A method and system for sorting and selecting with fixed precision based on KT algorithm

Abstract

The present invention discloses a method and system for sorting and selecting with fixed precision based on the KT algorithm. The method comprises: generating initial sample data required for sorting and selecting according to parameters input by a user or default parameters; performing initialization settings to generate different candidate solutions, and dividing all candidate solutions into several groups; based on elimination rules and a round-by-round mechanism, each processor independently performs a local selection process for a group of candidate solutions to obtain a local optimal candidate solution; each processor generates additional observations for the local optimal candidate solution according to specific rules; then calculating the sample mean of the additional observations of each local optimal candidate solution, and selecting the candidate solution with the largest sample mean as the best candidate solution. The present invention uses the KT algorithm to efficiently solve various large-scale fixed-precision sorting and selection problems, that is, under a pre-set precision, efficiently finding the solution with the best mean performance from a large-scale set of candidate solutions.

Description

Technical Field

[0001] The present invention relates to the technical field of sorting and optimization methods, and in particular to a method and system for sorting and optimization with fixed precision based on a KT algorithm. Background Art

[0002] The ranking and selection (R&S) problem is an important issue in the field of simulation. It aims to find the solution with the best average performance from a limited set of alternatives by conducting experiments and observing the random performance of each alternative. Depending on the constraints or objectives, algorithms for solving the ranking and selection problem can be roughly divided into two categories: fixed-precision algorithms and fixed-budget algorithms. The main goal of fixed-precision algorithms is to complete the selection with the smallest possible sample budget while satisfying certain statistical guarantees for the selected alternatives, such as the probability of correct selection (PCS) and the probability of good selection (PGS). The goal of fixed-budget algorithms is to find the best sample budget allocation for different solutions given a total sample budget, thereby maximizing PCS.

[0003] Traditional algorithms for solving sorting optimization problems are usually designed to handle problems with less than 500 alternatives, and are usually measured and compared through their empirical performance on some test problems (such as the total sample size). When the number of alternatives is large, such as the above, the use of parallel thinking is particularly important. At the same time, in recent years, computing technology has developed rapidly, making parallel computing environments widely available and easily accessible to ordinary users. In this context, using the powerful computing resources of these parallel computing environments to solve large-scale problems has become a hot area of ​​research. However, directly applying traditional algorithms to parallel computing environments to solve large-scale problems will lead to excessive simulation samples, comparisons between alternatives, and communication between processors.

[0004] Therefore, it is an urgent problem for those skilled in the art to propose a method and system for sorting and selecting with fixed precision based on the KT algorithm to solve the difficulties existing in the prior art. Summary of the Invention

[0005] The purpose of the present invention is to provide a method and system for fixed-precision sorting and optimization based on the KT algorithm, which uses the KT algorithm to efficiently solve various large-scale fixed-precision sorting and optimization problems, that is, to efficiently find the solution with the best mean performance from a large set of candidate solutions under a preset precision.

[0006] To achieve the above object, the present invention provides the following solutions:

[0007] A method for sorting and selecting with fixed precision based on the KT algorithm comprises the following steps:

[0008] S1. Generate the initial sample data required for sorting and selection based on the parameters input by the user or the default parameters;

[0009] S2. Perform initialization settings, generate different candidate solutions, and divide all candidate solutions into several groups;

[0010] S3, based on the elimination rule and round-by-round mechanism, each processor independently executes the local selection process of a set of candidate solutions to obtain the local optimal candidate solution;

[0011] S4. Each processor generates additional observations for the local optimal candidate solution according to specific rules; then calculates the sample mean of the additional observations of each local optimal candidate solution, and selects the candidate solution with the largest sample mean as the best candidate solution.

