298 results about "Combinatorial optimization problem" patented technology

The optimization device 90 includes a determining means 91 and an optimizing means 92. The determining means 91 determines a type of combinatorial optimization problem from a QUBO matrix obtained by QUBO modeling of a combinatorial optimization problem that includes a two-way one-hot condition as a constraint condition. The optimizing means 92 performs optimization process according to the determined type of combinatorial optimization problem.

The invention discloses an intelligent path planning method and system for an unmanned aerial vehicle. The method comprises the following steps: initializing related parameters and optimization algorithm parameters of the unmanned aerial vehicle; coding the path points of the unmanned aerial vehicle; calculating fitness values of all individuals in the population; generating a new population through selection, crossover and mutation operations; carrying out local search on individuals with better fitness in the population; dynamically adjusting algorithm parameters according to population fitness changes; and outputting the path with the highest fitness value as the optimal flight path of the unmanned aerial vehicle. According to the method, a local search mechanism and a dynamic adjustment strategy are introduced on the basis of a traditional genetic algorithm, the method is suitable for a combinatorial optimization problem, excellent performance is shown in unmanned aerial vehicle path planning, and a path with the shortest time can be efficiently found.

The invention discloses a power system unit commitment model solving method based on an improved simulated annealing algorithm, and the method comprises the steps: carrying out the optimization solving of a model objective function through employing the improved simulated annealing algorithm on the basis of a standard simulated annealing algorithm, and carrying out the optimization solving of a model objective function through adaptive temperature control and a multi-starting point strategy. And the global search capability and the convergence speed are improved when the unit commitment optimization problem of the power system is solved. Through the improvement of the invention, the algorithm can more effectively avoid falling into local optimum during solving, and the solving process is accelerated.

The invention relates to the technical field of deep learning model training optimization, in particular to a fine-grained resource scheduling method and system in DNN model parallel training, and the method comprises the steps: converting a DNN model into calculation graph representation, and constructing a calculation graph in combination with the topological information of calculation equipment; modeling calculation stage division in a DNN model training process and resource mapping from each calculation stage to calculation equipment as a stage division and resource mapping combined optimization problem; and iteratively solving the combinatorial optimization problem by using a heuristic algorithm, and optimizing the computing equipment resource mapping of each computing stage by dynamically adjusting the weight of constraint conditions and minimizing the overhead and load difference of the computing equipment. According to the method, the calculation cost, the communication overhead, the overall load balancing and the solving time are comprehensively considered in the target function, efficient calculation resource allocation and task scheduling optimization are achieved, the training and reasoning efficiency of a large-scale deep learning model is improved, and the method has a good application prospect in the field of deep neural network distributed parallel training.

The invention provides a method, a device and a system for solving a combinatorial optimization problem. The method comprises the steps of receiving a combination optimization problem input by a user; mapping the combinatorial optimization problem into a model; the model comprises a complex variable to be solved and a target function; the complex variables to be solved are used for describing solutions, meeting the requirements, of permutation and combination of things as discrete variables in a complex field; the objective function is used for describing permutation and combination of a plurality of things under a given constraint condition; determining a kinetic equation of the model in a complex field, wherein the kinetic equation is used for describing a time-based evolution process of a complex variable to be solved; the kinetic equation at least comprises a control variable, a multi-bifurcation parameter and an action parameter; solving the kinetic equation, and changing the value of the control variable to obtain q stable phases of the complex variable to be solved; the value of q is determined by a multi-bifurcation parameter; and obtaining a solution of the combinatorial optimization problem according to the q stable phases. The spin scale can be reduced and the solving time can be shortened when the combinatorial optimization problem is solved.

The embodiment of the invention discloses a quantum computer, a method and a related device for solving a combinatorial optimization problem, and the quantum computer comprises a measurement and control all-in-one machine which is configured to generate a control signal based on a constraint condition, the constraint condition is used for representing the constraint when a plurality of elements are combined; the quantum chip comprises at least one group of quantum bits, and the number of the quantum bits is the same as that of the elements; and the interface is connected with the measurement and control all-in-one machine and the quantum chip and is configured to control the quantum chip based on the control signal so as to generate an optimal combination of elements represented by the ground state of the group of quantum bits. By adopting the embodiment of the invention, the requirement of quantum bits can be effectively reduced, the method is suitable for a real quantum computer, and a practical and feasible quantum circuit can be operated to solve a combinatorial optimization problem.

