CN107368950A - A kind of resource allocation methods based on parallel batching - Google Patents
A kind of resource allocation methods based on parallel batching Download PDFInfo
- Publication number
- CN107368950A CN107368950A CN201710494779.0A CN201710494779A CN107368950A CN 107368950 A CN107368950 A CN 107368950A CN 201710494779 A CN201710494779 A CN 201710494779A CN 107368950 A CN107368950 A CN 107368950A
- Authority
- CN
- China
- Prior art keywords
- team
- agent
- batch
- resource allocation
- favorite
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q10/06311—Scheduling, planning or task assignment for a person or group
Landscapes
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Economics (AREA)
- Operations Research (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Marketing (AREA)
- Educational Administration (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of resource allocation methods based on parallel batching, step:(1) M group is set, N number of member, the number of group is N/M;(2) it is divided into several batches, the member of each batch reports the group of favorite, if this lot portion member reports identical group, the member of conflict is remained, and carries out next batch;If all members of this batch report identical group, equiprobability selects a member to add, and remaining member remains, and carries out next batch;(3) when encumbrance is equal to 0, algorithm terminates, or when being retained less than N/M, N/M member has conflict, and equiprobability selects a member to add, and loser adds a group having vacant position, and algorithm terminates when encumbrance is equal to 0;Otherwise repeat (2) (3).Policymaker's allocative efficiency is improved using technical scheme;By batch processing set of preferences, not only solves large-scale resource allocation problem, and there is good practicality and fairness.
Description
Technical field
The present invention relates to artificial intelligence and computational economics field, and in particular to a kind of resource based on parallel batching point
Method of completing the square.
Background technology
It is most important sustainability that resource is distributed between multiple Agent such as how efficient, effective and fair mode
One of problem.Recently, it has turned into the research topic of an emerging artificial intelligence, and different members of society (or colony) are usually
With different value orientation and conflicting Interest demands, it is the mankind in economic life that how various resources are made with distribution
Nearly all field major issue to be faced forever.History shows that constantly deeply understanding solves with trial to this problem,
Promote the progress and development of human society various aspects.And with the progress and development of society, the awareness of the obligations of citizens of members of society
Gradually awaken and form for two big world civilization of resource allocation social value pursuit:Economic benefit and social equality.Distribution
The mode of system action modern civilization society processing resource allocation problem, its reasonability are largely to this two big value
The response of pursuit.Therefore (can take into account economic benefit and social equality) distribution system reasonable in design is the one of modern economics
Individual important research content.Modern economics studies distribution system by the method for mathematical analysis, and its core content is to analyze
The Existence problems of ideal Distribution result in mathematical modeling, and for how to search out preferable allocation result and from calculating
Angle sees, it is very few that the difficulty of searching has a series of the problem of of being present in real operation aspect such as much to be then concerned about.
On the other hand, with the increasingly raising of human lives' level of socializations, the intelligent behavior of the mankind is more and more mainly
It is embodied in its interactive process with social environment.And various systems (particularly distribution system) are the important of modern society environment
Part.So in recent years, using aid in or realize intelligent behavior as the artificial intelligence theory of target with technical research increasingly
Pay close attention to towards multiple distribution system design problems from sharp Agent.Relevant research (includes from various real environmental restrictions
Repay distribution or free distribution, whether limited, centralised allocation or de-centralized distribution, resource to be allocated are divisible for the price of resource
Or it is indivisible, etc.) set out, take into full account the privately owned interests of individual and therewith caused rational behavior (i.e. from sharp
Agent pursues its private gainful maximization) on the basis of, design and analysis real operation aspect (particularly specific process
Change calculating aspect) on towards multiple distribution systems from sharp Agent.
However, the method for the indivisible commodity of distribution of many concentrations has been suggested, in these methods, Agent is needed
Their preference is fully disclosed to report to central authority (calculating final distribution) and pay resource allocation to them in some valency
Lattice, however, there is some shortcomings and limitations in these methods:
(1) heuristic process and winner determine that algorithm is very expensive;
(2) Agent must show their complete preferences, it may be that they are unwilling the thing done.
The content of the invention
In view of the shortcomings of the prior art, problem solved by the invention is that how to solve Agent resource allocations to take greatly, certainly
The low problem of plan executor's allocative efficiency.
In order to solve the above technical problems, the technical solution adopted by the present invention is a kind of resource allocation based on parallel batching
Method, comprise the following steps:
Step (1) assumes there is M={ A, B, C ... } individual team, there is N number of Agent (Agent quantity is team multiple),
Each team number is N/M.