[0012] Preferably, initialization settings are performed in S2 to generate different candidate solutions, and all candidate solutions are divided into several groups, specifically including:

[0013] Choose the probability of wrong selection α, set the number of candidate solutions k, parameter δ>0, the sample size n0 of the first stage, the number of processors m and the number of candidate solutions in a match g≥2; set represents the set of candidate solutions that are still competing at the beginning of round r in processor s;

[0014] Distribute k candidate solutions evenly to m processors so that each processor handles the selection of k / m candidate solutions. For example, for i=1,…,k, the set of candidate solutions competing in the first round in the first processor is

[0015] Preferably, in S2, based on the elimination rule and the round-by-round mechanism, each processor independently performs a local selection process of a set of candidate solutions to obtain the local optimal candidate solution, which specifically includes the following steps:

[0016] S201, initial grouping: For each processor, The candidate solutions in the group are grouped into g candidate solutions, and the remaining candidate solutions are grouped into a small group; therefore, there are Group, order represents the set of candidate solutions for the qth group of processors s in the rth round;

[0017] S202, round selection: In each round r, set the probability of incorrect selection α for this round of allocation r =α / 2 r, for each q group, set the candidate solution set of the qth group of processor s in the rth round And calculate the candidate solution set of processor s in the r+1th round Among them, KN(C,α r ,δ,n0) is the output of the KN program; the processor eliminates the corresponding candidate solutions according to the calculation results and adds the remaining candidate solutions to the candidate set for the next round;

[0018] S203, local optimal selection: For each processor, repeat S202 until At this time, each processor selects the local optimal candidate solution and sets I s for The index of the candidate solution in .

[0019] Preferably, the KN program in S202 outputs KN(C,α r ,δ,n0) is obtained by S2.2.1, S2.2.2 and S2.2.3 as follows:

[0020] S2.2.1, Initialization: For each candidate solution i∈C, generate s0 observations X i,1 ,X i,2 ,…, and calculate the sample mean of the observations Set the error probability control parameters And for For all j≠i, calculate the difference metric parameter and the difference significance parameter Among them, X i,l is the l-th observation value of candidate solution i, X j,l is the l-th observation value of candidate solution j, is the mean of s0 observations of candidate solution j;

[0021] Set the comprehensive difference significance parameter N r,i =max j≠i N r,i,j and the maximum comprehensive difference significance parameter N r,max =max i∈C N r,i ;if Then stop and select the one with the largest Otherwise, set t = s0 and go to S2.2.2;

[0022] S2.2.2, Screening: Set the currently selected candidate solution set C old =C and

[0023] S2.2.3, Stopping rule: If |C| = 1, stop and select the candidate whose index is in C as the best candidate; otherwise, if Then stop and select the index in C with the maximum Otherwise, an additional observation X is obtained from each candidate solution i∈X i,t+1 , set t=t+1, and go to S2.2.2.

[0024] Preferably, each processor in S4 generates additional observations for the local optimal candidate solution according to a specific rule; then calculates the sample mean of the additional observations of each local optimal candidate solution, and selects the candidate solution with the largest sample mean as the optimal candidate solution, which specifically includes the following steps:

[0025] S301, Initialization: For each processor, generate s0 observations for the local optimal candidate solution and calculate the sample variance based on the observations

[0026] S302, additional observation value generation: setting rounds and h(α r ,m,s0), where h(α r ,m,s0) is given by the constant α r ,m and s0 determine the Rinott constant, and for candidate solution I s Generate the following number of additional observations:

[0027] S303, best candidate solution selection: For each processor's local best candidate solution, based on s0 and The sample mean of the observations is calculated and the candidate with the largest sample mean is selected as the optimal candidate.

[0028] A system for fixed-precision sorting and optimization based on the KT algorithm, applying any of the above methods for fixed-precision sorting and optimization based on the KT algorithm, comprising a sample generator, an initialization module, a local selection module, and a best candidate solution generation module connected in sequence; wherein,

[0029] The sample generator is used to generate the initial sample data required for sorting and selection based on the parameters input by the user or the default parameters;

[0030] The initialization module is used to perform initialization settings, generate different candidate solutions, and divide all candidate solutions into several groups;

[0031] The local selection module is used to obtain the local optimal candidate solution by independently executing the local selection process of a set of candidate solutions on each processor based on the elimination rule and round-robin mechanism;

[0032] The best candidate solution generation module is used for each processor to generate additional observations for the local best candidate solution according to specific rules; then calculate the sample mean of the additional observations of each local best candidate solution, and select the candidate solution with the largest sample mean as the best candidate solution.

