English static mirror for SEO/GEO · AI-assisted translation · Read Chinese original

Using Measurement as the Gate: Photonic Quantum Computing and Boson Sampling Machines

Forum topic · 小凯 · 2026-11-07

Summary

This post, part of a quantum computing hardware series on zhichai.net, explains the photonic approach to quantum computing, whose philosophy differs from superconducting and trapped-ion platforms: qubits (photons) are ready-made, and the challenge is making them interact. It lists three photonic advantages: room-temperature operation, natural suitability for quantum networking, and readily available interferometry components like beam splitters (physical Hadamard gates) and phase shifters. The core obstacle is that photons barely interact, making two-qubit gates hard. The article presents two paths: (1) Boson sampling machines like Jiuzhang, which bypass universal computing by sampling photon distributions through interferometer networks, and (2) measurement-based quantum computing, where a large cluster state (pre-fabricated multi-photon entanglement) is prepared in advance, and computation proceeds by sequential measurements - the choice of measurement basis acts as the gate, with the program reduced to a measurement script. The post notes photonic integrated circuits (photonic waveguides, thin-film lithium niobate) as the key integration step, and identifies probabilistic photon sources and probabilistic fusion gates - whose success rates fall exponentially with circuit size - plus costly high-performance single-photon detectors, as the main weaknesses on the road to universal photonic quantum computing.

Using Measurement as the Gate: Photonic Quantum Computing and Boson Sampling Machines

> Prerequisite navigation: You need episode 049 (boson sampling and "Jiuzhang"), 021 (the beam splitter as a physical H gate), 016 (interference), 017 (changing axes). > Next episode preview: 064, "Tweezer Chessboard: Neutral Atom Arrays". > Related reading: The site article "Photonic Quantum Computing Moves into the Rack: Turing Quantum Gen3" documents the integration step featuring this episode's protagonist — a large optical table shrunk into a standard rack, the real-world version of "photonic integration" discussed at the end of this piece.

For the first two hardware routes (superconducting qubits, ion traps), the competitive focus is "whose qubits are better." The third route has a different philosophy: the qubits are ready-made — photons; the entire difficulty lies in "how to make them interact." And the answer it has produced may be the most subversive of all hardware platforms: the gate itself is implemented by measurement.

Isn't measurement the villain from episode 002 — the thing that "destroys everything with one look"? In this episode it completes its third transformation (it served as a process step in 035 and as a sentinel in 055) — it becomes the computing engine itself.

The Photon's Talent List

First, three talents photons bring as qubits:

One, robustness. Photons barely interact with the environment — in episode 006's fragility ranking they sit at the bottom (least fragile): room-temperature operation, no need for 0.01 K extreme cooling (only the detectors need refrigeration).

Two, they fly. A photon's job is propagation — the natural messenger of quantum networks (the best material for the dedicated lines of episode 025, and the lead actor of batch eight).

Three, they come with interferometry equipment built in. That half-silvered mirror from episode 021 (the beam splitter) is a physical H gate; add a phase shifter (the phase knob from episode 019), and every optical component of a single-qubit gate is available off the shelf — weaving dozens or hundreds of beam splitters and phase shifters into a network gives you a "photonic interference machine" (the body of "Jiuzhang" from episode 049).

Two Forks in the Road

After the talent list comes a dilemma: photons barely interact with each other — a two-qubit gate requires "interaction," and photons are famously aloof. If gates cannot be built, universal computing is out of reach. The road forks here:

Fork one: skip universal gates, go straight to being a sampler (episode 049). Jiuzhang's philosophy: since two-qubit gates are stillborn, don't do universal computing — herd photons into an interference maze and count the distribution of detection spots (boson sampling). No error correction, no deep circuits, yet the quantum-advantage demonstration still stands — the "bypass the weakness" route.

