[dynamics_options]
from qutip import Options
import numpy as np
Occasionally it is necessary to change the built in parameters of the dynamics solvers used by for example the qutip.mesolve and qutip.mcsolve functions. The options for all dynamics solvers may be changed by using the Options class qutip.solver.Options.
[dynamics_options]
options = Options()
the properties and default values of this class can be view via the print function:
[dynamics_options]
print(options)
Output:
[dynamics_options]
atol: 1e-08 rtol: 1e-06 method: adams order: 12 nsteps: 1000 first_step: 0 min_step: 0 max_step: 0 tidy: True num_cpus: 2 norm_tol: 0.001 norm_steps: 5 rhs_filename: None rhs_reuse: False seeds: 0 rhs_with_state: False average_expect: True average_states: False ntraj: 500 store_states: False store_final_state: False
These properties are detailed in the following table. Assuming options = Options():
table-striped
| Property | Default setting | Description |
|---|---|---|
| options.atol | 1e-8 | Absolute tolerance |
| options.rtol | 1e-6 | Relative tolerance |
| options.method | 'adams' | Solver method. Can be 'adams' (non-stiff) or 'bdf' (stiff) |
| options.order | 12 | Order of solver. Must be <=12 for 'adams' and <=5 for 'bdf' |
| options.nsteps | 1000 | Max. number of steps to take for each interval |
| options.first_step | 0 | Size of initial step. 0 = determined automatically by solver. |
| options.min_step | 0 | Minimum step size. 0 = determined automatically by solver. |
| options.max_step | 0 | Maximum step size. 0 = determined automatically by solver. |
| options.tidy | True | Whether to run tidyup function on time-independent Hamiltonian. |
| options.store_final_state | False | Whether or not to store the final state of the evolution. |
| options.store_states | False | Whether or not to store the state vectors or density matrices. |
| options.rhs_filename | None | RHS filename when using compiled time-dependent Hamiltonians. |
| options.rhs_reuse | False | Reuse compiled RHS function. Useful for repetitive tasks. |
| options.rhs_with_state | False | Whether or not to include the state in the Hamiltonian function callback signature. |
| options.num_cpus | installed num of processors | Integer number of cpus used by mcsolve. |
| options.seeds | None | Array containing random number seeds for mcsolver. |
| options.norm_tol | 1e-6 | Tolerance used when finding wavefunction norm in mcsolve. |
| options.norm_steps | 5 | Max. number of steps used to find wavefunction's norm to within norm_tol in mcsolve. |
| options.steady_state_average | False | Include an estimation of the steady state in mcsolve. |
| options.ntraj | 500 | Number of trajectories in stochastic solvers. |
| options.average_expect | True | Average expectation values over trajectories. |
| options.average_states | False | Average of the states over trajectories. |
| options.openmp_threads | installed num of processors | Number of OPENMP threads to use. |
| options.use_openmp | None | Use OPENMP for sparse matrix vector multiplication. |
As an example, let us consider changing the number of processors used, turn the GUI off, and strengthen the absolute tolerance. There are two equivalent ways to do this using the Options class. First way,
[dynamics_options]
options = Options() options.num_cpus = 3 options.atol = 1e-10
or one can use an inline method,
[dynamics_options]
options = Options(num_cpus=4, atol=1e-10)
Note that the order in which you input the options does not matter. Using either method, the resulting options variable is now:
[dynamics_options]
print(options)
Output:
[dynamics_options]
atol: 1e-10 rtol: 1e-06 method: adams order: 12 nsteps: 1000 first_step: 0 min_step: 0 max_step: 0 tidy: True num_cpus: 4 norm_tol: 0.001 norm_steps: 5 rhs_filename: None rhs_reuse: False seeds: 0 rhs_with_state: False average_expect: True average_states: False ntraj: 500 store_states: False store_final_state: False
To use these new settings we can use the keyword argument options in either the func:qutip.mesolve and qutip.mcsolve function. We can modify the last example as:
>>> mesolve(H0, psi0, tlist, c_op_list, [sigmaz()], options=options)
>>> mesolve(hamiltonian_t, psi0, tlist, c_op_list, [sigmaz()], H_args, options=options)or:
>>> mcsolve(H0, psi0, tlist, ntraj,c_op_list, [sigmaz()], options=options)
>>> mcsolve(hamiltonian_t, psi0, tlist, ntraj, c_op_list, [sigmaz()], H_args, options=options)