[0033] Preferably, the local selection module includes:

[0034] The initial grouping unit is used to group the The candidate solutions in the group are divided into groups, one group contains g candidate solutions, and the remaining candidate solutions are grouped into a group, which is divided into Group;

[0035] The round selection unit is used to set the probability of incorrect selection α in each round r r =α / 2 r , for each q group, set the candidate solution set of the qth group of processor s in the rth round And calculate the candidate solution set of processor s in the r+1th round The processor eliminates the corresponding candidate solutions based on the calculation results and adds the remaining candidate solutions to the candidate set for the next round;

[0036] The local optimal selection unit is used to repeat the steps in the round selection unit for each processor until At this time, each processor selects the local optimal candidate solution and sets I s for The index of the candidate solution in .

[0037] A non-transitory computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements any of the above-mentioned methods for fixed-precision sorting and optimization based on the KT algorithm.

[0038] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0039] The method of this application uses the KT algorithm to efficiently solve various large-scale fixed-precision sorting and optimization problems. That is, under a pre-set precision, it efficiently finds the solution with the best average performance from a large set of candidate solutions. The system of this application is a parallel computing open source software with simple operation, wide application scope, and scalable. It overcomes the various shortcomings of existing software technologies, such as high user threshold, narrow application scope, low scalability, high purchase cost, and untimely software updates. It mainly serves practitioners who need to solve practical large-scale sorting and optimization problems and scientific researchers who want to test their own parallel programs for solving large-scale sorting and optimization problems and compare them with other programs. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0041] Figure 1 A flow chart of a method for fixed-precision sorting and optimization based on the KT algorithm provided by the present invention;

[0042] Figure 2 This is a system structure diagram of a fixed-precision sorting and optimization based on the KT algorithm provided by the present invention. DETAILED DESCRIPTION

[0043] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0044] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0045] like Figure 1 As shown, the present invention provides a representation learning method based on a multimodal knowledge graph, comprising the following steps:

[0046] A method for sorting and selecting with fixed precision based on the KT algorithm comprises the following steps:

[0047] S1. Generate the initial sample data required for sorting and selection based on the parameters input by the user or the default parameters;

[0048] S2. Perform initialization settings, generate different candidate solutions, and divide all candidate solutions into several groups;

[0049] S3, based on the elimination rule and round-by-round mechanism, each processor independently executes the local selection process of a set of candidate solutions to obtain the local optimal candidate solution;

[0050] S4. Each processor generates additional observations for the local optimal candidate solution according to specific rules; then calculates the sample mean of the additional observations of each local optimal candidate solution, and selects the candidate solution with the largest sample mean as the best candidate solution.

[0051] Furthermore, S2 performs initialization settings, generates different candidate solutions, and divides all candidate solutions into several groups, including:

[0052] Choose the probability of wrong selection α, set the number of candidate solutions k, parameter δ>0, the sample size n0 of the first stage, the number of processors m and the number of candidate solutions in a match g≥2; set represents the set of candidate solutions that are still competing at the beginning of round r in processor s;

[0053] Distribute k candidate solutions evenly to m processors so that each processor handles the selection of k / m candidate solutions. For example, for i=1,…,k, the set of candidate solutions competing in the first round in the first processor is

[0054] Furthermore, in S2, based on the elimination rule and round-robin mechanism, each processor independently performs a local selection process for a set of candidate solutions to obtain the local optimal candidate solution. Specifically, the following steps are included:

[0055] S201, initial grouping: For each processor, The candidate solutions in the group are grouped into g candidate solutions, and the remaining candidate solutions are grouped into a small group; therefore, there are Group, order represents the set of candidate solutions for the qth group of processors s in the rth round;

[0056] S202, round selection: In each round r, set the probability of incorrect selection α for this round of allocation r =α / 2 r , for each q group, set the candidate solution set of the qth group of processor s in the rth round And calculate the candidate solution set of processor s in the r+1th round Among them, KN(C,α r ,δ,n0) is the output of the KN program, which is able to obtain the best result with at least α under the budget n0. rThe probability of correctly selecting the best candidate in the candidate set C is that the mean difference of the candidate solutions is at least δ; the processor eliminates the corresponding candidate solutions based on the calculation results and adds the remaining candidate solutions to the candidate set for the next round;

[0057] S203, local optimal selection: For each processor, repeat S202 until At this time, each processor selects the local optimal candidate solution and sets I s for The index of the candidate solution in .