Step (2) is randomly divided into ti∈{t1, t2..., tMIndividual orderly batch, per tiAgent inside individual batch is simultaneously
The team of oneself favorite is reported, if a part of Agent inside this batch has reported identical team, these are had conflict
Agent remain and be designated asThen next batch t is carried outi+1.If all Agent inside this batch
With Times identical team, when this situation occurs, a random function can be called, it is equiprobable to select one
Agent reports oneself favorite team, remaining Agent, then remains intoThen next batch is carried out.
Step (3) is as i=| M | when, at the end of this wheel batch, all people being retained can be obtained and be designated as
Step (4) is N` when the number of reservationtWhen=0, then algorithm terminates, or N`t<During N/M, there is N`tIndividual Agent mono-
If directly having conflict, then determine which Agent first reports the team of oneself favorite in the form of drawing lots, loser, which then adds one, to be had
The team of vacancy, until N`tAlgorithm terminates when=0.Otherwise, then repeat step (2)This is batch by batch
At the end of secondary, obtain all people being retained and be designated as ∪ N`tThen repeat step (4).
Wherein, in step (1), it is assumed that have a M team, N represents Agent quantity, N/M represent each team how many
Member, wherein N >=6.
In step (2), tiRepresent the i-th batch, i ∈ { 1,2 ... N/M }, and | ti|=N/M, N mod M=0, N expressions
Agent quantity,Represent in tiIdentical team Agent, t have been reported in batchi+1Represent to carry out i+1 batches.
In step (3), as i=| M | when, represent that first round batch terminates,Represent all reports in this wheel
Identical team Agent quantity.
In step (4), N`t=0, illustrate that all Agent have been assigned in that team of oneself favorite, that
The income (Utility) that each team member brings to this team reaches maximization, and algorithm terminates.Work as N`t<N/M, then
Show there is N` all the timetIndividual Agent has reported identical team, and when situation occurs in this, which determines by the way of lot
Agent first selects the team of oneself favorite, and loser is then added inside a team for having vacancy, until N`tAlgorithm knot when=0
Beam, otherwise repeat step (2).WhenWhen, show in this wheel, also i Agent is not allocated to certainly
In the team of own favorite, when there is situation in this, continue repeat step (3).
Carried using technical scheme by using parallel distribution system, the efficiency of decision-making executor distribution
It is high;By the processing set of preferences by batch, it is adapted to the large-scale resource allocation of processing, it is time-consuming big effectively solves resource allocation
The shortcomings that, while the efficiency of decision-making executor distribution is also improved, and there is good practicality.
Brief description of the drawings
Fig. 1 is an ordering of optimization preferences of the Agent to the team of favorite in the minds of oneself;
Fig. 2 is to work as to have two Agent to report identical team i.e. N`tOne of allocation result when=2;
Fig. 3 is to work as to have two Agent to report identical team i.e. N`tOther in which allocation result when=2;
Fig. 4 is the algorithm flow chart of the resource allocation obtained according to the resource allocation methods of parallel batching of the present invention;
Fig. 5 is to use and the identical set of preferences of figure one, the allocation result obtained using existing technology.
Embodiment
The embodiment of the present invention is further described with reference to the accompanying drawings and examples, but is not to this hair
Bright restriction.
The present invention considers the parallel distribution method of resource allocation system, and distributes resource in the form of batch processing.
Fig. 1 is the ordering of optimization preference that Agent most likes in the heart team to oneself.Wherein, below Agent numeral such as
" 1,2,3,4,5,6 " etc. representatives are Agent, and " A, B, C " represent the selectable team of Agent title.In addition, below team
Numeral represents favorable ratings of each Agent to team.
Fig. 2, Fig. 3 show a kind of resource allocation methods based on parallel batching, using parallel batching mechanism, make every
Individual Agent reports the preference information of oneself successively, without all reporting preference information, due to effectively being solved using batch processor system
Resource allocation of having determined takes the shortcomings that big, while the efficiency for also improving decision-making executor distribution is adapted to handle large-scale resource
Distribution.
Embodiment:
A kind of resource allocation methods based on parallel batching, comprise the following steps:
It is respectively t that step (1), which is randomly divided into two orderly batches,1And t2, it is assumed that t1=[1,2,3], t2=[4,5,6], t1
For the member of the inside with the team of Times oneself favorite, Agent1 have selected team C, Agent2 and 3 simultaneous selections team B,Then for the member inside t2 with the team of Times oneself favorite, Agent4 have selected team B, Agent5 and 6
Simultaneous selection team A,
Wherein, t1It is three Agent that first batch randomly generates;t2It is three Agent that second lot randomly generates,Represent in first batch Agent2 and 3 with Times identical team,Represent in second batch
The secondary middle Agent5 and 6 identical team with Times.