The invention discloses a quantum Isin model construction method for a security constraint unit commitment optimization problem, and relates to the field of quantum computation.The quantum Isin model construction method comprises the steps that a security constraint unit commitment optimization model is constructed, and parameters of a mixed integer programming problem are obtained; using a Benders decomposition method to decompose a mixed integer programming problem into a main problem and a sub-problem; substituting the optimal binary solution to solve the sub-problem to obtain a new cut plane and expand a cut plane set; constructing a compact high-dimensional quadratic function fitting cutting plane set; solving a positive semidefinite programming problem to obtain a high-dimensional quadratic function parameter; converting a quadratic unconstrained binary optimization model constructed based on a high-dimensional quadratic function into an Isin model; solving the Isin model to obtain a quantum bit state, and solving an optimal binary solution; and substituting the optimal binary solution into the above steps for iterative solution. According to the invention, the problem of huge consumption of quantum bit resources in the prior art is solved, especially the problem of difficulty in processing NP with complex constraints and more variables is solved.

A neuromorphic network may solve combinatorial optimization problems. The neuromorphic network may include variable neurons, a solution monitoring neuron, and one or more readout neurons. The variable neurons may each represent one binary variable in a combinatorial optimization problem. An internal state of a variable neuron may change as the variable flips. The internal state may be stored in a memory of the variable neuron. The variable neuron may spike when its internal state changes. One or more other variable neurons receiving the spike may determine whether to change their internal states based on the spike. The variable neurons may send their internal states to the solution monitoring neuron to compute a cost of the QUBO problem and determine whether a solution is found. A readout neuron may receive variable assignments resulting in the solution from at least some variable neurons and integrate the variable assignments into one message.

An information processing device according to an embodiment includes a solver device and an information processing circuit. The solver device stores first weight values and second weight values. The solver device solves a combination optimization problem minimizing a cost function including weight values obtained by combining the first weight values and the second weight values. The information processing circuit repeatedly acquires first data. The information processing circuit repeatedly acquires second data asynchronously with the timing of acquiring the first data. The information processing circuit updates the first weight values stored in the solver device on the basis of the first data, and updates a plurality of second weight values stored in the solver device on the basis of the second data.

The invention belongs to the technical field of image classification, particularly relates to an automatic optimization method for image classification migration based on evolutionary computation, and is particularly suitable for automatically searching and optimizing a migration strategy through an evolutionary algorithm in a scene with large data distribution difference between an original domain and a target domain so as to improve the accuracy and generalization ability of target domain image classification. According to the method, fine tuning strategy search in an image classification task is modeled as a combinatorial optimization problem under structural constraints, and a customized genetic algorithm of structural perception and a lightweight performance prediction mechanism are creatively fused; the optimal fine tuning configuration adaptive to the target image classification task is efficiently and automatically searched at low cost, and the classification accuracy and generalization ability of the model in small sample, cross-domain and resource limited scenes are remarkably improved.

The invention discloses a short-wave time difference positioning method based on particle swarm optimization, which is realized by aiming at a short-wave signal source positioning problem under the condition that ionized layer information is unknown through the following steps of: dynamically screening a receiving station combination with optimal geometric distribution by adopting a particle swarm optimization algorithm based on longitude and latitude and time difference observation data of a receiving station group; taking the main receiving station as a coordinate origin, and converting the geographic coordinates of the selected station into rectangular plane coordinates; constructing a time difference positioning equation set by using a Chan algorithm, and calculating a target position through pseudo-inverse solution and an iteration strategy; and filtering the ill-conditioned solution based on a geometric accuracy factor threshold, analyzing a positioning point cluster through sliding window density, and outputting median coordinates as a final positioning result. According to the method, the receiving station is innovatively selected and modeled as a combinatorial optimization problem, the positioning precision in a complex electromagnetic environment is remarkably improved through a collaborative architecture of particle swarm optimization and a Chan algorithm, and the problem of positioning drift caused by poor geometric distribution of stations in a traditional method is effectively solved.

An information processing device according to an embodiment executes processing on data. The information processing device includes a solver device and an information processing circuit. The circuit acquires data and generates a combination optimization problem based on the data. The circuit causes the solver device to solve the combination optimization problem. A first partial weight group being part of weight values is the same as a second partial weight group being other part of the weight values. The circuit writes the first partial weight group and the second partial weight group in a common region in a first memory, and gives, to the solver device, first information indicating a storage position of each of the weight values in the first memory. The solver device reads weight values from the first memory on the basis of the first information and obtains the solution.

