Errors in scalable quantum Computers (Englisch)

A functional quantum computer potentially outperforms any classical machine exponentially in a number of important computational tasks. Therefore, its physical implementation has to scale efficiently in the number of qubits, specifically in tasks such as treatment of external error sources. Due to the intrinsic complexity and limited accessibility of quantum systems, the validation of quantum gates is fundamentally difficult. Randomized Benchmarking is a protocol to efficiently assess the average fidelity of only Clifford group gates. In this thesis we present a hybrid of Randomized Benchmarking and Monte Carlo sampling for the validation of arbitrary gates. It improves upon the efficiency of current methods while preserving error amplification and robustness against imperfect measurement, but is still exponentially hard. To achieve polynomial scaling, we introduce a symmetry benchmarking protocol that validates the conservation of inherent symmetries in quantum algorithms instead of gate fidelities. Adiabatic quantum computing is believed to be more robust against environmental effects, which we investigate in the typical regime of a scalable quantum computer using renormalization group theory. We show that a k-local Hamiltonian is in fact robust against environmental influence but multipartite entanglement is limited to combined system-bath state which we conclude to result in a more classical behavior more susceptible to thermal noise. ; Ein Quantencomputer wäre in einer Reihe wichtiger Berechnungen exponenziell effizienter als klassische Computer, unter Vorraussetzung einer fehlerarmen und skalierbaren Implementierung. Aufgrund der intrinsischen Komplexität und beschränkten Auslesbarkeit von Quantensystemen ist die Validierung von Quantengattern ungleich schwerer als die klassischer. Das Randomized Benchmarking Protokoll leistet dies effizient, ist jedoch beschränkt auf Cliffordgatter. In dieser Arbeit präsentieren wir ein Hybridprotokoll aus Interleaved Randomized Benchmarking und Monte Carlo Sampling zur ...