CN104166783A - CUDA-based quick sequence distribution computing method for steady-state olefin copolymerization - Google Patents

CUDA-based quick sequence distribution computing method for steady-state olefin copolymerization Download PDF

Info

Publication number
CN104166783A
CN104166783A CN 201410324525 CN201410324525A CN104166783A CN 104166783 A CN104166783 A CN 104166783A CN 201410324525 CN201410324525 CN 201410324525 CN 201410324525 A CN201410324525 A CN 201410324525A CN 104166783 A CN104166783 A CN 104166783A
Authority
CN
Grant status
Application
Patent type
Prior art keywords
chain
cuda
monomer
step
sequence
Prior art date
Application number
CN 201410324525
Other languages
Chinese (zh)
Other versions
CN104166783B (en )
Inventor
陈曦
翁金祖
姚臻
Original Assignee
浙江大学
Priority date (The priority date 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 date listed.)
Filing date
Publication date

Links

Abstract

The invention discloses a CUDA-platform-based quick sequence distribution computing method applied to steady-state olefin copolymerization. For the steady-state process in an olefin reaction, the quick computing method for solving copolymer sequence distribution is provided under a CUDA platform according to the dynamics mechanism of a copolymerization reaction and on the basis of the Monte Carlo method, In the method, firstly, different probabilities required in the Monte Carlo method are provided according to the dynamics mechanism of the copolymerization reaction and the needed sequence distribution is finally obtained by executing the Monte Carlo method on the CUDA platform in a parallel mode. Due to the fact that the whole computing process is high in parallelism degree, the computing time is greatly shortened. Thus, the method is called as a quick sequence distribution computing technology.

Description

一种基于CUDA的稳态烯烃共聚快速序列分布计算方法 In a distributed computing method based on the steady-state flash sequence olefin copolymer CUDA

技术领域 FIELD

[0001] 本发明涉及基于CUDA平台下的高分子共聚稳态过程中的快速序列分布计算技术方法。 [0001] The present invention relates to a polymer-based platform CUDA copolymerization under steady state process in rapid sequence distributed computing art methods.

背景技术 Background technique

[0002] 随机数生成器,是指的能够生成随机数的函数或者程序模块。 [0002] The random number generator means capable of generating a random number function or program module. 在连续型随机变量的分布中,最简单而且最基本的分布是单位均匀分布,由该分布抽取的简单子样称为随机数序列,其中每一个个体称为随机数。 In the distribution of continuous random variable, the most simple and most basic unit of distribution is uniform, simple sub extracted by the random number sequence is called sample distribution, wherein each individual is called a random number. 独立性、均匀性是随机数必备的两个特点。 Independence, uniformity is a random number of two essential features. 包括蒙特卡罗计算方法在内的大多数算法都要求所采用的随机数序列服从均匀分布,即同一范围内的任一个数出现的概率相同。 Most algorithms including Monte Carlo calculation method are required, including the sequence of random numbers uniformly distributed used, i.e. the same probability of a any number appearing in the same range.

[0003] 蒙特卡罗方法,也称统计模拟方法,是二十世纪四十年代中期由于科学技术的发展和电子计算机的发明而被提出的一种以概率统计理论为指导的一类非常重要的数值计算方法。 [0003] Monte Carlo method, also known as statistical simulation method, is a mid-1940s due to the invention and development of computer science and technology have been presented to the theory of probability and statistics is a very important class of guidance numerical method. 该方法使用随机数(或更常见的伪随机数)来解决很多计算问题,与它对应的是确定性算法。 The method uses a random number (or more common pseudo-random number) is calculated to solve many problems, and it corresponds to a deterministic algorithm. 蒙特卡罗方法在化工领域已经得到认可和应用。 Monte Carlo method in the chemical industry has been recognized and applied. 在给定动力学机理的情况下,根据共聚系统稳态的各个状态值来计算出不同反应类型的概率;其次,设定的共聚体系中存在的分子链的数目,并进一步根据随机数生成器所生成的一系列随机数来重复的判定反应中各条链的反应情况,直到整个系统中的所有链都终止了为止。 In the case of a given kinetic mechanism, according to the respective steady state values ​​of the copolymerization system to calculate the probabilities of different types of reaction; secondly, setting the number of molecular chain present in the copolymerization system, and further according to the random number generator the generated series of random numbers repeated in the case of the reaction of the reaction is determined that the pieces of chain, so the chain until the entire system are terminated so far.

