Ph.D. Thesis - Physics
Ph.D. Thesis - Physics
Ph.D. Thesis - Physics
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Digital<br />
Analog<br />
Classical Quantum<br />
Space: 2n Time: T22n Space: n<br />
Time: Tn2 Space: 2n Space: n<br />
Time: T Time: T<br />
Table 1.1: A comparison of the resources required to implement the dynamics of digital and<br />
analog simulation of quantum systems, using both classical and quantum simulation. We<br />
consider a system of n qubits simulated for a total time T.<br />
the spatial resource requirement. In general, the time needed to simulate the system is also<br />
exponential, since an exponentiation of a 2 2n -element matrix is required to propagate the<br />
state vector forward in time. Further, the total simulation time T adds a constant factor to<br />
the total time required. For the analog case, suppose that the implementation is done using<br />
a set of voltages which specify the real and complex parts of the state vector amplitudes<br />
to the maximum possible precision. In this case, 2 n individual voltages will be required<br />
to specify the state, meaning that the spatial resource requirement is 2 n , the same as the<br />
digital case. However, the time required to perform the evolution is specified only by the<br />
total time T.<br />
Let us now consider the quantum-mechanical case. For digital quantum simulation,<br />
the space required scales as n, since each qubit in the model system may represent one<br />
qubit in the target system (and error correction adds only a polynomial number of qubits).<br />
The time required to implement the unitary evolution is proportional to n 2 , but only for<br />
Hamiltonians that can be modularly exponentiated efficiently [NC00]. The restrictions on<br />
simulable Hamiltonians are described in detail in Ref. [Llo96]. In the analog case, n model<br />
qubits again map to n target qubits, but the total simulation time is proportional only to<br />
T, as in the classical case.<br />
We summarize these results in Table 1.1, which describes the space and time resources<br />
required for implementation of the quantum dynamics, but does not include the error in the<br />
result. Although at first glance, it may seem that quantum methods are always superior<br />
to classical, and analog methods always superior to digital, this is not the case when the<br />
effects of errors are considered, or when probabilistic approaches are considered.<br />
1.4 Challenges for quantum simulation<br />
Quantum simulation is a tantalizing prospect, but there are good reasons why we don’t<br />
already have a large-scale quantum simulator. The three main reasons can be classified<br />
broadly as decoherence, precision limitations, and scalability. We discuss all of these in this<br />
section.<br />
29