Quantum Warm Starts for Faster Classical Optimisation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Classical optimisation algorithms outperform quantum approaches in efficiency due to the limitations of NISQ quantum computers, necessitating bespoke classical algorithms to correct quantum outputs, which are excessively demanding on quantum accuracy and fail to leverage existing classical algorithms effectively.
Innovation Solution
Utilize approximate quantum solutions as warm starts for classical algorithms, leveraging the quantum process's ability to probe solution spaces and encode valuable information, thereby improving the efficiency of classical algorithms by providing better initial inputs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If quantum computers are used to solve optimisation problems directly, then quantum computational power is leveraged, but the limited capabilities of NISQ hardware and excessive quantum accuracy requirements reduce effectiveness
Solution Approach 1:
The patent introduces a classical algorithm as an intermediary between the quantum computer and the final solution. The quantum computer generates approximate solutions that are then refined by a classical algorithm, allowing the system to leverage quantum computational power while avoiding the reliability limitations of NISQ hardware. This mediator approach enables both quantum and classical systems to contribute their strengths.
Solution Approach 2:
The patent segments the optimisation problem-solving process into two distinct phases: a quantum phase that generates approximate solutions and a classical phase that refines these solutions. This segmentation allows each system to operate within its capabilities - quantum for exploration and classical for exploitation - thereby resolving the contradiction between leveraging quantum power and managing accuracy limitations.
2Manufacturing precision
If bespoke classical algorithms are used to correct quantum outputs, then quantum results are improved, but the algorithms are highly specific and fail to leverage existing classical algorithms effectively
Solution Approach 1:
The patent employs a universal classical algorithm framework that can handle multiple types of optimisation problems and work with outputs from different quantum algorithms. Rather than creating bespoke correction algorithms for each specific case, the patent uses a general-purpose classical optimisation routine that can refine quantum outputs across various problem domains, thereby reducing algorithmic complexity while maintaining precision.
Solution Approach 2:
The patent adjusts parameters of existing classical algorithms to work effectively with quantum outputs. By modifying convergence criteria, initialization parameters, or search strategies of standard classical algorithms, the system achieves effective refinement of quantum solutions without requiring completely new bespoke algorithms, thus reducing complexity while maintaining accuracy.
3Measurement precision
If quantum computers provide complete solutions, then accuracy is maximized, but the process is excessively demanding on quantum accuracy beyond NISQ capabilities
Solution Approach 1:
The patent applies partial action by having the quantum computer perform only the portion of the computation it is suited for - generating approximate solutions - rather than attempting to solve the entire optimisation problem. The remaining refinement work is delegated to classical algorithms, thereby reducing the harmful factor of excessive quantum accuracy demands while still achieving high overall solution accuracy through the combined approach.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention relates to methods and apparatuses for improving the efficiency of solving optimisation problems. The invention includes the use of a quantum computer to provide a warm start to a known optimisation algorithm, thereby resulting in the known algorithm starting from a better (more promising) location in the solution space. This better starting point allows the known algorithm to converge on a good (i.e. toward an optimum) solution quicker, e.g. in fewer steps, and to reach better solutions sooner than known methods involving cold starts.