DeepQuark

Neural-network algorithms for multiquark bound states

Requirements

The codes are developed based on NetKet v3.17. A stable package version to run this code:

• Python >= 3.12
• JAX == 0.5.3
• NetKet == 3.17 See the NetKet installation guide for detailed instructions.

Example Usage

python multiquark/train.py AL1 ccqq 1 0  # potential quarks S I

The above code carries out the NN-VMC training of the doubly charmed tetraquark system $cc\bar q\bar q$ with total spin $S=1$ and isospin $I=0$ in the AL1 potential model. The default values of some other parameters are included in config_.py, you can override the default values by:

python multiquark/train.py AL1 ccqq 1 0 --nlayers 5 --nnodes 32 --bound 3 --sigma 0.05

For more information on the parameter inputs:

python multiquark/train.py -help

Reference

If you use this software, please cite our publication.