The invention discloses a carbon flow optimization method and device for edge calculation under vehicle-road cooperation and a medium, and relates to the technical field of traffic carbon emission. The method comprises the steps that a real-time state parameter sequence and environment information of a vehicle are acquired, and at least two candidate driving strategies are generated through a built-in carbon emission digital twinborn model; according to the candidate driving strategies, the corresponding expected carbon emission amount is calculated by inquiring the emission factor mapping table of the specific power of the vehicle and integrating; packaging the candidate driving strategy and the corresponding expected carbon emission into a bidding packet, and sending the bidding packet to an edge computing node on the roadside; based on the received bidding package, carbon right reverse auction is initiated, a combined optimization problem is solved by adopting a heuristic search algorithm, and a bid winning strategy set is determined; and generating a cooperative driving strategy according to the bid winning strategy set, and issuing the cooperative driving strategy to the corresponding vehicle end node. By means of the method, information distortion is restrained, and direct and real-time optimization control over regional total carbon emission is achieved.

The invention discloses an intelligent discharging method and device based on array combinatorial optimization and a storage medium, and relates to the technical field of image processing, and the method comprises the steps: analyzing a target graphic file to extract cutting piece information in the target graphic file; classifying the cutting piece information so as to classify the outline of the flexible material and sub cutting pieces in the flexible material; constructing an initial typesetting scheme according to the flexible material contour and the sub-cutting pieces; selecting a reference typesetting scheme from the initial typesetting schemes according to a preset fitness function; performing cross processing on the reference typesetting scheme; performing variation processing on the part of the reference typesetting schemes after the cross processing; carrying out correction processing on the reference typesetting scheme which is subjected to cross processing and is not subjected to variation processing and the reference typesetting scheme which is subjected to variation processing; and taking the reference typesetting scheme with the highest fitness as a target typesetting scheme. Typesetting of the sub-cutting pieces is regarded as a combinatorial optimization problem, and the material utilization rate is improved.

There is provided a quantum computing method for obtaining an optimal solution to a problem having a plurality of discrete variables, where the problem is represented by a cost function, the method comprising:-generating a graph structure from the cost function,-dividing the graph structure into at least two separate sub-graph structures, where each sub-graph structure comprises a subset of the plurality of variables,-dividing the sub-graph structures into at least two separate sub-graph structures, mapping each subgraph structure to a local cost function represented as a local cost Hamiltonian, determining, for each local cost Hamiltonian, all eigenstates corresponding to an energy below a predetermined cutoff energy using a quantum processing device, wherein each variable in the subset of the plurality of variables is represented by a qubit of the quantum processing device,-recombining the determined eigenstate, and-approximating a ground state from the recombined eigenstate, where the ground state represents the optimal solution.

The invention discloses a parallel computing method for a unit commitment optimization problem, which relates to the technical field of power systems and comprises the following steps: acquiring day-ahead optimization scheduling input data of a power system; decomposing the scheduling time period set with the high time resolution into a reference scheduling time period subset and a coordinated scheduling time period subset; building mutually independent unit commitment sub-optimization models in parallel and solving the mutually independent unit commitment sub-optimization models; according to a preset data restoration rule and the first unit optimal start-stop plan matrix of each unit combination sub-optimization model, restoring to obtain a second unit optimal start-stop plan matrix; determining a global set of units needing to be started and stopped and a set of remaining units; determining the start-up time period of each unit in the remaining unit set in a rolling identification form; and a unit commitment optimization model is constructed and solved. The method is used for solving the problem of unit commitment optimization, especially the problem of unit commitment optimization of a power system with high time resolution and the day-ahead clearing calculation of a power spot market.

The invention belongs to the technical field of energy consumption optimization of reconfigurable robots, and particularly relates to a deterministic optimization method and system for variable-structure energy consumption of a modular robot. The deterministic optimization method for the variable structure energy consumption of the modular robot comprises the following steps: determining an initial structure before the variable structure of the modular robot starts and a target structure after the variable structure of the modular robot ends; in the structure changing process of the modular robot, a motion replacement mechanism is used, so that the module can penetrate through the interior of the current structure to move; according to the method, the problem of minimum moving steps of all modules in the modular robot allosteric process is converted into the problem of combinatorial optimization of task allocation, a global optimal allosteric solution is obtained through configuration sliding and Hungary algorithm solving, and allosteric energy consumption optimization is achieved. According to the method, a motion substitution mechanism is provided, the globally optimal solution is rapidly and efficiently calculated through configuration sliding and the Hungary algorithm, the limitation that solving of a probability search method has randomness is overcome, and the certainty of the solution can be guaranteed.

