As I understand it, a Quantum Computer has qubits, where the system is in a state such as:
$$c_0 |S_0\rangle +\ c_1 | S_1\rangle \ + ... +\ c_n|S_n\rangle $$
where $c_0, c_1, ..., c_n$ are complex values that represent the entangled state that the qubits are really in.
Furthermore, we can manipulate these states in certain ways, such as by applying a unitary matrix, such as rotate around an axis.
However, these operations are performed by physical devices, and hence will not be perfect. For example, a quantum gate may be designed to rotate by $\theta$, but in practice, might rotate by $\theta + \epsilon$, for some small but nonzero $\epsilon$.
Some proposed Quantum Computations take up millions of such individual operations; what would prevent the small errors from each of these millions of operations accumulating, and actually overwhelm the desired result?
Note that I am not talking about decorrelation; where some physical qubit gets a completely unexpected value. Quantum Error Correction is supposed to handle that; however, QEC works by comparing various physical qubits; if they all drift off slightly, I don't see how QEC can correct for that.
Now, with conventional digital gates, they are designed so that each bit doesn't have to be exact; as long as it is close, the gate will act as designed (and produce an output which also might not be precisely correct, but close enough for the next gate).
Is there something similar for Quantum Computers?