Quantum superposition is the quantum-mechanical principle that a system can occupy a coherent linear combination of possible states rather than being restricted to one member of a chosen set of basis states. For a qubit, a general pure state can be written as alpha times state 0 plus beta times state 1, with complex amplitudes alpha and beta satisfying the normalisation condition.
The state is not equivalent to an ordinary classical mixture in which the system is secretly in either state but the observer does not know which one. A coherent superposition has relative phase information that can produce interference when the state is subsequently transformed and measured. Superposition is therefore a mathematical and physical property of quantum states, not simply a statement about uncertainty or incomplete knowledge. It is one of the core resources exploited by quantum algorithms, although useful quantum computation requires controlled interference and other operations in addition to superposition alone.
The distinction between superposition and classical uncertainty is fundamental. Suppose a classical bit is either 0 or 1 but an observer lacks information about which value it has; this is a probabilistic mixture. A qubit in a coherent superposition can have amplitudes for both basis states, and the relative phase between those amplitudes affects future measurement probabilities after suitable quantum operations.
Two states can have identical probabilities when measured immediately in one basis but behave differently under another operation because their relative phases differ. Consequently, superposition contains information that cannot be represented solely by assigning classical probabilities to mutually exclusive states. In density-matrix language, coherent off-diagonal terms distinguish a coherent superposition from an incoherent statistical mixture. Decoherence can suppress those coherence terms, turning behaviour increasingly like a classical mixture.
For a single qubit, the Hadamard gate provides a simple example of superposition. Applied to the computational-basis state 0, an ideal Hadamard operation produces a state with equal-magnitude amplitudes for 0 and 1, with the appropriate phase convention. Measuring immediately in the computational basis then gives either result with equal probability. Applying the same operation again returns the original basis state because the Hadamard transformation is its own inverse.
This demonstrates that the intermediate superposition is not merely a random selection between two classical values. The second operation acts on the amplitudes and their phases, causing interference that deterministically reconstructs the original state in the ideal case. The ability to create and later recombine coherent alternatives is central to quantum algorithms.
Superposition becomes computationally significant through interference. Quantum amplitudes are complex numbers, so they can have both magnitude and phase. When quantum paths or components lead to the same final state, their amplitudes add. Components with compatible phases can reinforce one another, whereas components with opposing phases can cancel. Algorithms exploit this behaviour to reshape the probability distribution of measurement results.
In a simplified example, an algorithm can place several candidate computational states into a superposition and then apply phase-dependent transformations so that unwanted candidates experience destructive interference while desired candidates experience constructive interference. The measurement then has a greater probability of producing useful results. This is why the popular statement that quantum computers "try all possibilities simultaneously" is incomplete: even if a quantum state contains amplitudes associated with many possibilities, the computation must manipulate those amplitudes so that measurement reveals useful information.
Superposition also applies to physical observables beyond computational qubits. In quantum mechanics, a system can be in a superposition of energy eigenstates, position states, spin states or other basis states, depending on the physical system and chosen description. The same physical state can be represented using different bases, and a state that is a basis state in one representation may be a superposition in another. This basis dependence is mathematically important in quantum information.
Measurement selects an observable and therefore a basis or set of measurement outcomes associated with that observable. A qubit in a computational-basis state can, for example, be a superposition of states in a different measurement basis. Thus superposition is not an assertion that a system possesses contradictory classical properties simultaneously in every possible sense; it describes the state relative to a chosen set of quantum alternatives and the observables used to interrogate it.
Maintaining superposition requires quantum coherence. Environmental interactions can reveal information about a system's state to its surroundings and cause the relative phase relationships needed for interference to degrade. This process is generally described as decoherence. Temperature, electromagnetic noise, material defects, fluctuating fields, imperfect isolation and control errors can all contribute to decoherence depending on the hardware platform.
Quantum computers therefore operate with elaborate methods to isolate and control qubits, such as cryogenic environments for superconducting devices, vacuum systems for trapped ions, optical control for atomic systems and specialised shielding or filtering. Coherence is not itself sufficient for useful computation: gates and measurements must also be accurate. Nevertheless, maintaining coherent superpositions long enough to perform the required operations is a fundamental engineering requirement.
Superposition should also be distinguished from entanglement. Superposition can exist for a single qubit and does not require multiple particles or qubits. Entanglement is a particular type of joint quantum state involving multiple subsystems whose correlations cannot be represented as independent states for the individual subsystems. A multi-qubit entangled state is still mathematically a superposition of computational-basis states, but not every superposition is entangled.
For example, a product state in which each of two qubits independently occupies a superposition is a multi-qubit superposition without entanglement. Entangling operations create correlations that cannot be separated into independent single-qubit descriptions. The concepts are therefore related but distinct: superposition concerns linear combinations of basis states, while entanglement concerns the non-separable structure of composite quantum states.
The practical importance of quantum superposition is that it supplies a state space in which quantum algorithms can manipulate amplitudes and phases before measurement. Superposition can support quantum simulation, interference-based algorithms, sensing protocols and other quantum-information tasks. However, it does not imply unlimited parallel classical computation or guaranteed speedup. Measurement generally yields a single outcome in the selected basis, and the useful information must be encoded into the outcome distribution.
Real systems also suffer from noise and decoherence, so maintaining and exploiting superposition is technologically demanding. Historically, the concept emerged from the general linear structure of quantum mechanics and became central to quantum-information theory as researchers recognised that coherent quantum states could themselves be treated as computational resources. In modern quantum computing, superposition is best understood as a precise property of quantum-state amplitudes and coherence that becomes computationally useful when combined with controlled gates, interference, entanglement and measurement.