To provide a spiking neuron apparatus able to efficiently implement a simulation of a spiking neuron. A spiking neuron apparatus using a coherent Ising machine, the coherent Ising machine including: a resonator unit for amplifying a plurality of optical pulses; a measurement unit for measuring phases and amplitudes of the optical pulses to obtain a measurement result; and a feedback configuration for computing and feeding back an interaction related to a certain optical pulse using a coupling coefficient of Ising Model on the basis of the measurement result, the feedback configuration feedback inputting correlation determined by two coupling coefficients with opposite signs to two predetermined optical pulses of the optical pulses, the spiking neuron apparatus simulating a state of a spiking neuron using one of values of two optical pulses finally obtained by the measurement unit.

To provide a simulation device that can efficiently arrive at a good solution by appropriately determining a range of a temperature parameter for searching a ground state of an interaction model using simulated annealing (SA).SOLUTION: A simulation device, which uses SA (simulated annealing) to find a solution to a combinatorial optimization problem, acquires a local optimal solution to an object function thereof by a local search method that changes a temperature parameter, calculates a statistic from the local optimum solution and a probability distribution function that a variable included in the object function follows, calculates, on the basis of the calculated statistic (standard deviation of the local optimum state), a maximum value (temperature parameter with a large standard deviation) and a minimum value (temperature parameter with a small standard deviation) that are a search range of the temperature parameter, and performs, in the search range of maximum and minimum values of the temperature parameter, a plurality of times of simulated annealing to find a good solution.SELECTED DRAWING: Figure 10

A data processing device includes: a first holding circuit that holds a representative value of each variable group of a plurality of variable groups, formed with at least two binary variables of different values or equal values included in an evaluation function of a combinatorial optimization problem, and that is a value of one of the at least two binary variables; a first memory that stores a plurality of first coupling coefficients that indicate magnitudes of interaction between the variable groups of the plurality of variable groups; and a second holding circuit that holds a first local field to be used to specify a first amount of change to be caused in a value of the evaluation function by a change in the representative value of each variable group of the plurality of variable groups.

The invention discloses an automatic guided vehicle path planning method based on a selection probability correction ant colony algorithm, relates to the field of intelligent optimization algorithms, and is suitable for logistics path optimization, robot navigation, intelligent traffic scheduling, network routing optimization and other application scenes needing to solve a combinatorial optimization problem. According to the method, an ant colony algorithm automatic guided vehicle path planning model is constructed, a straight movement preference probability correction mechanism is established, a sudden change point probability correction strategy is formulated, a pheromone redistribution rule is designed, the short-distance direction change frequency is reduced, the occurrence frequency of sudden change points is reduced, the attraction of a later-stage high-quality path is enhanced, and the path smoothness is improved.

This document teaches a computer implemented method for solving a combinatorial optimization problem of a cost function implemented on a digital computer system comprising a processor (10) adapted to execute a time evolving block decimation (TEBD) algorithm. The method comprises mapping the cost function to a Hamiltonian (H(x1, x2, ... xn)) in a mapping module (80), choosing an initial state of a vector space (V) representative of the cost function, applying a time-evolution operator (O) to the state to produce an updated state, iteratively applying the time-evolution operator to the updated state to produce a further updated state until a ground state is reached, and determining the cost function from the ground state of the Hamiltonian.

Provided is a quantum computing method for obtaining an optimal solution of a problem with multiple discrete variables, wherein the problem is represented by a cost function, the method comprising:—generating a graph structure from the cost function,—dividing the graph structure into at least two disjunct subgraph structures, wherein each subgraph structure comprises a subset of the multiple variables,—mapping each subgraph structure to a local cost function represented as local cost Hamiltonian,—determining, for each local cost Hamiltonian, all eigenstates corresponding to an energy below a predetermined cut off energy using a quantum processing device, wherein each variable of the subset of multiple variables is represented by a qubit of the quantum processing device,—recombining the determined eigenstates, and-approximating a ground state from the recombined eigenstates, wherein the ground state represents the optimal solution.

To provide an information processing device capable of predicting data to be acquired and executing processing using a combination optimization problem for the data.SOLUTION: An information processing device executes processing data. The information processing device acquires data, generates a combination optimization problem on the basis of the data, causes a solver device to solve the combination optimization problem, acquires a solution of the combination optimization problem from the solver device, generates output data on the basis of the solution, outputs the output data to a first device, and generates prediction data for predicting the data acquired after the output data is given to the first device. In addition, the information processing device acquires prediction data as one piece of data in the data acquisition.SELECTED DRAWING: Figure 7

According to a preferred aspect of the invention, provided is an information processing apparatus including a processor and a storage device. A penalty coefficient setting unit is implemented by the processor and the storage device using a solution finding function for obtaining a solution of a combinatorial optimization problem using a cost function and a constraint condition. The penalty coefficient setting unit sets a penalty function and a penalty coefficient based on the constraint condition related to a logical operation imposed between two variables of the cost function and a value of a model coefficient of the cost function such that the solution of the combinatorial optimization problem satisfies the constraint condition, and searches for the solution of the combinatorial optimization problem based on the penalty function and the penalty coefficient.

