model that is computationally more tractable -- the AI modeling is a model reduction exercise.
  • leplen 8y ago
    That's a pretty good summary. There are many different sets of approximations and simulation techniques that make different trade-offs of accuracy/scale. They're essentially using outputs of a higher-fidelity model to tune the free parameters of a lower fidelity model, and using ML to explore the parameter space efficiently. The higher fidelity model is itself still an approximation, but there's a lot of interest in this approach and quite a few groups doing work like this targeted towards various material properties, since there just isn't nanoscale experimental data to train models that depend on nanoscale material features.
    • steve_musk 8y ago
      Agreed, although I would hesitate to call the PBE functional a “first principles” model.
  • gumby 8y ago
    > If I'm reading this correctly, it sounds like they already have a high-fidelity 1st-principles model that is computationally intractable to solve at scale, so they are using ML techniques to create an surrogate model that is computationally more tractable -- the AI modeling is a model reduction exercise. So in the end is this really different from simulated annealing (which appealingly actually sounds like materials science process)?
    • wenc 8y ago
      It would seem yes?... unless I'm not seeing a deeper connection. Simulated annealing is a technique--like gradient descent--for finding the optimum of some model. Surrogate modeling on the other hand is the creation of a simpler approximation to a more complicated model. They would seem to me to be quite different things, the former being an optimization technique and the latter being a modeling one.