computer science//quantum computing

Quantum computing is computation carried out on physical systems whose state follows quantum mechanics, using qubits instead of bits, and it is pursued because for a few specific problems (simulating molecules and materials, factoring large numbers, some search and optimization tasks) quantum algorithms need far fewer steps than any known classical method. For everything else a quantum computer is slower, more expensive and less reliable than a laptop.

Quantum computing is computation carried out on physical systems whose state follows quantum mechanics, using qubits instead of bits, and it is pursued because for a few specific problems (simulating molecules and materials, factoring large numbers, some search and optimization tasks) quantum algorithms need far fewer steps than any known classical method. For everything else a quantum computer is slower, more expensive and less reliable than a laptop.

What all its parts share is the physics they exploit and the fragility it brings. A register of qubits can hold a superposition of many classical states at once, and an algorithm arranges interference so that wrong answers cancel and the right one becomes likely when the register is measured. Measurement then returns one classical result, with a probability; a computation is run many times and its statistics read. The same sensitivity that makes the states useful makes them easy to destroy: any stray coupling to the environment, a thermal photon, a control pulse slightly off, scrambles them (decoherence).

A quantum computer is a classical machine with a fragile accelerator in the middle.

An ordinary computer compiles the program into microwave or laser pulses, drives the qubits through control electronics, reads out the measurements and does all the post-processing; the quantum chip only runs the short stretch of operations between preparation and readout.

The core is extreme hardware. Superconducting qubits (Google, IBM) sit in a dilution refrigerator at about ten millikelvin, colder than outer space, with hundreds of control lines running down into it; trapped ions and neutral atoms work at room-temperature vacuum chambers with lasers instead, and photonic qubits use light. Each technology trades speed, connectivity and error rates differently.

Errors are the central engineering problem. Today's physical qubits fail roughly once every few hundred to few thousand operations, while the useful algorithms need billions; quantum error correction builds reliable logical qubits out of many physical ones, at an overhead of hundreds to a thousand per logical qubit.

The advantage depends on the problem and the algorithm. Shor's algorithm factors integers exponentially faster than the best known classical method, which is why public-key cryptography is migrating to post-quantum schemes; Grover's search gives only a quadratic speedup, often eaten by the error-correction overhead. Demonstrations so far are benchmark tasks chosen to be hard for classical machines and useless otherwise (quantum advantage).

For an engineer the practical reading is to separate three things the press blurs: real quantum hardware, quantum algorithms that would pay off on a fault-tolerant machine that does not yet exist, and quantum-inspired algorithms that run on ordinary computers.