Stochastic Gradient Thermostats with different numerical integrators

• Simulation 1 provides the results with a large range of different stepsizes
• Simulation 2 provides the estimate with stepsize 0.2
• Simulation 3 provides the estimate with stepsize 0.001

Please cite our AAAI paper if it helps your research:

@inproceedings{mSGNHT_AAAI2016,
  title={High-Order Stochastic Gradient Thermostats for Bayesian Learning of Deep Models},
  author={Li, Chunyuan and Chen, Changyou and Fan, Kai and Carin, Lawrence},
  booktitle={AAAI},
  Year  = {2016}
}