qiskit.opflow has been deprecated, the way I used to generate a UCC ansatz was to use
(Parameter('t') * Pauli('some Pauli')).exp_i(). Now this is not possible, and I am unsure how to exponential my
SparsePauliOp. If I have a parameterized
SparsePauliOp such as
SparsePauliOp(['XZ', 'YZ'], coeffs=[ParameterExpression(1.0*a), ParameterExpression(1.0*b)]) for some
a, b = Parameter('a'), Parameter('b'), how do I exponentiate this to get
exp(-i(XZ*a + YZ * b)) which I could then Trotterize to decompose into gates?
Take a look at PauliEvolutionGate. The UCC from Qiskit Nature in based on Qiskit's EvolvedOperatorAnsatz which in turn is built using PauliEvolutionGate.
There is also a migration guide done as
opflow was deprecated, which has a section on evolution.