A recording medium stores a program causing a computer to execute processing including: causing an Ising machine to calculate a solution of a first combinatorial optimization problem that searches for a combination which satisfies a condition that simultaneous measurement of expected values is possible; generating a first clique that indicates a combination of observables indicated in the solution of the first combinatorial optimization problem; generating a second group that includes a predetermined number of observables based on the number of observables in the first clique; causing the Ising machine to calculate a solution of a second combinatorial optimization problem that searches for a combination satisfies a condition that simultaneous measurement of the expected values is possible; generating a second clique that indicates a combination of observables indicated in the solution of the second combinatorial optimization problem; and repeating, until a predetermined end condition is satisfied.

The invention relates to a distributed quantum computing method, device and equipment for combinatorial optimization and a medium, and the method comprises the steps: modeling a to-be-solved combinatorial optimization problem to obtain a problem function and a global solution space, and constructing a corresponding Hamiltonian based on the problem function of the to-be-solved combinatorial optimization problem; randomly taking a value from the value domain of each bit in the global solution space to generate an initial global solution; extracting a first preset number of bit values from the initial global solution to construct a first subset; constructing a quantum circuit corresponding to the first subset; optimizing the first subset through the quantum circuit to obtain an updated global solution, calculating a Hamiltonian expected value corresponding to the updated global solution, and taking the global solution corresponding to the minimum Hamiltonian expected value as the optimal solution of the to-be-solved combinatorial optimization problem; according to the method, the problem of computing power demand of large-scale combinatorial optimization is effectively solved, the performance and precision of the algorithm are improved, and the method has wide applicability and flexibility.

An apparatus for training a reinforcement learning model according to an embodiment includes a starting point determinator configured to determine starting points from an input value of a combinatorial optimization problem, a multi-explorer configured to generate exploration trajectories by performing exploration from each of the starting points using a reinforcement learning model, a trajectory evaluator configured to calculate an evaluation value of each of the exploration trajectories using an evaluation function of the combinatorial optimization problem, a baseline calculator configured to calculate a baseline for the input value from the evaluation value of each exploration trajectory, an advantage calculator configured to calculate an advantage of each of the exploration trajectories using the evaluation value of each exploration trajectory and the baseline, and a parameter updater configured to update parameters of the reinforcement learning model by using the exploration trajectories and the advantages of each exploration trajectory.

The invention relates to a constellation design method and device based on evolutionary reinforcement learning, equipment and a memory. The method comprises the following steps: converting a constellation design task problem into a multi-satellite earth observation task planning problem; modeling a multi-satellite earth observation task planning problem into a multi-target combinatorial optimization problem; a double-layer circulation model of a double-layer architecture based on a multi-target genetic algorithm and a multi-agent deep reinforcement learning algorithm is adopted to solve a multi-target combinatorial optimization problem, and a constellation deployment design scheme is optimized; the double-layer circulation model is used for realizing task planning of dynamic target observation through inner-layer circulation based on multi-agent reinforcement learning and feeding back awards and loss of a mobile network to outer-layer circulation; through outer layer circulation based on a multi-target genetic algorithm, a group of satellite constellation parameters which are maximum in return and minimum in loss and meet constraint conditions are searched as a target, and a constellation deployment design scheme is optimized. According to the method, the scheme design efficiency and the representation capability of the time window data are improved.

A method, apparatus and system for solving a combinatorial optimization problem. The method comprises: receiving a combinatorial optimization problem inputted by a user; mapping the combinatorial optimization problem into a model, wherein the model comprises complex variables to be solved and an objective function, said complex variables are used for describing, as discrete variables in a complex field, solutions to permutations and combinations of objects that meet a requirement, and the objective function is used for describing the permutations and combinations of the plurality of objects under a given constraint condition; determining a kinetic equation of the model in the complex field, wherein the kinetic equation is used for describing a time-based evolution process of said complex variables, and the kinetic equation at least comprises a control variable, a multi-bifurcation parameter and an action parameter; solving the kinetic equation, and obtaining q stable phases of said complex variables by means of changing the value of the control variable, wherein the value of q is determined by means of the multi-bifurcation parameter; and obtaining solutions to the combinatorial optimization problem on the basis of the q stable phases. In the present application, the spin scale can be reduced when solving a combinatorial optimization problem, thereby shortening solving time.