Logical Qubit

A logical qubit is a qubit represented by an encoding within a larger physical quantum system, usually to protect its information through quantum error correction. Unlike a physical qubit, which refers to a hardware-level information carrier, a logical qubit is defined by its encoded states and operations. Many implementations distribute one logical qubit across multiple physical qubits, although other encodings use the larger state space of an oscillator. Encoding provides a structure for error protection; it does not, by itself, guarantee better performance than an unencoded qubit. (nature.com)

Encoded states

The basis states of a logical qubit are conventionally written $|0_L\rangle$ and $|1_L\rangle$. An arbitrary pure logical state has the form

These basis states span a two-dimensional code space within the physical system’s Hilbert space. The encoding must preserve both the amplitudes and their relative phase, and therefore protect quantum superpositions, not merely distinguish logical zero from logical one. (arxiv.org)

A simple illustration is the three-qubit repetition encoding,

It can correct a single bit-flip error under the corresponding restricted noise model, but it does not protect against arbitrary single-qubit errors, including phase flips. Full quantum error correction requires protection against both types. The encoding produces $\alpha|000\rangle+\beta|111\rangle$, not three independent copies of an unknown state, and thus does not violate the no-cloning theorem. (arxiv.org)

Error detection and recovery

Noise, including quantum decoherence, can move an encoded state into an error subspace or alter its logical information. Error correction uses measurements designed to reveal information about correctable errors without revealing the unknown logical state itself. (doi.org)

In a stabilizer code, commuting check operators define the code space. Their measurement outcomes form an error syndrome. A classical decoding algorithm interprets this syndrome and chooses a recovery operation. Different physical errors can have the same syndrome; successful decoding requires a recovery that restores the logical information, rather than necessarily identifying the exact physical error. (quantum.cloud.ibm.com)

Practical systems repeat syndrome measurements because errors continue to occur and the measurements themselves are imperfect. Some implementations apply physical corrections, while others track suitable corrections classically and account for them in subsequent operations or readout. (nature.com)

Code parameters and logical errors

A quantum code is commonly described by the notation $[[n-k-d|n,k,d]]$, where $n$ is the number of physical data qubits, $k$ the number of encoded logical qubits, and $d$ the code distance. Additional measurement and control qubits are generally not included in $n$. For a stabilizer code, distance is the minimum weight of a Pauli operator that acts nontrivially on the encoded information while escaping the code’s checks. “Weight” counts the physical qubits on which that operator acts nontrivially. (nature.com)

With ideal recovery, a distance-$d$ code can correct arbitrary errors affecting up to

physical qubits. This mathematical guarantee differs from the performance of an actual error-correction circuit, whose gates, measurements, and resets can introduce further errors. (quantum.cloud.ibm.com)

A logical error is an error in the encoded information that remains after recovery. Logical error rates must be specified for a particular task—for example, per correction cycle, per logical gate, or over a storage interval—because these quantities are not interchangeable. (nature.com)

Physical overhead and error thresholds

There is no fixed conversion between physical and logical qubits. Resource requirements depend on the code, hardware connectivity, noise, decoder, and required logical accuracy. A common rotated surface code patch of distance $d$ uses $d^2$ data qubits and $d^2-1$ check qubits to encode one logical qubit; other code families can encode several logical qubits in a shared block. (nature.com)

Below an appropriate error threshold, increasing code distance can suppress logical errors. The threshold depends on the code and its implementation, rather than being a universal physical-qubit error rate. Above threshold, adding physical resources need not improve reliability. Correlated noise and leakage outside the intended qubit states can also limit error suppression. (nature.com)

Bosonic encodings provide a different resource model: a logical qubit can occupy selected states of a quantum oscillator, such as a microwave cavity, with auxiliary hardware used for control and error detection. Consequently, counting two-level physical qubits alone does not fully describe their overhead. (nature.com)

Logical operations and fault tolerance

Logical quantum gates manipulate encoded information while maintaining its protection. A fault-tolerant implementation must control error propagation so that faults during gates or syndrome extraction do not overwhelm the code’s correction capability. Protecting idle quantum information is therefore only one part of building an error-corrected quantum computer. (learning.quantum.ibm.com)

Logical-qubit demonstrations accordingly distinguish several achievements: preparing encoded states, detecting errors, correcting them repeatedly, outperforming an unencoded reference, and implementing protected gates. Exceeding a defined break-even benchmark establishes an advantage for the measured task, but does not alone establish a complete fault-tolerant processor. For example, research published online on December 9, 2024 demonstrated surface-code memories with logical errors decreasing as code distance increased, while separately identifying further requirements for large-scale computation. (nature.com)

References

1. Stabilizer Codes and Quantum Error Correctionarxiv.org
2. Stabilizer codes | IBM Quantum Learningquantum.cloud.ibm.com
3. Fault-tolerant quantum computation | IBM Quantum Learninglearning.quantum.ibm.com
4. Controlling error propagation | IBM Quantum Learningquantum.cloud.ibm.com
5. High-threshold and low-overhead fault-tolerant quantum memorynature.com
6. Logical quantum processor based on reconfigurable atom arraysnature.com
7. Beating the break-even point with a discrete-variable-encoded logical qubitnature.com
8. Beating the break-even point with a discrete-variable-encoded logical qubitdoi.org
9. Quantum error correction below the surface code thresholdnature.com