---
fixes:
- |
Fixed bug in :class:`.QNSPSA` which raised a type error when the computed fidelities
happened to be integers but the perturbation was of type float.