[0004] 统一计算设备架构(CUDA),是显卡产商NVIDIA推出的运算平台,是一种通用并行计算架构。 [0004] Compute Unified Device Architecture (the CUDA), the graphics card manufacturers NVIDIA Release computing platform, a general purpose parallel computing architecture. 由于它包含了指令集架构以及并行计算引擎,因此能够解决很多复杂的计算问题,并且大幅度的缩短计算时间,计算效率得到明显的提升。 Because it contains the instruction set architecture and parallel computing engine, it is possible to solve many complex computational problems, and greatly reduce the computing time and computational efficiency significantly improved. 序列分布,是指不同种类单体构成的序列在分子链中出现频率的一类分布。 Sequence distribution, refers to the sequence of the different types of monomers occurs in a class of frequency distributions of the molecular chain. 在高分子化学领域,聚合物的性能指标包括常见的熔融指数、平均分子量、分子量分布,然而在共聚体系中这些指标并不能完全的描述聚合物的性能,除了严格的交替共聚和嵌段共聚外,同一个大分子内各个单体的排列是不规则的,因此就存在着链段的序列分布。 In the field of polymer chemistry, performance polymers include common melt index, average molecular weight, molecular weight distribution, but these indicators do not fully describe the properties of the polymer in the copolymerization system, in addition to strictly alternating copolymerization and block copolymerization outer , with the arrangement of the respective monomers within a macromolecule is irregular, therefore there is a sequence distribution segment.

发明内容 SUMMARY

[0005] 本发明的目的是针对稳态烯烃共聚体系中自由基聚合反应的应用场景,提供一种基于CUDA的稳态烯烃共聚快速序列分布计算方法。 [0005] The object of the present invention is directed to a steady state scenarios olefin copolymer radical polymerization reaction system, to provide an olefin-based steady-state calculation method CUDA rapid sequence distribution copolymer.

[0006] 本发明的技术方案如下: [0006] aspect of the present invention is as follows:

[0007] 基于CUDA的稳态烯烃共聚快速序列分布计算方法包括如下步骤: [0007] Based on the steady-state flash sequence olefin copolymer CUDA calculated distribution comprising the steps of:

[0008] a.读取稳态烯烃共聚体系的状态值,包括链增长、链转移、链终止反应的动力学常数以及各类单体、链转移剂、链终止剂的浓度; . [0008] a steady state value read-olefin copolymer system, comprising a chain growth, chain transfer, the kinetic constants of chain termination reactions and a variety of monomers, the chain transfer agent concentration of chain terminating agent;

[0009] b.计算蒙特卡罗方法所需要的各个概率值,包括以各类单体结尾的活性链发生链增长反应的概率Pi以及以各类单体结尾的活性链向各类单体链增长的概率Pu,以各类单体结尾的活性链发生链增长反应的概率即将各类单体结尾的链增长化学反应速率乘以相应单体的浓度之后的加和,除以各类单体结尾的链增长、链转移、链终止化学反应速率乘以相应单体的浓度之后的加和,以各类单体结尾的活性链向各类单体链增长的概率即将向各类单体链增长化学反应速率乘以对应单体的浓度,除以各类单体结尾的链增长化学反应速率乘以相应单体的浓度之后的加和;以公式表示: [0009] b. Calculating respective probability values ​​required Monte Carlo method, the probability of the active strand comprises chain extension reaction occurring types of monomers Pi and ending at the end of the active strand of various types of various types of monomer to monomer chains growth probability Pu, with a probability of occurrence of the living chain end of the chain extension reaction of the monomer is about various types of various types of chain-end chemical reaction rate of the monomer multiplied by the corresponding monomers after concentration and addition, various types of monomer divided by the end of the chain growth, chain transfer, chemical reaction rate multiplied by the chain termination after the addition and the concentration of the respective monomers, the probability of an active chain end of the various types of monomer to chain growth of various types of monomers is about to various types of monomer chains increase chemical reaction rates multiplied by the corresponding monomer concentration divided by the growth of various types of chemical reaction rate of the monomer chain end concentration multiplied by the corresponding monomers after adding and; in formula:

[0010] [0010]

Figure CN104166783AD00041

[0012] 其中,Rpi表示聚合反应中以单体i结尾的活性链链增长速率;Rti表示以单体i结尾的活性链链转移速率;Rdi表示以单体i结尾的活性链链终止速率;[j]、[m]分别表示单体j、单体m的浓度;kpij、kpim分别表示以单体i结尾的活性链向单体j、单体m发生链增长的化学反应速率; [0012] wherein, Rpi denotes a polymerization reaction to increase the active chain end of the rate of the monomer i; Rti represents an active chain end of the transfer rate of the monomer i; Rdi represents an active chain end of the termination rate of the monomer i; [J], [m] denote the concentration of monomer j, the m monomer; kpij, kpim represent the chemical reaction rate, the occurrence of a chain of monomer m the active chain end of the monomer to the monomer J i;

[0013] c.将步骤b计算得到的概率值从CPU平台传递到CUDA平台上; . [0013] c calculated probability value obtained in step b is transferred from the CPU to the internet platform CUDA;

[0014] d.在CUDA平台上开辟用于记录序列信息的存储空间以及数目等于模拟总链数的线程数; . [0014] d the total number of threads of the open-chain and storage space for recording information is equal to the number of sequences in the simulation platform CUDA;

