Dear All,
This weeks research day on THURSDAY will be at the Computer Science
Building CAB. Note the different day, time and building.
Speaker: Frederic Dupuis
Title: Secure Two-Party Quantum Evaluation of Unitaries Against Specious
Adversaries
Time: 3 pm
Place: CAB H 52
Abstract:
We show that any two-party quantum computation, specified by a unitary
which simultaneously acts on the registers of both parties, can be
securely implemented against a quantum version of classical semi-honest
adversaries that we call specious.
We first show that no statistically private protocol exists for swapping
qubits against specious adversaries. The swap functionality is modeled
by a unitary transform that is not sufficient for universal quantum
computation. It means that universality is not required in order to
obtain impossibility proofs in our model. However, the swap transform
can easily be implemented privately provided a classical bit commitment
scheme.
We provide a simple protocol for the evaluation of any unitary transform
represented by a circuit made out of gates in some standard universal
set of quantum gates. All gates except one can be implemented securely
provided one call to swap made available as an ideal functionality. For
each appearance of the remaining gate in the circuit, one call to a
classical NL-box is required for privacy. The NL-box can easily be
constructed from oblivious transfer. It follows that oblivious transfer
is universal for private evaluations of unitaries as well as for
classical circuits.
Unlike the ideal swap, NL-boxes are classical primitives and cannot be
represented by unitary transforms. It follows that, to some extent,
this remaining gate is the hard one, like the AND gate for classical
two-party computation.
Joint work with Jesper Buus Nielsen (Aarhus University) and Louis
Salvail (University of Montreal)
Cheers,
Dejan