Georgia Institute of Technology
Quantum Measurement Error Mitigation In The NISQ Era Via Software And Pulse Level
Abstract
dc:description.abstractThe objective of the proposed research is to reduce the measurement error of Quantum Machines in the Noisy Intermediate-Scale Quantum (NISQ) era. Current existing quantum computers in IBM and Google have an average measurement error rate between 1-4%. Experiments show that fast and robust qubit measurement becomes critical as the size of quantum computers increases. A typical measurement contains three steps: 1) sending analog measurement pulses, 2) sampling the reflected pulses, and 3) performing software digital signal processing. In the last step, software discriminators classify the state of a qubit as state |0⟩ or |1⟩. In the most popular software discriminators, each qubit is trained individually and thus no correlation information such as qubit-qubit crosstalk effect is considered, so the they becomes noisy when the system is large or the circuit is complex. Although recently full-system discriminators (FSDs), which use support vector machines and neural networks to classify all the qubits on the quantum machine, have been proposed and achieved higher accuracy than traditional discriminators, it is hard for these methods to scale to large systems because their training and evaluating cost grows exponentially. To address the above problems, we propose Wish-Disc or Wish Discriminator that sequentially uses per-qubit discriminators (PQD) and group-discriminators (GD) that contains small groups of qubits. The probabilistic nature of the qubits has been utilized to further improve the accuracy with a little cost of time. Wish-Disc first tries using fast but less accurate PQDs, and if the confidence of the PQDs result is less than a confidence threshold, then the slower but more accurate group-discriminators will be used. By smartly partitioning the qubits into several small groups, properly training the discriminators, and selecting the probabilistic thresholds, our Wish-Disc can outperform the state-of-the-art algorithms while maintaining scalability. Finally, the pulse level measurement can improve the measurement fidelity by using a dedicated measurement pulse of each qubit.
Degree
thesis:*- Level thesis:degree_level
- Masters
- Department dc:contributor.department
- Electrical and Computer Engineering
- Grantor dc:publisher
- Georgia Institute of Technology
- Year dc:date.issued
- 2022
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Wen, Jiahao
- Advisor dc:contributor.advisor
-
- Qureshi, Moinuddin K.
- Committee members dc:contributor.committeemember
-
- Khan, Asif I.
- Raychowdhury, Arijit
Subjects
dc:subject × 3Rights
- Language dc:language.iso
- en_US
Identifiers
dc:identifier.*- Handle dc:identifier.uri
- http://hdl.handle.net/1853/67217