[0015] e.并行的执行所有线程,在每个线程里依靠步骤b中得到的概率,顺序地判断相应活性链是否发生链增长;若否,停止该线程的模拟计算并将得到的序列信息按线程编号在存储空间中进行存储;若是,则继续进行判断向哪一类单体进行链增长并记录对应的序列信息; . [0015] e parallel execution of all threads, relying probability obtained in step b in each thread sequentially determines whether the corresponding activity chain chain growth occurs; if not, the thread stops simulation and the resulting sequence information by thread number is stored in a storage space; if yes, proceed to a judgment which a chain monomer corresponding to the sequence information and records;

[0016] f.重复步骤e,直到获得停止信息并退出; . [0016] f repeating steps e, until stop information and exit;

[0017] g.将记录的序列信息从CUDA平台传递到CPU平台上; . [0017] g recording sequence information transmitted from the internet to the CPU platform CUDA;

[0018] h.统计所有的序列信息,得到所需要的序列分布。 [0018] h. All statistical sequence information sequence to obtain the desired distribution.

[0019] 步骤b中所述的蒙特卡罗方法的模拟平台为CUDA。 [0019] Step b, the Monte Carlo simulation platform for CUDA. 步骤b中所述的蒙特卡罗方法的模拟方式为每个模拟线程都只进行一条链的模拟过程。 In step b the Monte Carlo method in an analog manner for each analog simulation process with only a thread chain. 步骤b中所述的蒙特卡罗方法的模拟顺序为并行的执行多个蒙特卡罗模拟线程。 Monte Carlo simulation of the sequence of step b according to the plurality of parallel execution threads Monte Carlo simulation.

[0020] 步骤e中所述的记录信息为链中所有的序列信息。 [0020] Step e in the information recording sequence information for all chain.

[0021] 基于CUDA的稳态烯烃共聚中的快速序列分布计算方法的核心思想是:用每个线程进行共聚反应中每条链的模拟,并将这些线程在CUDA平台下运行实现,从而实现快速计算。 [0021] The core idea of ​​the method of calculating steady-state flash sequence distribution of olefin comonomer is CUDA: simulation of each chain copolymerization reaction with each thread, and these threads run implemented in CUDA platform, enabling fast calculation. 方法是:首先,设定需要模拟计算的链的数目以及一些系统的参数值;其次,根据给定的共聚动力学反应机理以及稳态下对应的动力学参数值和系统状态值,计算出以不同共聚物类型结尾的活性链进行各种动力学反应的概率以及在此基础上进一步向不同反应物类型反应的概率等;然后,将前面的参数值以及计算出来的概率传到CUDA平台上,以链数为线程数进行重复运算,直到满足设定的链的数目或者最大链长为止;最后,将CUDA上模拟计算出来的每条链的信息传回主程序,统计出所需要的不同序列的数目,再把这些值进行归一化整理后得到所需要的共聚物序列分布。 The method is: first, set the number of chains need to simulate some of the system and calculated parameter values; Secondly, according to the copolymerization kinetics of the reaction mechanism and kinetics parameter values ​​and the corresponding values ​​of the system state at the given steady-state, in order to calculate different types of copolymers probability of ending various active strand and kinetics of the reaction in a further reaction to a different type of reaction on the basis of probability; then, the foregoing parameter values ​​and the probability calculated on the spread CUDA platform, the number of threads is a chain operation is repeated until the set number of chains meet or until the maximum length of the chain; Finally, each chain information calculated on the CUDA simulated return the main routine, the statistics of the different sequences required number, then these values ​​were normalized to give a copolymer sequence distribution finishing required.

[0022] 本发明与现有技术相比的有益效果是:由于蒙特卡罗的模拟计算过程以多线程的形式分配给CUDA平台运行,因此并行度较高、计算时间大幅度缩短,从而实现了快速序列分布计算方法。 [0022] Advantageous effects of the present invention compared to prior art: Because the Monte Carlo simulation of the process assigned to run as CUDA internet multithreading, a higher degree of parallelism, greatly reduce the computing time, thereby achieving fast sequence distribution calculation.

附图说明 BRIEF DESCRIPTION

[0023] 图1是基于CUDA的稳态烯烃共聚快速序列分布计算方法的主程序模块流程图; [0023] FIG 1 is a main program module copolymerizable rapid sequence flowchart of a method based on distributed computing steady CUDA-olefin;

[0024]图2是本发明中的蒙特卡罗模拟计算模块流程图; [0024] FIG 2 is a flowchart of the calculation module Monte Carlo simulation of the present invention;

[0025] 图3是本发明得到的序列分布图。 [0025] FIG. 3 is a sequence distribution of the present invention is obtained.

