Quantum Circuit Noise Learning Layers for Error Mitigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Quantum error mitigation using learned noise models is limited to quantum circuits with a small set of unique quantum gate layers, and the noise properties of noisy quantum computers drift rapidly, making it difficult to learn noise models quickly enough to reflect current device conditions, leading to a mismatch between actual noise characteristics and the noise model.
Innovation Solution
Divide each target layer of a quantum circuit into sub-layers, group them into a reduced set of learning layers, learn noise models for each sub-layer based on gate crosstalk, and combine these models to form a complete set of noise models for the target layers, leveraging low device crosstalk to reduce the number of learning experiments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If noise models are learned for each unique layer of a quantum circuit, then measurement precision is improved, but loss of time increases because learning each layer takes too long and noise drifts before models are complete
Solution Approach 1:
The patent divides each unique layer into multiple sub-layers (e.g., based on qubit groups or gate types). Instead of learning one noise model for the entire layer, separate noise models are learned for each sub-layer. This segmentation reduces the complexity and time required for learning each individual model while maintaining overall accuracy through composition of the sub-layer models.
Solution Approach 2:
The patent merges multiple sub-layer noise models to form a complete layer noise model. By learning smaller sub-layer models that can be combined, the system achieves the same overall accuracy as learning a full layer model but in significantly less time, since the sub-layers can be learned in parallel and require fewer parameters each.
2Productivity
If noise models are learned quickly for circuits with many unique layers, then productivity is improved, but measurement precision deteriorates because there isn't enough time to learn each layer accurately before noise drift
Solution Approach 1:
By segmenting layers into sub-layers, the patent enables faster learning of individual models that can be performed in parallel, improving productivity while maintaining the ability to achieve accurate representations through composition of multiple sub-layer models.
Solution Approach 2:
The patent learns noise models for sub-layers individually rather than requiring complete layer models before use. This partial action approach allows the system to progressively build up accurate noise characterizations without waiting for complete layer learning, improving productivity while maintaining precision through incremental model composition.
3Measurement precision
If each unique layer has its own noise model, then measurement precision is improved, but device complexity increases making the system harder to manage and update
Solution Approach 1:
The patent segments layers into sub-layers with potentially reusable noise models. Sub-layers that are identical or similar across different layers can share the same noise model, reducing the total number of unique models to manage while still providing accurate layer-specific characterizations when needed.
Solution Approach 2:
The patent creates noise models for sub-layers that can serve multiple purposes and be reused across different layers. A sub-layer noise model learned for one layer can be applied to other layers containing the same sub-layer structure, reducing device complexity through model reuse while maintaining measurement precision through appropriate model selection and composition.
Data Source
AI summary
A method, system, and computer program product for learning noise models to perform quantum error mitigation. Each target layer of a quantum circuit is divided into a set of sub-layers. Each of the sub-layers for each target layer of the quantum circuit is grouped into a reduced set of learning layers, which enables each sub-layer's noise model to be learned from fewer layers (learning layers). A learning layer refers to a layer that is used in combination with other learning layers to form the minimally complete layer set for learning all the layer components used in the quantum circuit. The noise models for each of the sub-layers are then learned on the reduced set of learning layers. Such learned noise models are combined to form a complete set of noise models for the target layers of the quantum circuit and used to perform quantum error mitigation on the quantum circuit.