Abstract
In network science, graphs obtained by taking the Kronecker or tensor power of the adjacency matrix of an initiator graph are used to construct complex networks. In this paper, we analytically prove sufficient conditions under which such Kronecker graphs can be searched by a continuous-time quantum walk in optimal ΘN time. First, if the initiator is regular and its adjacency matrix has a dominant principal eigenvalue, meaning its unique largest eigenvalue asymptotically dominates the other eigenvalues in magnitude, then the Kronecker graphs generated by this initiator can be quantum searched with probability 1 in πN/2 time, asymptotically, and we give the critical jumping rate of the walk that enables this. Second, for any fixed initiator that is regular, nonbipartite, and connected, the Kronecker graphs generated by it are quantum searched in ΘN time. This greatly extends the number of Kronecker graphs on which quantum walks are known to optimally search. If the fixed, regular, connected initiator is bipartite, however, then search on its Kronecker powers is not optimal, but is still better than a classical computer's O(N) runtime if the initiator has more than two vertices.
| Original language | English (US) |
|---|---|
| Article number | 062334 |
| Journal | Physical Review A |
| Volume | 98 |
| Issue number | 6 |
| DOIs | |
| State | Published - Dec 27 2018 |
| Externally published | Yes |
Fingerprint
All Science Journal Classification (ASJC) codes
- Atomic and Molecular Physics, and Optics
Cite this
Optimal quantum-walk search on Kronecker graphs with dominant or fixed regular initiators. / Glos, Adam; Wong, Thomas.
In: Physical Review A, Vol. 98, No. 6, 062334, 27.12.2018.Research output: Contribution to journal › Article
}
TY - JOUR
T1 - Optimal quantum-walk search on Kronecker graphs with dominant or fixed regular initiators
AU - Glos, Adam
AU - Wong, Thomas
PY - 2018/12/27
Y1 - 2018/12/27
N2 - In network science, graphs obtained by taking the Kronecker or tensor power of the adjacency matrix of an initiator graph are used to construct complex networks. In this paper, we analytically prove sufficient conditions under which such Kronecker graphs can be searched by a continuous-time quantum walk in optimal ΘN time. First, if the initiator is regular and its adjacency matrix has a dominant principal eigenvalue, meaning its unique largest eigenvalue asymptotically dominates the other eigenvalues in magnitude, then the Kronecker graphs generated by this initiator can be quantum searched with probability 1 in πN/2 time, asymptotically, and we give the critical jumping rate of the walk that enables this. Second, for any fixed initiator that is regular, nonbipartite, and connected, the Kronecker graphs generated by it are quantum searched in ΘN time. This greatly extends the number of Kronecker graphs on which quantum walks are known to optimally search. If the fixed, regular, connected initiator is bipartite, however, then search on its Kronecker powers is not optimal, but is still better than a classical computer's O(N) runtime if the initiator has more than two vertices.
AB - In network science, graphs obtained by taking the Kronecker or tensor power of the adjacency matrix of an initiator graph are used to construct complex networks. In this paper, we analytically prove sufficient conditions under which such Kronecker graphs can be searched by a continuous-time quantum walk in optimal ΘN time. First, if the initiator is regular and its adjacency matrix has a dominant principal eigenvalue, meaning its unique largest eigenvalue asymptotically dominates the other eigenvalues in magnitude, then the Kronecker graphs generated by this initiator can be quantum searched with probability 1 in πN/2 time, asymptotically, and we give the critical jumping rate of the walk that enables this. Second, for any fixed initiator that is regular, nonbipartite, and connected, the Kronecker graphs generated by it are quantum searched in ΘN time. This greatly extends the number of Kronecker graphs on which quantum walks are known to optimally search. If the fixed, regular, connected initiator is bipartite, however, then search on its Kronecker powers is not optimal, but is still better than a classical computer's O(N) runtime if the initiator has more than two vertices.
UR - http://www.scopus.com/inward/record.url?scp=85059439924&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85059439924&partnerID=8YFLogxK
U2 - 10.1103/PhysRevA.98.062334
DO - 10.1103/PhysRevA.98.062334
M3 - Article
VL - 98
JO - Physical Review A
JF - Physical Review A
SN - 2469-9926
IS - 6
M1 - 062334
ER -