具体实施方式 detailed description

[0026] 以烯烃共聚反应体系为例,对本发明的技术方案进行进一步说明。 [0026] In olefin copolymerization reaction system as an example, the technical solution of the present invention will be further described.

[0027] I实例背景介绍 [0027] I Examples Background

[0028] 本发明中,以烯烃共聚反应体系为实施案例。 [0028] In the present invention, olefin copolymerization reaction system implementation case. 烯烃由于其原材料丰富且价格低廉、容易加工成型、综合性能优良,是一类产量巨大、应用十分广泛的高分子材料。 Olefin because of its rich raw materials and low cost, easy molding processing, excellent overall performance, is a kind of huge production, it is widely used polymer materials. 在烯烃聚合体系中,由于通过共聚不仅能够扩大聚合物的品种,而且可以让一些难以进行均聚的单体参与聚合反应,因此共聚体系的应用比较广泛。 In the olefin polymerization system, since the species can be increased by copolymerizing not only polymers, but also can be difficult for some homopolymerization of the monomers participating in the polymerization reaction, the more extensive application copolymerization system.

[0029] 2聚合反应机理 [0029] The polymerization mechanism 2

[0030] 本发明中,以两元共聚为例,考虑烯烃共聚反应体系中带有末端效应的链增长反应与链转移反应,如表I所示。 [0030] In the present invention, a two copolymerized as an example, consider the olefin copolymer a chain reaction in the reaction system with end effects and the chain transfer, as shown in Table I.

[0031] 表I烯烃共聚反应机理 [0031] TABLE I-olefin copolymerization reaction mechanism

[0032] [0032]

Figure CN104166783AD00061

[0033] 其中,A与B分别代表了二元共聚反应的两种单体类型;P/是链长为r并以A结尾的活性链;P/是链长为r并以B结尾的活性链;比是链长为r的死聚物链;匕是空活性位;H2为氢气;A1为助催化剂;kp,kt分别是链增长与链转移的动力学反应速率常数。 [0033] wherein, A and B represent the two monomer types dipolymerization reaction; P / r a chain length of A and ending with the active strand; P / r and a chain length of B at the end of the active chain; r than the chain length of the oligomer chains death; dagger empty active site; H2 of hydrogen; A1 as cocatalyst; kp, kt rate constants are kinetic chain propagation and chain transfer.

[0034] 3主程序模块 [0034] The main program module 3

[0035] 主程序模块主要负责模拟工作的准备工作以及最后的数据整理工作,如图1所示。 [0035] The main program module is responsible for simulated operation of the preparations and the last data collation, as shown in FIG. 准备工作有:根据给定的系统状态值计算出进行蒙特卡罗模拟所需要的每个参数值,并开辟与设定的总链数相等的线程数。 Preparations are: the parameter value for each Monte Carlo simulation based on a given desired value calculation system state, and to open the total number of threads is set equal to the chain. 数据整理工作有:统计每个线程模拟计算出来的每条链中8种序列(AAA, AAB,ABA, ABB, BAA, BAB, BBA,BBB)各自所占的比重,统计这些比重信息得到所需要的序列分布。 Collation data are: statistical simulation Each thread out of eight in each chain sequence (AAA, AAB, ABA, ABB, BAA, BAB, BBA, BBB) occupied by each of the specific gravity, the specific gravity of these statistics to obtain the desired information the sequence distribution.

[0036] 4蒙特卡罗模拟计算模块 [0036] Monte Carlo simulation module 4

[0037] 每个线程下的蒙特卡罗模拟计算流程图如图2所示,具体步骤如下: [0037] calculating the flowchart shown in FIG. 2 in a Monte Carlo simulation for each thread, the following steps:

[0038]步骤一:获取从主程序传过来的系统参数值(rn, rA, rB, fA, fB, qtA, qtB, Pp;LS, PpLS;A),跳到步骤二; [0038] Step a: acquiring system parameter values ​​(rn, rA, rB, fA, fB, qtA, qtB, Pp; LS, PpLS; A) ​​passed over from the main program, jumps to step II;

[0039] 步骤二:将链长(r)置零,每条链的最后三个单体的标记(LS、SS、TS,分别是链最后一个单体的标记、链倒数第二个单体的标记、链倒数第三个单体的标记)置零,8种序列的记录值(ηΑΑΑ, ηΑΑΒ, η舰,η.,ηΒΑΑ, ηΒΑΒ, ηΒΒΑ, ηΒΒΒ)直零,跳到少骤二; [0039] Step 2: chain length (r) to zero, marking the last three monomers each strand (LS, SS, TS, namely, a single strand of the last marker, the penultimate monomer chain mark, marking the third countdown chain monomers) zero, eight kinds of recording a sequence of values ​​(ηΑΑΑ, ηΑΑΒ, η ship, η, ηΒΑΑ, ηΒΑΒ, ηΒΒΑ, ηΒΒΒ) straight zero, skip to step two less.;

