create_arbitrary_state gives different results when too many qubits are specified

[apozas]

When using the function grove.alpha.arbitrary_state.arbitrary_state(vector, qubits), inserting qubits that will not be used in the qubits option modifies the output program and the results. Only when len(qubits) =ceil(log2(len(vector))) (i.e., only when the number of qubits specified equals the minimum number of qubits for encoding the state) the result is correct. Here is a MWE:

import pyquil.api as api
import pyquil.quil as pq
from pyquil.gates import I
from grove.alpha.arbitrary_state.arbitrary_state import create_arbitrary_state

good = pq.Program()
bad = pq.Program()
qvm = api.QVMConnection()
# To show that the problem is independent of the position of the qubits
for i in range(6):
    good.inst(I(i))
    bad.inst(I(i))
good.inst(create_arbitrary_state([0,0,0,0,1], [2,3,4]))
print('Good: ', qvm.wavefunction(good))
bad.inst(create_arbitrary_state([0,0,0,0,1], [2,3,4,5]))
print('Bad: ', qvm.wavefunction(bad))

The result being

Good: (1+0j)|010000>
Bad: (0.7071067812+0j)|010000> + (0.3535533906+0j)|100000> + (0.3535533906+0j)|100100> + (0.3535533906+0j)|101000> + (0.3535533906+0j)|101100>