Step (2) is as 2=N/3, at the end of such first round batch, can obtain all people being retained and be designated as
Wherein, N`=[2,3,5,6] is the member of all report identical team in first batch and second lot
Set,
Step (3) is due to | N` | > 2, so algorithm continues repeat step (1),Two are randomly divided into again
Individual batch in order is respectively t1And t2, then t1=[2,3,5], t2=[6], t1The member of the inside with Times oneself favorite team,
Agent5 have selected team A, Agent2 and 3 simultaneous selections team B,Then t2The member of the inside is simultaneously
The team of oneself favorite, Agent6 is reported to have selected team A.Last N`=[2,3], then repeat step (3), work as N`t=2, adopt
Determine which Agent first reports the team of oneself favorite with the form of lot, if 2 win.Then 2 selection B, otherwise 2 selection
C。
Pass through above-mentioned technical proposal can obtain, and Fig. 1 is that Agent most likes a team ordering of optimization preference to show oneself in the heart
Illustration.Fig. 2 is the exemplary plot of the resource allocation obtained according to resource allocation methods of the present invention based on parallel batching.Figure
In 2, when Agent2 and 3 has reported identical team at the same time, 2 team for first selecting oneself favorite are determined by the way of lot, its
Secondary is that Agent3 selects last to have the team of vacancy.And in Fig. 3, when Agent2 and 3 has reported identical team at the same time, use
The mode of lot determines 3 team for first selecting oneself favorite, next to that Agent2 selects last to have the team of vacancy.
It can thus be seen that Fig. 5 prior art is compared with Fig. 2 with Fig. 3 present invention, parallel batching side of the invention
Method, the fairness of allocation result can be preferably represented, and the efficiency and speed ratio that perform are very fast.
Carried using technical scheme by using parallel distribution system, the efficiency of decision-making executor distribution
It is high;By the processing set of preferences by batch, it is adapted to the large-scale resource allocation of processing, it is time-consuming big effectively solves resource allocation
The shortcomings that, while the efficiency of decision-making executor distribution is also improved, and there is good practicality.
Embodiments of the present invention are made that with detailed description above in association with drawings and examples, but the present invention is not limited to
Described embodiment.To those skilled in the art, without departing from the principles and spirit of the present invention, it is right
These embodiments carry out various change, modification, replacement and modification and still fallen within protection scope of the present invention.
Claims (5)
1. a kind of resource allocation methods based on parallel batching, comprise the following steps:
Step (1) assumes there is M={ A, B, C ... } individual team, has N number of Agent (Agent quantity is team multiple), each
Team number is N/M;
Step (2) is randomly divided into ti∈{t1, t2..., tMIndividual orderly batch, per tiAgent inside individual batch with Times from
The team of own favorite, if a part of Agent inside this batch has reported identical team, these there is conflict
Agent, which is remained, to be designated asThen next batch t is carried outi+1;If all Agent inside this batch are same
Times identical team, when this situation occurs, a random function can be called, it is equiprobable to select one
Agent reports oneself favorite team, remaining Agent, then remains intoThen next batch is carried out;
Step (3) is as i=| M | when, at the end of this wheel batch, all people being retained can be obtained and be designated as
Step (4) is N` when the number of reservationtWhen=0, then algorithm terminates, or N`t<During N/M, there is N`tIndividual Agent has always
If conflict, then determine which Agent first reports the team of oneself favorite in the form of drawing lots, loser then adds one and had vacant position
Team, until N`tAlgorithm terminates when=0;Otherwise, then repeat step (2) ifThis wheel batch terminates
When, obtain all people being retained and be designated as ∪ N`tThen repeat step (4).
2. the resource allocation methods according to claim 1 based on parallel batching, it is characterised in that;In step (1),
Assuming that there is M team, N represents Agent quantity, and N/M represents each how many member of team, wherein N >=6.
3. the resource allocation methods according to claim 1 or 2 based on parallel batching, it is characterised in that;In step (2)
In, tiRepresent the i-th batch, i ∈ { 1,2 ... N/M }, and | ti|=N/M, N mod M=0, N represent Agent quantity,Table
Show in tiIdentical team Agent, t have been reported in batchi+1Represent to carry out i+1 batches.