[0058] Furthermore, the KN program in S202 outputs KN(C,α r ,δ,n0) is obtained by S2.2.1, S2.2.2 and S2.2.3 as follows:

[0059] S2.2.1, Initialization: For each candidate solution i∈C, generate s0 observations X i,1 ,X i,2 ,…, and calculate the sample mean of the observations Set the error probability control parameters And for For all j≠i, calculate the difference metric parameter and the difference significance parameter Among them, X i,l is the l-th observation value of candidate solution i, X j,l is the l-th observation value of candidate solution j, is the mean of s0 observations of candidate solution j;

[0060] Set the comprehensive difference significance parameter N r,i =max j≠i N r,i,j and the maximum comprehensive difference significance parameter N r,max =max i∈C N r,i ;if Then stop and select the one with the largest Otherwise, set t = s0 and go to S2.2.2;

[0061] S2.2.2, Screening: Set the currently selected candidate solution set C old =C and

[0062] S2.2.3, Stopping rule: If |C| = 1, stop and select the candidate whose index is in C as the best candidate; otherwise, if Then stop and select the index in C with the maximum Otherwise, an additional observation X is obtained from each candidate solution i∈C i,t+1 , set t=t+1, and go to S2.2.2.

[0063] Furthermore, each processor in S4 generates additional observations for the local optimal candidate solution according to specific rules. Then, the sample mean of the additional observations of each local optimal candidate solution is calculated, and the candidate solution with the largest sample mean is selected as the optimal candidate solution. Specifically, the following steps are included:

[0064] S301, Initialization: For each processor, generate s0 observations for the local optimal candidate solution and calculate the sample variance based on the observations

[0065] S302, additional observation value generation: setting rounds and h(α r ,m,s0), where h(α r ,m,s0) is given by the constant α r ,m and s0 determine the Rinott constant, and for candidate solution I s Generate the following number of additional observations:

[0066] S303, best candidate solution selection: For each processor's local best candidate solution, based on s0 and The sample mean of the observations is calculated and the candidate with the largest sample mean is selected as the optimal candidate.

[0067] like Figure 2 As shown, a system based on the fixed precision sorting and optimization of the KT algorithm, applying any of the above methods based on the fixed precision sorting and optimization of the KT algorithm, comprises a sample generator, an initialization module, a local selection module, and an optimal candidate solution generation module connected in sequence; wherein,

[0068] The sample generator is used to generate the initial sample data required for sorting and selection based on the parameters input by the user or the default parameters;

[0069] The initialization module is used to perform initialization settings, generate different candidate solutions, and divide all candidate solutions into several groups;

[0070] The local selection module is used to obtain the local optimal candidate solution by independently executing the local selection process of a set of candidate solutions on each processor based on the elimination rule and round-robin mechanism;

[0071] The best candidate solution generation module is used for each processor to generate additional observations for the local best candidate solution according to specific rules; then calculate the sample mean of the additional observations of each local best candidate solution, and select the candidate solution with the largest sample mean as the best candidate solution.

[0072] Furthermore, the local selection module includes:

[0073] The initial grouping unit is used to group the The candidate solutions in the group are divided into groups, one group contains g candidate solutions, and the remaining candidate solutions are grouped into a group, which is divided into Group;

[0074] The round selection unit is used to set the probability of incorrect selection α in each round r r =α / 2 r , for each q group, set the candidate solution set of the qth group of processor s in the rth round And calculate the candidate solution set of processor s in the r+1th round The processor eliminates the corresponding candidate solutions based on the calculation results and adds the remaining candidate solutions to the candidate set for the next round;

[0075] The local optimal selection unit is used to repeat the steps in the round selection unit for each processor until At this time, each processor selects the local optimal candidate solution and sets I s for The index of the candidate solution in .

