# Techniques: Securing Quantum Computations in the NISQ Era

## What the paper contributes (one paragraph)
The paper gives a blind and verifiable protocol for delegating a deterministic
quantum computation with classical inputs and outputs to an untrusted, noisy
quantum server. A client with limited quantum power keeps its data and algorithm
hidden and checks that the computation ran correctly, while tolerating genuine
device noise that would make earlier verification schemes abort. Correctness is
certified by interleaving the target computation with indistinguishable test runs
and taking a majority vote, so confidence rises exponentially close to one
without full fault tolerance: a classical repetition code replaces costly
fault-tolerant constructions. A malicious server can cheat only with
exponentially small probability, the overhead is a polynomial number of
repetitions on the same hardware, and test-failure noise up to 25 per cent
(12.5 per cent on a planar resource state) is tolerated.

## Techniques offered
- **Robust verifiable blind delegation** - delegates a deterministic classical-input,
  classical-output quantum computation to an untrusted noisy server so the input
  and algorithm stay hidden and the output is verified, without aborting on
  ordinary device noise.
  - guarantee: interactive soundness; a malicious server can cheat only with an exponentially small success probability.
  - quote: "We address this problem with an efficient, robust, blind, verifiable scheme to delegate deterministic quantum computations with classical inputs and outputs."
- **Noise-robust test-round verification** - interleaves the computation with
  indistinguishable test runs and a majority vote, amplifying confidence
  exponentially close to one using a classical repetition code instead of full
  fault tolerance.
  - guarantee: statistical bound; tolerates test-failure noise up to 25 per cent, and 12.5 per cent on a planar graph resource state.
  - quote: "in our setting, full fault-tolerance is not needed to amplify the confidence level exponentially close to 1."

## Where it could apply
- Target primitive(s): none
- Target application group(s): crypto_security, networking_edge
- Code family / hardware assumptions: none (measurement-based, planar graph resource state)

## Caveats
none