Quantum error correction with realistic measurement data

Patent metadata

Field	Value
Publication number	US-11521104-B2
Application number	US-202117179625-A
Country	US
Kind code	B2
Filing date	Feb 19, 2021
Priority date	Feb 19, 2021
Publication date	Dec 6, 2022
Grant date	Dec 6, 2022

Abstract

A quantum computing system computes soft information quantifying an effect of soft noise on multiple rounds of a syndrome measurement that is output by a quantum measurement circuit. The soft noise arises due to imperfections in a readout device that introduce variability in repeated measurements of ancilla qubits and is distinct from quantum noise arising from bit-flips in data qubits that are indirectly measured by the ancilla qubits. The quantum computing system applying decoding logic to identify fault locations within the quantum measurement circuit based on the computed soft information.

First claim

What is claimed is:

1. A system comprising:
a soft information computation engine that:
receives multiple rounds of a syndrome measurement output by a quantum measurement circuit, the syndrome measurement including a soft value measured for each bit of a plurality of bits in a quantum circuit, the soft value measured for each of the bits being either of a first range corresponding to a hard outcome indicative of non-detection of a fault at a corresponding circuit location or of a second range corresponding to a hard outcome indicative of at least one fault detected at the corresponding circuit location;
computes soft information based on the soft values for each of the bits, the soft information indicating a confidence in the hard outcome associated with the bit and also quantifying an effect of soft noise on the multiple rounds of a syndrome measurement output by a quantum measurement circuit, the soft noise introducing variability in repeated measurements of ancilla qubits due to at least one of imperfections in a readout device and limited measurement time for the repeated measurements; and
a decoding unit that uses the computed soft information to identify fault locations that collectively explain the syndrome measurement output, the soft information biasing the decoding unit toward selecting solution sets including fault locations that correspond to a subset of the bits characterized by low-confidence hard outcomes in the syndrome measurement.

2. The system of claim 1 , wherein the soft information computing engine is further configured to:
receive, from the quantum measurement circuit, multiple rounds of a soft outcome vector, the soft outcome vector containing real number values representing measurements of a plurality of syndrome bits, each of the syndrome bits providing information about one or more errors affecting a quantum measurement;
generate a decoding graph defining nodes corresponding to the plurality of syndrome bits, the nodes being connected to one another by horizontal edges and vertical edges; and
compute a set of soft edge weights, each of the soft edge weights corresponding to one of the vertical edges in the decoding graph and being based on a real number measurement value of a syndrome bit corresponding to endpoints of the vertical edge.

3. The system of claim 2 , wherein the decoding unit is further configured to:
determine a minimum weight solution for the decoding graph based on the computed set of soft edge weights.

4. The system of claim 2 , wherein the decoding unit uses the soft edge weights as edge weights for vertical edges of the decoding graph.

5. The system of claim 2 , wherein the soft information computing engine computes the set of soft edge weights based on the real number measurement value of each of the syndrome bits and also based on a hard outcome value for each of the syndrome bits.

6. The system of claim 1 , wherein the soft noise is distinct from quantum noise arising from bit-flips in data qubits that are indirectly measured by the ancilla qubits.

7. The system of claim 1 , wherein the decoding unit implements logic of at least one of a union find (UF) decoder and a minimum weight perfect matching (MWPM) decoder.

8. A method comprising:
receiving multiple rounds of a syndrome measurement output by a quantum measurement circuit, the syndrome measurement including a soft value measured for each bit of a plurality of bits in a quantum circuit, the soft value measured for each of the bits being either of a first range corresponding to a hard outcome indicative of non-detection of a fault at a corresponding circuit location or of a second range corresponding to a hard outcome indicative of at least one fault detected at the corresponding circuit location;
computing soft information based on the soft values for each of the bits, the soft information indicating a confidence in the hard outcome associated with the bit and also quantifying an effect of soft noise on the multiple rounds of a syndrome measurement output by a quantum measurement circuit, the soft noise introducing variability in repeated measurements of ancilla qubits due to at least one of limited measurement time and imperfections in a readout device, the soft noise being distinct from quantum noise arising from bit-flips in data qubits that are indirectly measured by the ancilla qubits; and
identifying fault locations within the quantum measurement circuit based on the computed soft information, the identified fault locations collectively explaining the syndrome measurement output, the soft information biasing a decoding unit toward selecting solution sets including fault locations that correspond to a subset of the bits characterized by low-confidence hard outcomes in the syndrome measurement.

9. The method of claim 8 , wherein computing the soft information further comprises:
receiving, from the quantum measurement circuit, multiple rounds of a soft outcome vector, the soft outcome vector containing real number values representing measurements of a plurality of syndrome bits providing soft information errors affecting a quantum measurement;
generating a decoding graph defining nodes corresponding to the plurality of syndrome bits, the nodes being connected to one another by horizontal edges and vertical edges; and
computing a set of soft edge weights, each of the soft edge weights corresponding to one of the vertical edges in the decoding graph and being based on real number measurement values of a syndrome bit corresponding to endpoints of the vertical edge.

10. The method of claim 8 , wherein identifying fault locations further comprises:
determining a minimum weight solution for a decoding graph based on the computed soft information.

11. The method of claim 10 , wherein the decoding graph is a three-dimensional graph that includes multiple layers of a 2D grid having nodes that correspond to measurement locations of the syndrome bits within the quantum measurement circuit, the multiple layers of the 2D grid being separated by vertical edges each representing a time step between repeated measurements of the nodes within the 2D grid.

12. The method of claim 10 , wherein determining the minimum weight solution further comprises:
using the computed set of soft edge weights as edge weights for corresponding vertical edges of the decoding graph.

13. The method of claim 10 , wherein determining the minimum weight solution further comprises:
deriving edge weights for vertical edges of the decoding graph, each of the edge weights being based on a soft edge weight of the computed soft edge weights and a further based on a computed hard edge weight.

14. The method of claim 10 , wherein determining the minimum weight solution further comprises:
building a distance graph with at least one of a union find (UF) decoder and a minimum weight perfect matching (MWPM) decoder.

15. One or more non-transitory computer-readable storage media encoding computer executable instructions for executing a computer process comprising:
receiving multiple rounds of a syndrome measurement output by a quantum measurement circuit, the syndrome measurement including a soft value measured for each bit of a plurality of bits in a quantum circuit, the soft value measured for each of the bits being either of a first range corresponding to a hard outcome indicative of non-detection of a fault at a corresponding circuit location or of a second range corresponding to a hard outcome indicative of at least one fault detected at the corresponding circuit location;
computing soft information based on the soft values for each of the bits

Assignees

• Microsoft Technology Licensing Llc
• Microsoft Licensing Tech Llc

Inventors

• Delfosse Nicolas Guillaume
• Pattison Christopher Anand
• Beverland Michael
• Da Silva Marcus Palmer

Classifications

• G06F16/9024

  Graphs; Linked lists (G06F16/9027 takes precedence) · CPC title

• G06F11/1048

  using arrangements adapted for a specific error detection or correction feature · CPC title

• G06N10/70 Primary

  Quantum error correction, detection or prevention, e.g. surface codes or magic state distillation · CPC title

• G06N10/00 Primary

  Quantum computing, i.e. information processing based on quantum-mechanical phenomena · CPC title