Ising Computing Device Initial State Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing Ising computing devices face challenges in efficiently calculating initial energy and local fields, leading to high computational overhead and difficulties in improving processing performance for combinatorial optimization problems.
Innovation Solution
The proposed optimization apparatus and method utilize an Ising computing device with an initial setting control unit that generates desired initial states by setting spin states to expected values, allowing for parallel calculation of local fields and reducing computational overhead.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the Ising computing device calculates initial energy and local fields sequentially using traditional methods, then calculation accuracy is maintained, but computational overhead increases and processing performance deteriorates
Solution Approach 1:
The patent divides the calculation of initial energy and local fields into independent parallel segments. Each spin's local field and energy contribution can be calculated independently and simultaneously, rather than sequentially. This segmentation enables parallel processing architecture that reduces computational overhead while maintaining accuracy.
Solution Approach 2:
The patent performs preliminary calculation of spin states and their contributions to energy and local fields before the main optimization process. By pre-computing these values and storing them in appropriate registers, the system avoids redundant calculations during the optimization iterations, significantly reducing computational overhead.
2Productivity
If the Ising computing device starts from arbitrary initial states, then solution diversity is improved, but the time to reach lowest energy state increases
Solution Approach 1:
The patent implements an initial setting control unit that prepares desired initial states before the optimization process begins. This preliminary action allows the system to start from strategically chosen states that balance solution diversity with convergence speed, rather than using arbitrary or purely random initial states.
Solution Approach 2:
The patent enables dynamic adjustment of initial state parameters through the initial setting control unit. By controlling parameters such as spin state distributions and energy levels in the initial configuration, the system can optimize the trade-off between exploration (solution diversity) and exploitation (convergence speed) based on the specific problem characteristics.
Data Source
Figure 1
Figure 2
Figure 3A
AI summary
An optimization apparatus includes: a temperature control unit that controls a temperature value indicating a temperature; an energy change amount calculation unit that calculates a change amount of energy represented by an evaluation function in a case where a state transition is performed by changing a state of any one of a plurality of state variables included in the evaluation function representing the energy; a determination unit that stochastically determines whether or not to accept the state transition based on a correlation between the change amount of the energy and a threshold calculated based on the temperature value and a random number value; an expected value holding unit that holds an expected value of each of the states of the plurality of state variables; an expected value comparison unit that compares each of the expected values held by the expected value holding unit with the corresponding one of the values of the states of the state variables and extracts each unequal state variable; a confirmation unit that selects the state variable extracted by the expected value comparison unit and changes the state of the selected state variable until the values of the states of the state variables are all equal to the expected values, and selects the state variable for which the state transition is accepted by the determination unit and changes the state of the selected state variable after the values of the states of the state variables are once equal to the expected values; an energy calculation unit that calculates post-transition energy after the state of the state variable selected by the confirmation unit is changed; and a search unit that sets the post-transition energy as lowest energy in a case where the post-transition energy calculated by the energy calculation unit is less than the lowest energy.