mesolve slow

[Joaquin Guimbaogaspar]

Hi everyone,

I am having some issues when using the me solver either for small systems or large systems. In all cases, the solver is slower than the es solver, which makes no sense for me, specially for the case of large systems. 

In this specific case, the system consists on 5 atoms interacting with eachother and with a single-mode cavity field. The attached files contain matlab QOtoolbox version and the Qutip version of the code.

The code just set up the parameters of the system and builds the Hamiltonian and Lindblad equation. Finally it solves de equation with mesolve, and compute a 2op_2t correlation function. For the case of Qutip, the mesolve spends about 15 minutes for solving the equation, and 3 hours the correlation function. For the case of matlab toolbox, it takes a few seconds solving the equation and one minte computing the correlation function. Finally, if I try ode2solve in qutip, it solves the equation in just 45 seconds.

Is there  something I am doing terribly wrong with the mesolver in qutip?, or is just that the matlab toolbox has a much better solver?

Thank you very much,

Joaquin Guimbao

[Simon Cross]

For those following the list, Joaquin also posted this question to the issue tracker and there is a discussion taking place there: https://github.com/qutip/qutip/issues/1591.