4. the resource allocation methods according to claim 1 or 2 based on parallel batching, it is characterised in that;In step (3)
In, as i=| M | when, represent that first round batch terminates,Represent all Agent for having reported identical team in this wheel
Quantity.
5. the resource allocation methods according to claim 1 or 2 based on parallel batching, it is characterised in that;In step (4)
In, N`t=0, illustrate that all Agent have been assigned in that team of oneself favorite, then each team member
The income (Utility) brought to this team reaches maximization, and algorithm terminates;Work as N`t<N/M, then show there is N` all the timetIt is individual
Agent has reported identical team, when situation occurs in this, determines which Agent first selects oneself by the way of lot
The team of favorite, loser is then added inside a team for having vacancy, until N`tAlgorithm terminates when=0, otherwise repeat step
(2);WhenWhen, show in this wheel, also i Agent is not allocated in the team of oneself favorite,
When there is situation in this, continue repeat step (3).
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710494779.0A CN107368950A (en) | 2017-06-26 | 2017-06-26 | A kind of resource allocation methods based on parallel batching |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710494779.0A CN107368950A (en) | 2017-06-26 | 2017-06-26 | A kind of resource allocation methods based on parallel batching |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107368950A true CN107368950A (en) | 2017-11-21 |
Family
ID=60305225
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710494779.0A Pending CN107368950A (en) | 2017-06-26 | 2017-06-26 | A kind of resource allocation methods based on parallel batching |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107368950A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110807683A (en) * | 2019-09-27 | 2020-02-18 | 贝壳技术有限公司 | Message-based house source fusion method |
-
2017
- 2017-06-26 CN CN201710494779.0A patent/CN107368950A/en active Pending
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110807683A (en) * | 2019-09-27 | 2020-02-18 | 贝壳技术有限公司 | Message-based house source fusion method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Simaria et al. | A genetic algorithm based approach to the mixed-model assembly line balancing problem of type II | |
CN107544251B (en) | Single machine scheduling method for minimizing total stall period based on distributed robust model | |
CN103235743B (en) | A kind of based on decomposing and the multiple goal test assignment dispatching method of optimum solution follow-up strategy | |
CN104572297B (en) | A kind of Hadoop job scheduling methods based on genetic algorithm | |
CN104842564B (en) | A kind of 3 D-printing multitask Optimization Scheduling based on NSGA II | |
CN103679388B (en) | Production scheduling method and production scheduling system | |
CN103279818A (en) | Method for cloud workflow scheduling based on heuristic genetic algorithm | |
CN104077634B (en) | active-reactive type dynamic project scheduling method based on multi-objective optimization | |
CN109784656A (en) | A kind of discrete manufacture collaborative production planning scheduling method | |
CN110097190A (en) | A kind of intelligent perception method for allocating tasks based on dual-time limitation | |
CN107423914A (en) | A kind of method for scheduling task and system based on cloud manufacture | |
CN116430736A (en) | Multi-agent autonomous cooperative allocation method for aerospace measurement and control | |
CN107368950A (en) | A kind of resource allocation methods based on parallel batching | |
CN102547726A (en) | Bidding auction-based frequency spectrum allocation method with collusion inhibition effect in cognitive radio system | |
Korgin et al. | Experimental and theoretical comparison of several resource allocation mechanisms | |
Han et al. | Multiple rules decision-based DE solution for the earliness-tardiness case of hybrid flow-shop scheduling problem | |
Korgin et al. | Experimental gaming analysis of ADMM dynamic distributed optimization algorithm | |
CN114021895B (en) | Neighborhood structure-based minimum total cost IT operation and maintenance personnel scheduling method and system | |
CN109343966A (en) | A kind of cluster organization method and device of unmanned node | |
Esquivel et al. | Parameter settings and representations in Pareto-based optimization for job shop scheduling | |
CN110968428B (en) | Cloud workflow virtual machine configuration and task scheduling collaborative optimization method | |
CN107832864A (en) | The net bidded under environment about special train distribution and pricing method | |
CN106970840A (en) | A kind of Method for HW/SW partitioning of combination task scheduling | |
CN103440533B (en) | The confining method of the non-bottleneck ability of job shop under a kind of cloud manufacturing mode | |
Kwong et al. | Determination of fault-tolerant fabric-cutting schedules in a just-in-time apparel manufacturing environment |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20171121 |
|
WD01 | Invention patent application deemed withdrawn after publication |