We already have the Jordan-Wigner transform (more generally, the entire procedure from pyscf to the Qiskit qubit Hamiltonian). We could look for something on the same level, something higher level, or even something lower level.
I favor looking for other features of complexity on the level of the Jordan-Wigner transformation, or more ambitious.
I would be cautious about investing time in lower level improvements in efficiency that are not currently part of a bottle neck. But, these still might be worth it because:
Once you are committed to installing qiskit-alt, the marginal cost of using another feature is not so high (you still are leaving Python, which means, debugging, for example, might be harder)
These may be useful in catalyzing ideas
An accumulation of such changes might have significant effects.
We already have the Jordan-Wigner transform (more generally, the entire procedure from pyscf to the Qiskit qubit Hamiltonian). We could look for something on the same level, something higher level, or even something lower level.
I favor looking for other features of complexity on the level of the Jordan-Wigner transformation, or more ambitious.
I would be cautious about investing time in lower level improvements in efficiency that are not currently part of a bottle neck. But, these still might be worth it because: