Skip to content

Latest commit

 

History

History
 
 

bimol_train

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
 
 
 
 
 
 
 
 
 
 
 
 

Train the model for bimolecular annihilation A+A->0

For the reaction A + A -> 0.

Create data directories

Run the included Python scripts

python create_data_dirs.py

Running

First use cmake:

mkdir build
cd build
cmake ..

Input data

This requires the stochastic simulations in the bimol_train_stoch_sim directory.

Learning the hidden layer model

From the build directory:

make main_hidden
cd ../bin
./main_hidden

The data is in the data_learned_hidden directory.

The progress can be monitored with the hidden_monitor_progress notebook in the mathematica directory.

After learning, the complete trajectories can be constructed by running from the build directory:

make learned_trajs_hidden
cd ../bin
./learned_trajs_hidden

The data is in the data_learned_hidden/trajs directory.

Learning the visible layer model

As above, replacing hidden with visible.

Paper figures

Paper figures can be generated using the notebooks in the mathematica directory:

  • hidden_diff_eq_rhs,visible_diff_eq_rhs visualize the learned right hand sides of the differential equations.
  • correlation calculates the correlations in the stoch sims in bimol_train_stoch_sim
  • colored_trajs creates the nearest-neighbor colored initial lattice configurations.