We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Quantum Physics

Title: Unitarization Through Approximate Basis

Authors: Joshua Cook
Abstract: We introduce the problem of unitarization. Unitarization is the problem of taking $k$ input quantum circuits that produce orthogonal states from the all $0$ state, and create an output circuit implementing a unitary with its first $k$ columns as those states. That is, the output circuit takes the $k$th computational basis state to the state prepared by the $k$th input circuit. We allow the output circuit to use ancilla qubits initialized to $0$. But ancilla qubits must always be returned to $0$ for any input. The input circuits may use ancilla qubits, but we are only guaranteed the they return ancilla qubits to $0$ on the all $0$ input.
The unitarization problem seems hard if the output states are neither orthogonal to or in the span of the computational basis states that need to map to them. In this work, we approximately solve this problem in the case where input circuits are given as black box oracles by probably finding an approximate basis for our states. This method may be more interesting than the application. This technique is a sort of quantum analogue of Gram-Schmidt orthogonalization for quantum states.
Specifically, we find an approximate basis in polynomial time for the following parameters. Take any natural $n$, $k = O\left(\frac{\ln(n)}{\ln(\ln(n))}\right)$, and $\epsilon = 2^{-O(\sqrt{\ln(n)})}$. Take any $k$ input quantum states, $(|\psi_i \rangle)_{i\in [k]}$, on polynomial in $n$ qubits prepared by quantum oracles, $(V_i)_{i \in [k]}$ (that we can control call and control invert). Then there is a quantum circuit with polynomial size in $n$ with access to the oracles $(V_i)_{i \in [k]}$ that with at least $1 - \epsilon$ probability, computes at most $k$ circuits with size polynomial in $n$ and oracle access to $(V_i)_{i \in [k]}$ that $\epsilon$ approximately computes an $\epsilon$ approximate orthonormal basis for $(|\psi_i \rangle)_{i\in [k]}$.
Comments: Review Significantly improves presentation of results, adds more details
Subjects: Quantum Physics (quant-ph); Computational Complexity (cs.CC)
Cite as: arXiv:2104.00785 [quant-ph]
  (or arXiv:2104.00785v2 [quant-ph] for this version)

Submission history

From: Joshua Cook [view email]
[v1] Thu, 1 Apr 2021 22:11:05 GMT (19kb)
[v2] Mon, 13 Sep 2021 22:45:21 GMT (30kb)

Link back to: arXiv, form interface, contact.