TensorNetworkAD.jl
This is a package with the goal to implement the algorithms described in Differentiable Programming Tensor Networks, namely implementing automatic differentiation (AD) on Corner Transfer Matrix Renormalization Group (CTMRG) and Tensor Renormalization Group (TRG), demonstrating two applications:
- Gradient based optimization of iPEPS
- Direct calculation of energy densities in iPEPS via derivatives of the free energy
We aimed for readable and easy to extend code that demonstrates advantages of julia (seamless integration of different packages, performance, readability) and some cutting edge physics.