Fork two: pre-"weld" the circuit into one blob — use measurement as the gate. This is the more subversive path: since "executing gates one by one" is hard, pre-fabricate all the entanglement the computation needs in one shot — one large networked multi-photon entangled state (jargon: a cluster state). Then what does computation become? Becomes measuring this blob one qubit at a time, in order, along chosen axes:

  • Which qubit you measure (the order) sets the direction the computation advances;
  • Which axis you measure along (episode 017's axis change!) sets which "gate" is executed.
  • The choice of measurement = the choice of gate. The programmable part of the whole "computer" is compressed into a measurement script — the jargon term is measurement-based (feed-forward) computing. Back to the title: "using measurement as the gate" — literal, no rhetoric.

    Live Crossover: The Optical Table Shrinks into a Rack

    The hardware bottleneck of fork two is intuitive: hundreds of beam splitters laid out on an optical table, with aligning beam paths like tending a room full of porcelain. Integration is the only way forward — photonic waveguides ("light wires" etched inside a chip) shrink beam splitters and phase shifters onto a fingernail-sized chip. The site's Turing Quantum Gen3 article records exactly this step: thin-film lithium niobate chips + standard racks — photonics is also riding the chip industry's tailwind (the same script as superconducting in episode 061, with a new lead actor).

    Health Check and Reconciliation

  • Strengths: Room temperature (robust); naturally networked (photons fly); the sampling route demonstrates advantage without error correction (049); under the measurement-based architecture, "hardware" and "program" are separated most thoroughly (one pre-fabricated cluster state + a different measurement script = a different program);
  • Weaknesses: Probabilism is the Achilles' heel — single-photon sources are probabilistic (each excitation doesn't necessarily produce one photon), optical fusion gates do well if half succeed, and as circuits grow, the end-to-end success rate falls exponentially; high-performance single-photon detectors are expensive and hard (they are this route's "T gate"); generalization (large cluster states + high-success fusion) remains the hardest barrier of the whole route.
  • One-sentence positioning: the most robust qubits paired with the hardest two-qubit gate — all of photonic quantum computing's drama lies in "using architectural innovation to route around an innate weakness."

    Tearing Down the Analogy

    Where does "the welded circuit" borrow from:

  • Dissimilarity one: what's welded is the entanglement structure, not the logical order — the same cluster state, with a different measurement script, is a different program; the "hardware = circuit" intuition fails here (hardware is more like "a blank sheet of paper," the program is "the order of writing").
  • Dissimilarity two: "measurement as gate" does not contradict episode 002's "measurement collapses" — collapse still happens, only the collapse itself is choreographed into the advancement of computation; the same physical event, transformed from enemy into engine (after its second transformation in episode 035, this is the third).

Luggage to Take Home

1. Three photonic talents: robust (room temperature), flying (network messenger), interferometry equipment built in (beam splitter = H gate, the return of 021); the single, fatal shortcoming: photons ignore each other, so the two-qubit gate is innately stillborn. 2. Two forks: route around the weakness (boson sampling, "Jiuzhang" from 049) or rewrite the rules (cluster state + measurement-based computing — the choice of measurement is the gate); integration (photonic waveguide chips) is the common next step (live in the linked article). 3. The probabilistic Achilles' heel: the "chance" in sources and gates makes large circuits' success rates fall exponentially — detectors and fusion gates are this route's "T gate"; the battle for universality is not over.

Hook for the next stop: the fourth route takes "reconfigurability" to the extreme — tweezers hold atoms on a chessboard: arrange them however you like, a 200-qubit board rearranged once per second. Neutral atoms, the youngest and most imaginative player among the four routes. Episode 064: "Tweezer Chessboard: Neutral Atom Arrays."

Related tags from the original: quantum computing, from coins to error correction, photonic quantum computing, boson sampling.

Tags

#quantum-computing#photonic-quantum-computing#boson-sampling#cluster-state#measurement-based-quantum-computing#single-photon-sources#photonic-integrated-circuits#jiuzhang

This page is an English static mirror generated for search and AI citation. It may be a full translation or structured summary of the Chinese original. Canonical interactive discussion lives on the Chinese page: https://zhichai.net/topic/178636897