Hybrid Eigenstate Search for Complete TTNS Extremal States
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing algorithms for determining extremal eigenstates of tensor network matrices are either computationally expensive or prone to omitting states, especially when initial states are close to other eigenstates, leading to inefficiencies in eigenproblem calculations.
Innovation Solution
A hybrid eigenstate determination algorithm combining a limited multistate optimization algorithm and a single-state optimization algorithm, which iteratively optimizes a subset of estimated extremal eigenstates using orthogonality constraints to converge on the desired eigenstates of a tree tensor network operator.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If a limited multistate optimization algorithm is used to rapidly determine extremal eigenstates, then computational speed is improved, but eigenstates may be omitted when initial states are close to other eigenstates
Solution Approach 1:
The algorithm incorporates a feedback mechanism where the limited multistate optimization is followed by a single-state optimization step that uses the previously determined eigenstates as a basis. This feedback loop allows the system to identify and correct omitted eigenstates by checking orthogonality conditions and refining the state set iteratively, thus maintaining reliability while preserving computational speed.
Solution Approach 2:
The algorithm dynamically adjusts the number of states optimized in each iteration based on convergence criteria and orthogonality checks. By adapting the optimization scope during execution rather than using a fixed approach, the system can rapidly converge when few states are needed while automatically expanding the search when eigenstates are at risk of being omitted, balancing speed and reliability.
2Reliability
If a single-state optimization algorithm is used to determine a large set of extremal eigenstates without omitting states, then completeness is improved, but computational cost and time increase significantly
Solution Approach 1:
The algorithm segments the eigenstate determination process into two distinct phases: a limited multistate optimization phase that rapidly identifies a subset of eigenstates, and a single-state optimization phase that refines and completes the set. This segmentation allows the system to leverage the speed of multistate methods for the bulk of the computation while using single-state methods only where necessary, dramatically reducing overall computational time while maintaining completeness.
Solution Approach 2:
The algorithm performs partial optimization by determining only a subset of eigenstates through the limited multistate approach, then completes the remaining states through targeted single-state optimization. This partial action strategy avoids the excessive computational cost of applying single-state optimization to all states from the beginning, reducing computational time while ensuring no eigenstates are omitted through the completion phase.
3Productivity
If algorithms optimize multiple eigenstates simultaneously, then productivity is improved, but complexity of the algorithm increases
Solution Approach 1:
The algorithm segments the complex multistate optimization into two simpler, sequential processes: limited multistate optimization followed by single-state optimization. Each segment has reduced complexity compared to a full simultaneous multistate optimization, making the overall algorithm more manageable and easier to implement while maintaining high productivity through the combined effect of both segments.
Data Source
AI summary
One example includes a hybrid eigenstate determination algorithm. The hybrid eigenstate determination algorithm includes a limited multistate optimization algorithm configured to determine a state set comprising estimated extremal eigenstates of a bundled tree tensor network state (TTNS) based on predefined algorithm parameters. The bundled TTNS can be associated with a quantity of extremal eigenstates of a tree tensor network operator (TTNO) to be determined. The hybrid eigenstate determination algorithm also includes a single-state optimization algorithm configured to select at least one estimated extremal eigenstate of the determined state set and to sequentially optimize the selected at least one estimated extremal eigenstate of the state set to convergence to determine a respective at least one of the extremal eigenstates of the TTNO.