Quantum Circuit Learning System for Molecular Inference
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for implementing the variational quantum eigensolver (VQE) in quantum chemistry computing, such as the quantum-classical hybrid neural network, face challenges with inference accuracy when dealing with multiple molecules or increased qubits, requiring numerous iterative calculations and being inefficient for practical applications.
Innovation Solution
A quantum circuit learning system that includes a quantum computing unit and a learning control unit, utilizing a quantum circuit with a first and second block circuit to construct and transform Hartree-Fock states, reducing the number of circuit parameters and improving convergence, thereby enhancing inference accuracy and reducing calculation costs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a parameterized quantum circuit with high expressive power is used to solve quantum chemistry computing tasks, then inference accuracy is improved in simple systems, but the number of circuit parameters increases leading to poor convergence and high calculation cost when dealing with multiple molecules or increased qubits
Solution Approach 1:
The quantum circuit is divided into multiple block circuits, where each block circuit processes a specific subset of qubits and corresponds to a specific molecule or molecular fragment. This segmentation allows the circuit to handle complex molecules by combining results from simpler block circuits, reducing the number of parameters needed for each individual block while maintaining overall inference accuracy.
Solution Approach 2:
The patent introduces a new dimension of organization by arranging block circuits in a hierarchical structure with different levels. Each block circuit operates on a specific dimension (subset of qubits), and the results are combined across dimensions to produce the final inference. This dimensional organization reduces the parameter space complexity compared to a single monolithic circuit.
2Measurement precision
If a parameterized quantum circuit with high expressive power is used, then inference accuracy is improved, but calculation cost increases due to the need for many iterative calculations
Solution Approach 1:
By segmenting the quantum circuit into block circuits that each handle specific molecules or molecular fragments, the patent reduces the number of iterative calculations required. Each block circuit can be optimized independently and processed in parallel, significantly reducing the overall calculation cost while maintaining high inference accuracy through the combination of block results.
Solution Approach 2:
The patent uses multiple copies of similar block circuit structures to handle different molecules or molecular fragments. These copied block circuits share common parameter sets and can be trained simultaneously, reducing the total computational effort compared to training a single large circuit for all molecules.
3Adaptability or versatility
If the number of qubits is increased to handle more complex molecules, then the capability to process larger molecules is improved, but the number of circuit parameters increases leading to poor convergence
Solution Approach 1:
The patent segments the qubits into multiple groups, with each group assigned to a separate block circuit. This allows the system to handle a large total number of qubits by distributing them across multiple smaller, manageable block circuits. Each block circuit maintains a reasonable number of parameters for good convergence, while the collective system can process molecules with many more qubits.
Solution Approach 2:
The block circuits are designed with universal parameter sets that can be applied across different molecules and molecular fragments. This universality allows the same block circuit structure to handle various molecular sizes and types, improving adaptability without proportionally increasing the number of parameters, as the parameters are shared across multiple applications.
Data Source
AI summary
Quantum circuit includes 1st block and 2nd block. 1st block includes gate operation layer and measurement layer. Gate operation layer includes encoding gate parameterized with encoding parameter including encoded input information for constructing 1st HF state, and transformation gate parameterized with learning parameter for transforming 1st HF state into 1st quantum state. Measurement layer outputs measurement value of 1st quantum state. 2nd block includes gate operation layer. Gate operation layer includes 2nd encoding gate parameterized with encoding parameter including encoded measurement value for constructing 2nd HF state, and transformation gate parameterized with learning parameter for transforming 2nd HF state into 2nd quantum state.