[0076] Specifically, the system also includes a visualization module for interacting with the initialization module, the local selection module, and the best candidate solution generation module to provide users with real-time feedback functions, display the progress and results of the current algorithm, and allow users to make adjustments according to their needs.

[0077] The best candidate modules include:

[0078] Initialization unit: used to generate n0 observations for the local optimal candidate solution for each processor and calculate the sample variance based on these observations

[0079] Additional observation generation unit: used to set and h(α r ,m,n0), where h(α r ,m,n0) is given by the constant α r ,m and n0 determine the Rinott constant, and for candidate solution I s Generate the following number of additional observations:

[0080] Best candidate solution selection unit: for each processor's local best candidate solution, based on the above n0 and The sample mean of the observations is calculated and the candidate with the largest sample mean is selected as the optimal candidate.

[0081] A non-transitory computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements any of the above-mentioned methods for fixed-precision sorting and optimization based on the KT algorithm.

[0082] The large-scale fixed-precision sorting and optimization method based on the KT algorithm has a wide range of applications and plays an important role in real life. Here are a few examples to illustrate:

[0083] First, in the field of logistics management, by sorting and optimizing a large number of distribution routes and resource allocation plans, we can improve efficiency, reduce transportation costs, and effectively deal with the limited nature of resources.

[0084] In addition, in financial investment, large-scale fixed-precision sorting and optimization algorithm software can help screen out high-return, low-risk options from massive investment portfolios, thereby optimizing asset allocation and improving investment returns.

[0085] In supply chain management, large-scale fixed-precision sorting and optimization algorithm software can help companies quickly identify optimal solutions for raw material procurement, product sales channel selection, and other processes, enhancing supply chain flexibility and resilience. In the healthcare industry, by evaluating and prioritizing multiple treatment options, large-scale fixed-precision sorting and optimization algorithm software can help medical staff make rapid and appropriate treatment decisions while ensuring accuracy, thereby improving the quality of medical services.

[0086] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, or of course, by hardware. Based on this understanding, the essence of the above technical solution or the part that contributes to the existing technology can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or certain parts of the embodiments.

[0087] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the method and core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims

1. A method for sorting and selecting with fixed precision based on KT algorithm, characterized in that: The following steps are involved: S1. Generate the initial sample data required for sorting and selection based on the parameters input by the user or the default parameters; S2. Perform initialization settings, generate different candidate solutions, and divide all candidate solutions into several groups; S3, based on the elimination rule and round-by-round mechanism, each processor independently executes the local selection process of a set of candidate solutions to obtain the local optimal candidate solution; S4. Each processor generates additional observations for the local optimal candidate solution according to a specific rule; then calculates the sample mean of the additional observations of each local optimal candidate solution, and selects the candidate solution with the largest sample mean as the optimal candidate solution; In S3, based on the elimination rule and the round-by-round mechanism, each processor independently performs a local selection process of a set of candidate solutions to obtain the local optimal candidate solution, which specifically includes the following steps: S301, initial grouping: For each processor, The candidate solutions in the group are grouped into g candidate solutions, and the remaining candidate solutions are grouped into a small group; therefore, there are Group, order represents the set of candidate solutions for the qth group of processors s in the rth round; S302, round selection: In each round r, set the probability of incorrect selection α allocated in this round r =α / 2 r , for each q group, set the candidate solution set of the qth group of processor s in the rth round And calculate the candidate solution set of processor s in the r+1th round Among them, KN(C,α r ,δ,n0) is the output of the KN program; the processor eliminates the corresponding candidate solutions according to the calculation results and adds the remaining candidate solutions to the candidate set for the next round; S303, local optimal selection: For each processor, repeat S302 until At this time, each processor selects the local optimal candidate solution and sets I s for The index of the candidate solution in; The KN program in S302 outputs KN(C, α r ,δ,n0) is obtained by S3.2.1, S3.2.2 and S3.2.3 as follows: S3.2.1, Initialization: For each candidate solution i∈C, generate s0 observations and calculate the sample mean of the observations Set the error probability control parameters And for For all j≠i, calculate the difference metric parameter and the difference significance parameter Among them, X i,l is the l-th observation value of candidate solution i, X j,l is the l-th observation value of candidate solution j, is the mean of s0 observations of candidate solution j; Set the comprehensive difference significance parameter N r,i =max j≠i N r,i,j and the maximum comprehensive difference significance parameter N r,max =max i∈ C N r,i ;if Then stop and select the one with the largest Otherwise, set t = s0 and go to S3.2.2; S3.2.2, Screening: Set the currently selected candidate solution set C old =C and S3.2.3, Stopping rule: If |C| = 1, then stop and select the candidate whose index is in C as the best candidate; otherwise, if Then stop and select the index in C with the maximum Otherwise, an additional observation X is obtained from each candidate solution i∈C i,t+1 , set t = t + 1, and go to S3.2.2; In S4, each processor generates additional observations for the local optimal candidate solution according to a specific rule; then calculates the sample mean of the additional observations of each local optimal candidate solution, and selects the candidate solution with the largest sample mean as the optimal candidate solution, which specifically includes the following steps: S401, Initialization: For each processor, generate s0 observations for the local optimal candidate solution and calculate the sample variance based on the observations S402, Additional Observation Generation: Setting Rounds α r =α / 2 m and h(α r ,m,s0), where h(α r ,m,s0) is given by the constant α r ,m and s0 determine the Rinott constant, and for candidate solution I s Generate the following number of additional observations: S403, best candidate solution selection: For each processor's local best candidate solution, based on s0 and The sample mean of the observations is calculated and the candidate with the largest sample mean is selected as the optimal candidate.

