# Coherence Time **Coherence time** is how long a [[Qubit|qubit]] keeps its quantum information before noise scrambles it. It is the qubit's "shelf life." ## First principles A qubit's power comes from the precise relationship (amplitude and phase) between $|0\rangle$ and $|1\rangle$. The environment — stray fields, vibrations, heat, stray photons — constantly nudges the qubit, randomizing that relationship. This loss is called **decoherence**. Two characteristic times describe it: - $T_1$ — **relaxation**: how long until the qubit randomly flips, e.g. $|1\rangle \to |0\rangle$ (energy leaks out). - $T_2$ — **dephasing**: how long the *phase* relationship stays intact. $T_2$ is usually the stricter limit. > [!intuition] A spinning coin > Think of a coin spinning on a table, encoding "heads-and-tails at once." Coherence time is how long it keeps spinning cleanly before friction and bumps make it wobble and fall to a definite face. ## Why it matters A computation is a race: you must finish your [[Gate Fidelity|gates]] before coherence runs out. The figure of merit is not the raw time but the **ratio of coherence time to gate time** — how many operations fit inside the window. For [[Quantum Error Correction]] this is doubly important: error correction must run continuously for many [[Syndrome Extraction]] cycles, so the longer the underlying coherence, the more rounds of protection you can apply. Encodings like [[Nuclear-Spin Qubits]] are prized precisely for their long coherence. ## Related - [[Qubit]] - [[Gate Fidelity]] - [[Nuclear-Spin Qubits]] - [[Quantum Error Correction]]