[0040] 步骤三:利用随机数生成器生成一个(0,I)随机数,根据活性链最后一个单体的标记LS来判断生成的随机数与具体哪个概率值进行比较,从而判定是执行链增长还是链转移;如果是链增长,则跳到步骤四;如果是链转移,则跳到步骤二十二; [0040] Step Three: using a random number generator generates a (0, I) a random number, according to the last active strand single marker LS generated random number to determine which of the particular probability value, thereby determining the implementation chain growth or chain transfer; if the chain is, skip to step four; if it is a chain transfer, skip to step two 52;

[0041] 步骤四:执行r自加操作,并再次生成一个(0,I)随机数,同样根据活性链最后一个单体的标记LS来判断生成的随机数与具体哪个概率值进行比较,从而判定是执行单体A链增长还是单体B链增长;如果是单体A链增长,则跳到步骤五;如果是单体B链增长,则跳到步骤六; [0041] Step Four: r from performing add operation, and generates a (0, I) a random number again, also in accordance with the last active strand single marker LS generated random number to determine which of the particular probability value is compared to A determination is performed monomeric chain extender or chain extender monomer B; if A is a monomeric chain growth, skip to step five; if monomer B is a chain, then jumps to step six;

[0042] 步骤五:更新所有的标记,也就是把SS赋给TSjE LS赋给SSJE LS赋为A,跳到步骤七; [0042] Step Five: update all the marks, which is assigned to the SS TSjE LS assigned SSJE LS assigned to A, skip to Step seven;

[0043] 步骤六:更新所有的标记,也就是把SS赋给TSjE LS赋给SSJE LS赋为B,跳到步骤七; [0043] Step Six: Update all the tags, which is assigned to the SS TSjE LS assigned SSJE LS assigned to B, skip to Step seven;

[0044] 步骤七:根据TS的值判断该链倒数第三个单体是否为A,如果是,跳到步骤八;如果不是,则跳到步骤九; [0044] Step 7: The value of the TS determines whether the chain is antepenultimate monomers A, if so, skip to step eight; if not, skip to Step 9;

[0045] 步骤八:根据SS的值判断该链倒数第二个单体是否为A,如果是,跳到步骤十;如果不是,则跳到步骤十一; [0045] Step Eight: The value of the SS determines whether the chain penultimate monomer is A, if it is, go to Step 10; if not, skip to Step 11;

[0046] 步骤九:根据SS的值判断该链倒数第二个单体是否为A,如果是,跳到步骤十二;如果不是,则跳到步骤十三; [0046] Step 9: The value of the SS determines whether the chain penultimate monomer is A, if it is, go to step 52; if not, skip to step thirteen;

[0047] 步骤十:根据LS的值判断该链最后一个单体是否为A,如果是,跳到步骤十四;如果不是,则跳到步骤十五; [0047] Step 10: The determination value LS of the last chain of one monomer whether A, if so, skip to step fourteen; if not, skip to step fifteen;

[0048] 步骤十一:根据LS的值判断该链最后一个单体是否为A,如果是,跳到步骤十六;如果不是,则跳到步骤十七; [0048] Step 11: The determination value LS of the last chain of one monomer whether A, if so, skip to step sixteen; if not, skip to step seventeen;

[0049] 步骤十二:根据LS的值判断该链最后一个单体是否为A,如果是,跳到步骤十八;如果不是,则跳到步骤十九; [0049] Step 12: The determination value LS of the last chain of one monomer whether A, if so, skip to step eighteen; if not, skip to step nineteen;

[0050] 步骤十三:根据LS的值判断该链最后一个单体是否为A,如果是,跳到步骤二十;如果不是,则跳到步骤二十一; [0050] Step Thirteen: LS based on the value of the chain is determined whether the last monomer A, if so, skip to step twenty; if not, skip to step twenty-one;

[0051] 步骤十四:执行ηωΑ自加操作,跳到步骤三; [0051] Step 14: performing self ηωΑ add operation jumps to step III;

[0052] 步骤十五:执行ηωΒ自加操作,跳到步骤三; [0052] Step five: performing ηωΒ increase since operation jumps to step III;

[0053] 步骤十六:执行Iiaba自加操作,跳到步骤三; [0053] Step six: performing Iiaba increase since operation jumps to step III;

[0054] 步骤十七:执行η-自加操作,跳到步骤三; [0054] Step seven: performing self-η- add operation jumps to step III;

[0055] 步骤十八:执行ηΒΑΑ自加操作,跳到步骤三; [0055] Step eight: performing ηΒΑΑ increase since operation jumps to step III;

[0056] 步骤十九:执行ηΒΑΒ自加操作,跳到步骤三; [0056] Step nine: performing ηΒΑΒ increase since operation jumps to step III;

[0057] 步骤二十:执行ηΒΒΑ自加操作,跳到步骤三; [0057] Step Twenty: performing ηΒΒΑ increase since operation jumps to step III;

[0058] 步骤二i^一:执行nBBB自加操作,跳到步骤三; [0058] i ^ a step two: performing nBBB increase since operation jumps to step III;

[0059]少骤_■十_■:保存所葡要的Ίη 息(n—,Haab> ^aba> n_,nBAA, nBAB, nBBA, nBBB),矛王序JS7TT结束。 [0059] _ ■ step least ten _ ■: saving the information to the glucosamine Ίη (n-, Haab> ^ aba> n_, nBAA, nBAB, nBBA, nBBB), the end of the lance Wang Xu JS7TT.

