A cluster optimized Python based simulation framework to simulate 1D Rydberg tweezer dynamics utilizing tensor network methods.
The parameters of the simulation can be changed in mpsPhonons.py. Different initial states and hamiltonians can be added in the file initialStatesClass.py.
During the simulation, after every timestep all results are stored to a new dataframe in HDF5 file format.
This leads to large files with redundant values while adding a layer of security for timeouts or crashing simulations.
Moreover, a SLURM array job launched with runArray.slurm creates many of those files.
To remove the redundancy and combine array result data after a successful simulation ,the script minify_h5.py can be used.
Data can be plotted and analyzed with plotMultipleDataSets.py.
Old simulation data with corresponding figures can be found in ./data/
The following features are supported already:
- Supply a list of
$\Omega$ -values to simulate - Supply a list of
$V$ -values to simulate - Specify the number of atoms
$N$ - Specify a list of single- and multi-site observables you want to measure the time-evolution during the simulation
- Supply a list of time-points you want to calculate the observables
- Specify numeric tolerance and timeout thresholds
- Specify trap-frequencies
$\omega_\mathrm{trap}$ which model delocalization of atoms in a tweezer-trap - Calculate
$V$ directly via interatomic distance$a_0$ and given coefficients$C_3$ and$C_6$ - correct formula is then:
$V(r_j,r_k)=V(a_0)+\sum_\nu^{\inf} d/dr^\nu V(r)|_{a_0} (a_j^\dagger+a_j-a_k^\dagger - a_k)$ - Check whether higher order potential expansions are worth
- color{blue}{They are, when
$\omega$ is small.}\textcolor{red}{CHECK: When the spreading$\Delta x$ in the harmonic traps is larger than$a_0$ .} - Set atom at one harmonic oscillation level corresponding to a specific temperature
$(n=1,2,3...)$ - Specify a distribution of levels
$n$ for the atoms. For example$80%$ in$n=1$ ,$20%$ in$n=2$ . - Add long-range interaction beyond nearest-neighbour interactions
- Specify the order of long-range interaction
- Supply a µ-wave tuned interaction potential (maybe a potential of this form can also be written as
$C_3/r^3+C_6/r^6$ , so the simulation is already able to handle it) - Do simulation with realistic potential with the
$C_3$ and$C_6$ values corresponding to a specific pair-state. We have to change the Hamiltonian for this. We combine now the phonons and spins to a local quantum object of dimension$2M$ . In this fashion we can extend the mps-simulation with phonon-modes. - Estimate errors of observables
- Add dissipative processes like radiative decay, dephasing etc.
- Specify different potentials for trapped and anti-trapped Rydberg atoms. Excited Rydberg atoms are less or even anti-trapped compared to well trapped ground-states. Not relevant, if traps are off. Maybe do Quench dynamics?
- Do simulation with exact calculated pair-potential out of i.e. pairinteraction