Quantum Decision Trees and Semidefinite Programming.
- Michael
- Mario
We reformulate the notion of quantum query complexity in terms of inequalities and equations for a set of positive matrices, which we view as a quantum analogue of a decision tree. Using the new formulation we show that: 1. every quantum query algorithm needs to use at most n quantum bits in addition to the query register. 2. For any function f there is an algorithm that runs in polynomial time in terms the truth table of f and (for {var_epsilon} > 0) computes the {var_epsilon}-error quantum decision tree complexity of f. 3. Using the dual of our system we can treat lower bound methods on a uniform platform, which paves the way to their future comparison. In particular we describe Ambainis's bound in our framework. 4. The output condition on quantum algorithms used by Ambainis and others is not sufficient for an algorithm to compute a function with {var_epsilon}-bounded error: we show the existence of algorithms whose final entanglement matrix satisfies the condition, but for which the value of f cannot be determined from a quantum measurement on the accessible part of the computer.
- Research Organization:
- Los Alamos National Laboratory
- Sponsoring Organization:
- DOE
- OSTI ID:
- 975874
- Report Number(s):
- LA-UR-01-6417
- Country of Publication:
- United States
- Language:
- English
Similar Records
Local induction of decision trees: Towards interactive data mining
Finding cycles and trees in sublinear time.