[0060] 5对比效果 [0060] 5 contrast

[0061] 本发明中,设定值如下: [0061] In the present invention, the set values ​​are as follows:

[0062] rn = 1000,rA = 5.0,rB = 0.2,fA = 0.6,fB = 0.4,qtA = 0.5,qtA = 0.5 [0062] rn = 1000, rA = 5.0, rB = 0.2, fA = 0.6, fB = 0.4, qtA = 0.5, qtA = 0.5

Figure CN104166783AD00081

[0068] Pp0 = PpAfA+PpBfB,Ppoa = PAAfA+PBAfB [0068] Pp0 = PpAfA + PpBfB, Ppoa = PAAfA + PBAfB

[0069] 其中,rn表示数均链长;rA表示以A结尾的活性链向A链增长的反应速率除以以A结尾的活性链向B链增长的反应速率;rB表示以B结尾的活性链向B链增长的反应速率除以以B结尾的活性链向A链增长的反应速率;fA表示单体A的浓度除以A的浓度与B的浓度的加和;fB表示单体B的浓度除以A的浓度与B的浓度的加和;qtA表示以A结尾的链转移、链终止化学反应速率乘以相应单体的浓度之后的加和除以以A和B结尾的链转移、链终止化学反应速率乘以相应单体的浓度之后的加和;qtB表示以B结尾的链转移、链终止化学反应速率乘以相应单体的浓度之后的加和除以以A和B结尾的链转移、链终止化学反应速率乘以相应单体的浓度之后的加和表示单体A在共聚物中的平均摩尔分率;巧表示单体B在共聚物中的平均摩尔分率;PpA表示以A结尾的活性链发生链增长反应的概率^表示以A [0069] where, rn represents the number average chain length; of rA represents the reaction rate at the end of the reaction rate of A-chain to A-chain active growth divided by the active end of the A chain to the B chain growth; rB represents ending B activity the reaction rate of the chain to the B-chain-reaction rate divided by the active end of the B chain to the a chain growth; fA represents the concentration of the monomer concentration divided by the concentration of a and B and a plus; fB represents a monomer B divided by the concentration of a and B the concentration of the additive; Qta a represents the chain transfer to the end of the chain terminator, after the chemical reaction rate multiplied by the concentration of the respective monomers added and divided by the end of the a and B chain transfer, after termination of the chain multiplied by the concentration of chemical reaction rates and adding corresponding monomers; QTB expressed in the B chain transfer ends, the chemical reaction rate multiplied by the chain termination after the addition and the concentration of the respective monomers a and B divided by the end of the after the transfer of the chain, the chain termination reaction rate multiplied by the concentration of the chemical additive and the corresponding monomers a represent average molar fraction of the monomer in the copolymer; Qiao represents an average molar fraction of the monomer B in the copolymer; represents PPA chain activity occurs with a probability of end of the a chain-reaction represented by a ^ 尾的活性链发生向A链增长反应的概率;PpB表示以B结尾的活性链发生链增长反应的概率;PBA表示以B结尾的活性链发生向A链增长反应的概率;Pp(l与Ppcia分别是用来进行判断链增长与具体向哪一类单体链增长的初始概率。 The probability of the A chain extension reaction of the active end of the chain occurs; ppb represents probability active chain to chain growth reaction end B; the PBA represents the active chain end of the B occurrence probability of the A chain-reaction; Pp (l and Ppcia They are used for determining the probability of chain growth and particularly to the initial monomer which chain growth.

[0070] 根据这些参数值,套入前面的模块中,可以得到序列分布如下图3所示: [0070] The values ​​of these parameters, set into the front of the module, the sequence distribution can be obtained as shown in Figure 3:

[0071] 整个程序分别在CPU平台下串行运行、在CUDA平台下并行运行,各自所需时间如下表2所示: [0071] Serial entire program are running in CPU platform, running in parallel CUDA platform, each required time as shown in Table 2:

[0072] 表2运行时间表 [0072] Table 2 Run schedule

[0073] [0073]

Figure CN104166783AD00091

[0074] 进一步的,可以计算得到加速效果: [0074] Further, the acceleration effect can be calculated:

Figure CN104166783AD00092

[0076] 可以看到,在CUDA平台下进行蒙特卡罗的模拟计算能够大幅度的提升计算效率,从而实现快速序列分布计算技术。 [0076] It can be seen Monte Carlo simulation can greatly enhance computational efficiency in CUDA platform, enabling rapid sequence distributed computing techniques.

