Skip to content

Results are limited to Matrix Product State emulator (EMU_MPS) v2.1.1

Product news
You are reading Matrix Product State emulator (EMU_MPS) v2.1.1, an older version of this documentation. Go to the latest version (v2.11.0)

Summary of the TDVP algorithm

Emu-mps uses a second order 2-site TDVP to compute the time-evolution of the system (see here for details (external)). Briefly, the algorithm repeatedly computes the time-evolution for 2 neighbouring qubits while truncating the resulting MPS to keep the state small. It does this by

  • evolving qubit 1 and 2 forwards in time by dt/2dt/2
  • evolving qubit 2 backwards by dt/2dt/2
  • evolving qubit 2 and 3 forwards in time by dt/2dt/2

...

  • evolving qubit n−1n-1 and nn forward in time by dtdt
  • evolving qubit n−1n-1 backwards in time by dt/2dt/2
  • evolving qubit n−2n-2 and n−1n-1 forward in time by dt/2dt/2

...

  • evolving qubit 1 and 2 forwards in time by dt/2dt/2

The fact that we sweep left-right and the right-left with timesteps of dt/2dt/2 makes this a second-order TDVP.