2. The method of fixing the precision and selecting the best based on the KT algorithm according to claim 1, characterized in that: In S2, initialization settings are performed to generate different candidate solutions, and all candidate solutions are divided into several groups, specifically including: Choose the probability of wrong selection α, set the number of candidate solutions k, parameter δ>0, the sample size n0 of the first stage, the number of processors m and the number of candidate solutions in a match g≥2; set represents the set of candidate solutions that are still competing at the beginning of round r in processor s; Distribute k candidate solutions evenly to m processors so that each processor handles the selection of k / m candidate solutions; for i = 1, ..., k, the set of candidate solutions competing in the first round in the first processor is 3. A fixed precision sorting and optimization system based on the KT algorithm, characterized in that: A method for sorting and selecting with fixed precision based on the KT algorithm according to any one of claims 1-2 is applied, characterized in that it comprises a sample generator, an initialization module, a local selection module, and a best candidate solution generation module connected in sequence; wherein, The sample generator is used to generate the initial sample data required for sorting and selection according to the parameters input by the user or the default parameters; The initialization module is used to perform initialization settings, generate different candidate solutions, and divide all candidate solutions into several groups; The local selection module is used to obtain the local optimal candidate solution by independently executing a local selection process of a group of candidate solutions on each processor based on elimination rules and a round-by-round mechanism; The optimal candidate solution generation module is used for each processor to generate additional observation values ​​for the local optimal candidate solution according to specific rules; then calculate the sample mean of the additional observation values ​​of each local optimal candidate solution, and select the candidate solution with the largest sample mean as the optimal candidate solution.

4. The system for fixed-precision sorting and optimization based on the KT algorithm according to claim 3 is characterized in that: The local selection module includes: The initial grouping unit is used to group the The candidate solutions in the group are divided into groups, one group contains g candidate solutions, and the remaining candidate solutions are grouped into a group, which is divided into Group; The round selection unit is used to set the probability of incorrect selection α in each round r r =α / 2 r , for each q group, set the candidate solution set of the qth group of processor s in the rth round And calculate the candidate solution set of processor s in the r+1th round The processor eliminates the corresponding candidate solutions based on the calculation results and adds the remaining candidate solutions to the candidate set for the next round; The local optimal selection unit is used to repeat the steps in the round selection unit for each processor until At this time, each processor selects the local optimal candidate solution and sets I s for The index of the candidate solution in .

5. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for fixed-precision sorting and optimization based on the KT algorithm as described in any one of claims 1 to 2 is implemented.

Images