Claims (5)

  1. 1.一种基于CUDA的稳态烯烃共聚快速序列分布计算方法,其特征在于包括如下步骤: a.读取稳态烯烃共聚体系的状态值,包括链增长、链转移、链终止反应的动力学常数以及各类单体、链转移剂、链终止剂的浓度; b.计算蒙特卡罗方法所需要的各个概率值,包括以各类单体结尾的活性链发生链增长反应的概率Pi以及以各类单体结尾的活性链向各类单体链增长的概率Pij: R1,: P —_H_ 'Rpi + Rti + Kii k..[/IP _ PU L-, J Σο] m 其中,Rpi表示聚合反应中以单体i结尾的活性链链增长速率;Rti表示以单体i结尾的活性链链转移速率;Rdi表示以单体i结尾的活性链链终止速率;[j]、[m]分别表示单体j、单体m的浓度;kpU、kpim分别表示以单体i结尾的活性链向单体j、单体m发生链增长的化学反应速率; c.将步骤b计算得到的概率值从CPU平台传递到CUDA平台上; d.在CUDA平台上开辟用于记录 1. A fast method for calculating the distribution of comonomer sequences based on steady-state CUDA-olefins, comprising the steps of: a steady state value read-olefin copolymer systems, including chain growth, chain transfer, the kinetics of chain termination. constants and a variety of monomers, the chain transfer agent concentration of chain terminating agent;. b probability values ​​calculated for each Monte Carlo method required, including the probability of occurrence active chain end of the chain extension reaction of the various types of monomer as well as Pi various types of monomers to the active strand probability of chain growth of various types of monomer end Pij: R1 ,: P -_H_ 'Rpi + Rti + Kii k .. [/ IP _ PU L-, J Σο] m wherein, Rpi represents an active polymerization rate of the monomer chain growth ending i; Rti represents an active chain end of the transfer rate of the monomer i; Rdi represents an active chain end of the termination rate of monomer i; [j], [m] They represent the concentration of monomer j, the m monomer; kpU, kpim represent the active chain end of the monomer to the monomer i j, m monomer chain growth rate of a chemical reaction; C probability calculated in step b. value passed from the CPU to the internet platform CUDA;. d on the open platform for recording CUDA 列信息的存储空间以及数目等于模拟总链数的线程数; e.并行的执行所有线程,在每个线程里依靠步骤b中得到的概率,顺序地判断相应活性链是否发生链增长;若否,停止该线程的模拟计算并将得到的序列信息按线程编号在存储空间中进行存储;若是,则继续进行判断向哪一类单体进行链增长并记录对应的序列信息; f.重复步骤e,直到获得停止信息并退出; g.将记录的序列信息从CUDA平台传递到CPU平台上; h.统计所有的序列信息,得到所需要的序列分布。 Information storage and the number of columns is equal to the total number of threads of the analog chain;. E parallel execution of all threads, relying probability obtained in step b in each thread sequentially determines whether the corresponding activity chain chain growth occurs; if not , sequence information of the simulation is stopped and the resulting thread by thread number is stored in a storage space; if yes, proceed to a judgment which a chain monomer corresponding to the sequence information and records; F repeats to step e. until stop information and exit;. g sequence recorded information is transferred from the CPU CUDA platform to platform; the sequence count all the sequence information to obtain the desired distribution of h.
  2. 2.根据权利要求1所述的一种基于CUDA的稳态烯烃共聚快速序列分布计算方法,其特征在于步骤b中所述的蒙特卡罗方法的模拟平台为CUDA。 2. A method according to claim 1 fast calculation method based on the stationary-olefin comonomer sequence distribution of CUDA, wherein said step (b) Monte Carlo simulation platform for CUDA.
  3. 3.根据权利要求1所述的一种基于CUDA的稳态烯烃共聚快速序列分布计算方法,其特征在于步骤b中所述的蒙特卡罗方法的模拟方式为每个模拟线程都只进行一条链的模拟过程。 According to claim 1, wherein one of the calculation method based on the steady-state flash olefin comonomer sequence distribution of CUDA, characterized in that the Monte Carlo simulation mode in said step b for each simulation for only one strand of thread the simulation process.
  4. 4.根据权利要求1所述的一种基于CUDA的稳态烯烃共聚快速序列分布计算方法,其特征在于步骤b中所述的蒙特卡罗方法的模拟顺序为并行的执行多个蒙特卡罗模拟线程。 According to claim 1, wherein one of the calculation method based on the steady-state flash sequence distribution copolymer CUDA-olefins, characterized in that the Monte Carlo simulation order of said step b is performed in a plurality of parallel Monte Carlo simulation thread.
  5. 5.根据权利要求1所述的一种基于CUDA的稳态烯烃共聚快速序列分布计算方法,其特征在于步骤e中所述的记录信息为链中所有的序列信息。 According to one of the fast sequence of claim 1 calculated based on steady-state distribution of olefin comonomer CUDA, characterized in that the information recorded in said step e for all chains of sequence information.
CN 201410324525 2014-07-09 2014-07-09 Cuda calculation method in a distributed olefin comonomer fast steady state based on the sequence CN104166783B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 201410324525 CN104166783B (en) 2014-07-09 2014-07-09 Cuda calculation method in a distributed olefin comonomer fast steady state based on the sequence

