tkwant.manybody.WaveFunction

class tkwant.manybody.WaveFunction(psi_init, tasks, comm=None)

Evolve a many-particle wavefunction in time.

Initialize the manybody state.

Parameters

• psi_init (dict of tkwant.onebody.WaveFunction) – Dictionary with all initial one-body states. For load balancing the dictionary should be distributed over all MPI ranks.

• tasks (dict of tkwant.onebody.Task) –

  Dictionary containing the weighting factor for each one-body state. Each item must have at least the following attribute:

  • weight : ndarray, weighting factor

  tasks must include all one-body states stored in psi_init and must be the same on all MPI ranks.

• comm (Intracomm, optional) – The MPI communicator over which to parallelize the computation. By default, use the tkwant global MPI communicator.

Methods

add_distributed_onebody_states(states, tasks)

Add several onebody wavefunctions to the manybody state.

Parameters

• state (dict of tkwant.onebody.WaveFunction) – Dictionary of onebody wavefunctions to be added

• task (dict of tkwant.onebody.Task) –

  Task info of the states. Each tasks item must contain at least the weight factors of the onebody states in the manybody average as attribute:

  • weight : ndarray, weighting factor

Notes

The states present in the states dictionary can be distributed over all MPI ranks. Each single onebody state must be unique and not dublicated on any other MPI rank.

add_onebody_state(state, task, key=None, rank=None)

Add a onebody wavefunction to the manybody state.

Parameters

• state (tkwant.onebody.WaveFunction) – onebody wavefunction

• task (tkwant.onebody.Task) –

  Task info of the state. task must contain at least the weight factors of the onebody states in the manybody average as attribute:

  • weight : ndarray, weighting factor

• key (dict key, optional) – Idendentifier of the state. A free key is attributed if not present.

• rank (int, optional) – MPI rank where the state should be added.

Returns

key – Idendentifier of the added onebody state.

Return type

dict key

Notes

The state must be present on all MPI ranks.

add_perturbation(qt)

set perturbation

delete_onebody_state(key)

Delete a onebody wavefunction corresponding to its key.

Parameters

key (dict key) – Idendentifier of the onebody state that should be deleted.

evaluate(observable, root=0)

Evaluate the expectation value of an operator at the current time.

Parameters

• observable (callable or kwant.operator) – An operator to evaluate the expectation value. Must have the calling signature of kwant.operator.

• root (int or None, optional) – MPI return rank on which to return the result. If root is an integer, it must be in the range of valid MPI ranks 0 <= root < self.comm.size. In that case, the calculated result is returned only on that specific MPI rank where rank == root, whereas the result is None on all other MPI ranks with rank != root. Alternatively, if root is None, the calculated result is returned on all MPI ranks. By default, the result is returned on MPI rank zero only.

Returns

result – The expectation value of observable, integrated over all occupied bands. The result might not be returned on all MPI ranks; note the explanation above for input parameter root.

Return type

ndarray

Notes

Returning the result on all MPI ranks (by setting root=None), might be slower, as an additional broadcast step is needed.

evolve(time)

Evolve all wavefunctions up to time.

Parameters

time (int or float) – time argument up to which the solver should be evolved

get_free_key()

Get a new free key.

Returns

key – New unused key. All numbers larger then the returned key are also valid empty keys.

Return type

int

get_keys()

Get the keys of all onebody wavefunctions forming the manybody state.

Returns

keys – List of all state identifier keys present in the solver.

Return type

list

get_onebody_state(key, root=0)

Get the onebody wavefunction corresponding to its key.

Parameters

• key (dict key) – Idendentifier key of the state.

• root (int or None, optional) – MPI return rank on which to return the state. If root is an integer, it must be in the range of valid MPI ranks 0 <= root < self.comm.size. In that case, the state is returned only on that specific MPI rank where rank == root, whereas the result is None on all other MPI ranks with rank != root. Alternatively, if root is None, the calculated state is returned on all MPI ranks. By default, the state is returned on MPI rank zero only.

Returns

state – onebody wavefunction

Return type

tkwant.onebody.WaveFunction

weight_shape()

Get the shape of the weighting factor.

Returns

shape – Shape of the weight factor array.

Return type

tuple