I am currently using OpenMPS to calculate the dynamics of a system of bosons in a lattice where I add some external harmonic potential. By now I have only made some tests adapting the example code for Bose Hubbard quench dynamics. The idea is to basically calculate the time-evolution of the width of the distribution of particles in the trap after a sudden change of the trapping frequency (something like "breathing" oscillations).
The result should be some coherent oscillations in this quantity, but at some point it totally breaks down (see attached figure "echo.pdf"). It always happens around the same value of t (total time evolution takes t=15), regardles of the size of the time step, type of change in the potential, and even the choice of convergence parameters (I am currently using the default values, which I expected should be enough for this small system). Weirdly, the Loschmidt echo seems to work fine (see echo.pdf). The main problem is some discontinuity in the time evolution of the density operator 'n', I think. Any ideas on how I could get rid of this? I am attaching the code too, sorry about the mess but I am trying several different things.
I had some troubles running your script, but as far as I believe I did not change the physics or calls to OSMPS. (E.g. mix of tabs and spaces for indentation ...) In addition, I could not see where you did the plots in the script. I have attached the modified script and the plot I get, where I replaced the width with the occupation of the 6th site. That works for me. Krylov works too, although it might have in general problems with sudden quenches.
If the problem persists, we need a full working script including the plotting to help you. The more I have to guess, the more likely I debug something else instead of your problem. As it looks now, it would be good to know the operating system, the version of OSMPS, the compiler, and the compiler optimizations.
Hello everyone,
I am currently using OpenMPS to calculate the dynamics of a system of bosons in a lattice where I add some external harmonic potential. By now I have only made some tests adapting the example code for Bose Hubbard quench dynamics. The idea is to basically calculate the time-evolution of the width of the distribution of particles in the trap after a sudden change of the trapping frequency (something like "breathing" oscillations).
The result should be some coherent oscillations in this quantity, but at some point it totally breaks down (see attached figure "echo.pdf"). It always happens around the same value of t (total time evolution takes t=15), regardles of the size of the time step, type of change in the potential, and even the choice of convergence parameters (I am currently using the default values, which I expected should be enough for this small system). Weirdly, the Loschmidt echo seems to work fine (see echo.pdf). The main problem is some discontinuity in the time evolution of the density operator 'n', I think. Any ideas on how I could get rid of this? I am attaching the code too, sorry about the mess but I am trying several different things.
Thanks,
Rafael B.
Hello Rafael,
I had some troubles running your script, but as far as I believe I did not change the physics or calls to OSMPS. (E.g. mix of tabs and spaces for indentation ...) In addition, I could not see where you did the plots in the script. I have attached the modified script and the plot I get, where I replaced the width with the occupation of the 6th site. That works for me. Krylov works too, although it might have in general problems with sudden quenches.
If the problem persists, we need a full working script including the plotting to help you. The more I have to guess, the more likely I debug something else instead of your problem. As it looks now, it would be good to know the operating system, the version of OSMPS, the compiler, and the compiler optimizations.
Best regards,
Daniel
Hello Daniel,
Thanks for the feedback and the files. It seems to be working well there indeed, so it must be some issue with the data handling.
I will try to fix this and I'll write back if necessary.
Best,
Rafael