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 201410324525 CN104166783B (en) 2014-07-09 2014-07-09 Cuda calculation method in a distributed olefin comonomer fast steady state based on the sequence

Publications (2)

Publication Number Publication Date
CN104166783A true true CN104166783A (en) 2014-11-26
CN104166783B CN104166783B (en) 2017-03-22

Family

ID=51910592

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 201410324525 CN104166783B (en) 2014-07-09 2014-07-09 Cuda calculation method in a distributed olefin comonomer fast steady state based on the sequence

Country Status (1)

Country Link
CN (1) CN104166783B (en)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5301118A (en) * 1991-11-18 1994-04-05 International Business Machines Corporation Monte carlo simulation design methodology
CN102063544A (en) * 2011-01-04 2011-05-18 浙江大学 Multicore parallel solving method for computation of polymer molecular weight distribution
CN102289559A (en) * 2011-05-30 2011-12-21 复旦大学 Monte Carlo simulation prediction radical copolymerization of the copolymer sequence distribution system

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5301118A (en) * 1991-11-18 1994-04-05 International Business Machines Corporation Monte carlo simulation design methodology
CN102063544A (en) * 2011-01-04 2011-05-18 浙江大学 Multicore parallel solving method for computation of polymer molecular weight distribution
CN102289559A (en) * 2011-05-30 2011-12-21 复旦大学 Monte Carlo simulation prediction radical copolymerization of the copolymer sequence distribution system

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
郭海凤: "基于CUDA平台的伪随机数产生器系统研究", 《计算机技术与发展》 *

Also Published As

Publication number Publication date Type
CN104166783B (en) 2017-03-22 grant

Similar Documents

Publication Publication Date Title
McCollum et al. The sorting direct method for stochastic simulation of biochemical systems with varying reaction execution behavior
Degnan et al. Gene tree distributions under the coalescent process
Rogers et al. A fast method for approximating maximum likelihoods of phylogenetic trees from nucleotide sequences
Buback et al. Consistent values of rate parameters in free radical polymerization systems
Rannala et al. Improved reversible jump algorithms for Bayesian species delimitation
Yang et al. Statistical properties of the branch-site test of positive selection
Bevington et al. The termination reaction in radical polymerizations. II. Polymerizations of styrene at 60° and of methyl methacrylate at 0 and 60°, and the copolymerization of these monomers at 60°
Ermolaeva et al. The age of the Arabidopsis thaliana genome duplication
Hendrix et al. miRTRAP, a computational method for the systematic identification of miRNAs from high throughput sequencing data
Leo et al. Biodoop: bioinformatics on hadoop
Campbell et al. MAKER-P: a tool-kit for the rapid creation, management, and quality control of plant genome annotations
Tirosh et al. Comparative analysis indicates regulatory neofunctionalization of yeast duplicates
Bryant et al. QSRA–a quality-value guided de novo short read assembler
Nakhleh et al. Towards the development of computational tools for evaluating phylogenetic network reconstruction methods
Merz et al. Intramolecular reactions in vinyl polymers as a means of investigation of the propagation step
Hoffman et al. Batch Polymerization
Deaton et al. A software tool for generating non-crosshybridizing libraries of DNA oligonucleotides
Blum et al. On statistical tests of phylogenetic tree imbalance: the Sackin and other indices revisited
Kolaczkowski et al. Recurrent adaptation in RNA interference genes across the Drosophila phylogeny
Zhang et al. CMDS: a population-based method for identifying recurrent DNA copy number aberrations in cancer from high-resolution data
Szász Limit theorems for the distributions of the sums of a random number of random variables
Yang et al. Unsupervised binning of environmental genomic fragments based on an error robust selection of l-mers
Iedema et al. Synthesis of branched polymer architectures from molecular weight and branching distributions for radical polymerisation with long‐chain branching, accounting for topology‐controlled random scission
Lawrence Dependency of intervals between events in superposition processes
Fernandez Correlation of pause sites in MDV‐1 RNA replication with kinetic refolding of the growing chain: A Monte Carlo simulation of the Markov process

Legal Events

Date Code Title Description
C06 Publication
C10 Entry into substantive examination
C14 Grant of patent or utility model