Metropolis Algorithm Circuit for Ising Model Ground State Search
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for searching the ground state of the Ising model, such as Markov chain Monte Carlo (MCMC) and simulated annealing, require complex calculations and large physical quantities, making it difficult to implement in a limited circuit area, especially for large-scale apparatuses.
Innovation Solution
An information processing apparatus with array circuits that include units with memory for storing node states and coupling coefficients, and a logic circuit that determines subsequent states using a temperature schedule and random variables following an exponential distribution, reducing the physical quantity of the circuit required.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If complex calculations are performed using Markov chain Monte Carlo method for ground state search, then calculation accuracy is improved, but circuit area and physical quantity increase
Solution Approach 1:
The patent divides the calculation process into discrete time steps and segments the state transition logic into modular units. Each unit handles specific operations (energy calculation, random number generation, state update) that can be implemented with simple logic circuits rather than complex sequential calculations, thereby reducing circuit area while maintaining accuracy.
Solution Approach 2:
The patent replaces complex sequential mechanical calculation systems with parallel stochastic processes. By using random number generators and simple logic circuits to perform state transitions according to Metropolis criteria, the system achieves accurate ground state search without requiring large circuits for complex arithmetic operations.
2Productivity
If parallel processing is implemented for high-speed calculation, then productivity is improved, but device complexity increases
Solution Approach 1:
The patent segments the parallel processing system into identical, independent calculation units that can operate simultaneously. Each unit handles a specific spin or node in the Ising model, and all units follow the same simple logic structure, enabling parallel execution without requiring complex inter-unit coordination circuits.
Solution Approach 2:
The patent designs universal calculation units that can handle multiple functions (energy calculation, random number generation, state transition) within a single standardized module. This universality allows parallel processing to be achieved by replicating the same simple unit rather than designing complex specialized circuits for each function.
3Measurement precision
If temperature schedule is applied for simulated annealing, then solution quality is improved, but loss of time increases
Solution Approach 1:
The patent implements periodic temperature updates in a systematic schedule, where the temperature parameter is adjusted at regular intervals during the annealing process. This periodic action allows the system to maintain high solution quality by properly cooling the system while preventing excessive calculation time through structured, predictable temperature changes rather than continuous adjustment.
Data Source
AI summary
It is possible to perform a stochastic process based on a metropolis algorithm while reducing a physical quantity of a circuit. Provided is an information processing apparatus including one or a plurality of array circuits. In this apparatus, each of the array circuits includes a plurality of units, and each of the plurality of units includes a first memory that stores a value indicating a state of one node of a coupling model, a second memory that stores a coupling coefficient indicating coupling from a node of another unit connected to an unit of the second memory, and a logic circuit that determines a value indicating a subsequent state of the one node based on a value indicating a state of the node of the other unit and the coupling coefficient. Further, the logic circuit sets a first random variable in accordance with an exponential distribution of a parameter θ as an input.