双相演化

此词条暂由Henry翻译。

}}

Dual phase evolution (DPE) is a process that drives self-organization within complex adaptive systems. It arises in response to phase changes within the network of connections formed by a system's components. DPE occurs in a wide range of physical, biological and social systems. Its applications to technology include methods for manufacturing novel materials and algorithms to solve complex problems in computation. 双相演化（DPE）是复杂适应系统中驱动自组织的过程。它的产生是对系统组成部分所形成的连接网络中的相位变化的响应。DPE发生在广泛的物理、生物和社会系统中。它在技术上的应用包括制造新材料的方法和解决复杂计算问题的算法。

Introduction介绍

DPE is capable of producing social networks with known topologies, notably small-world networks and scale-free networks. In the absence of social interaction, the uptake of an opinion promoted by media is a Markov process. The effect of social interaction under DPE is to retard the initial uptake until the number converted reaches a critical point, after which uptake accelerates rapidly.

DPE能够产生具有已知拓扑结构的社交网络，特别是小世界网络和无标度网络。在缺乏社会互动的情况下，媒体对某一观点的接受是一个 Markov process马尔可夫过程。DPE下的社会互动效应是延迟最初的吸收，直到转化的数量达到临界点，之后的吸收将迅速加速。

Dual phase evolution (DPE) is a process that promotes the emergence of large-scale order in complex systems. It occurs when a system repeatedly switches between various kinds of phases, and in each phase different processes act on the components or connections in the system. DPE arises because of a property of graphs and networks: the connectivity avalanche that occurs in graphs as the number of edges increases. 双相演化（DPE）是一个促进复杂系统中大规模有序涌现的过程。当一个系统在不同的阶段之间反复切换，并且在每个阶段中，不同的过程作用于系统中的组件或连接时，就会发生这种情况。DPE的产生是由于图和网络的一个性质：当边的数目增加时，图中的连接性雪崩将会发生。

Social networks provide a familiar example. In a social network the nodes of the network are people and the network connections (edges) are relationships or interactions between people. For any individual, social activity alternates between a local phase, in which they interact only with people they already know, and a global phase in which they can interact with a wide pool of people not previously known to them. Historically, these phases have been forced on people by constraints of time and space. People spend most of their time in a local phase and interact only with those immediately around them (family, neighbors, colleagues). However, intermittent activities such as parties, holidays, and conferences involve a shift into a global phase where they can interact with different people they do not know. Different processes dominate each phase. Essentially, people make new social links when in the global phase, and refine or break them (by ceasing contact) while in the local phase.

社交网络提供了一个熟悉的例子。在社交网络中，网络的节点是人，网络连接（边缘）是人与人之间的关系或互动。对于任何个人来说，社会活动都是在“局部阶段”和“全局阶段”之间交替进行的，在前者中个体只与他们已经认识的人进行互动，在后者中可以与他们以前不认识的大量人进行互动。历史上，这些阶段是由于时间和空间的限制而强加给人们的。人们把大部分时间花在局部阶段，只与周围的人（家人、邻居、同事）交流。然而，诸如聚会、假日和会议之类的间歇式活动涉及到一个全局阶段的转变，在这个阶段，他们可以与不认识的不同的人进行互动。不同的过程控制着每个阶段。从本质上讲，人们在全局阶段建立新的社会联系，在局部阶段则通过停止联系来改善或打破这种联系。

DPE models of socio-economics interpret the economy as networks of economic agents. Several studies have examined the way socioeconomics evolve when DPE acts on different parts of the network. One model interpreted society as a network of occupations with inhabitants matched to those occupations. In this model social dynamics become a process of DPE within the network, with regular transitions between a development phase, during which the network settles into an equilibrium state, and a mutating phase, during which the network is transformed in random ways by the creation of new occupations.

社会经济学的DPE模型将经济解释为经济主体网络。有几项研究考察了当DPE作用于网络的不同部分时，社会经济的演化方式。一个模型将社会解释为一个职业网络，其居民与这些职业相匹配。在这个模型中，社会动态成为网络内的DPE过程，在发展阶段（网络进入平衡状态）和变异阶段（网络通过创造新的职业以随机方式转变）之间有规律的过渡。

The DPE mechanism DPE机制

Another model interpreted growth and decline in socioeconomic activity as a conflict between cooperators and defectors. The cooperators form networks that lead to prosperity. However, the network is unstable and invasions by defectors intermittently fragment the network, reducing prosperity, until invasions of new cooperators rebuild networks again. Thus prosperity is seen as a dual phase process of alternating highly prosperous, connected phases and unprosperous, fragmented phases.

另一个模型将社会经济活动的增长和衰退解释为合作者和叛逃者之间的冲突。合作者形成了通向繁荣的网络。然而，网络是不稳定的，叛逃者的入侵断断续续地破坏了网络，降低了繁荣，直到新的合作者再次入侵重建网络。因此，繁荣被视为一个高度繁荣、相互联系的阶段和不繁荣、支离破碎的阶段交替的双相过程。

The following features are necessary for DPE to occur. 发生DPE需要以下特性

Underlying network底层网络

In a forest, the landscape can be regarded as a network of sites where trees might grow. Some sites are occupied by living trees; others sites are empty. In the local phase, sites free of trees are few and they are surrounded by forest, so the network of free sites is fragmented. In competition for these free sites, local seed sources have a massive advantage, and seeds from distant trees are virtually excluded. Even if a few isolated trees do find free ground, their population is prevented from expanding by established populations, even if the invaders are better adapted to the local environment. A fire in such conditions leads to an explosion of the invading population, and possibly to a sudden change in the character of the entire forest.

在森林中，景观可以看作是树木生长的场所网络。有些地方被活树占据，有些地方则是空的。在局部阶段，没有树木的站点很少，而且被森林包围，因此无树站点的网络是碎片化的。在对这些无树站点的竞争中，局部种子来源具有巨大的优势，而来自遥远树木的种子几乎被排除在外。即使一些孤立的树木确实找到了可用的土地，它们的种群也会被已建立的种群所阻止，即使入侵者能够更好地适应当地的环境。在这种情况下发生的火灾会导致入侵种群的爆炸，并可能导致整个森林性质的突然变化。

DPE occurs where a system has an underlying network. That is, the system's components form a set of nodes and there are connections (edges) that join them. For example, a family tree is a network in which the nodes are people (with names) and the edges are relationships such as "mother of" or "married to". The nodes in the network can take physical form, such as atoms held together by atomic forces, or they may be dynamic states or conditions, such as positions on a chess board with moves by the players defining the edges.

DPE发生在系统有底层网络的地方。也就是说，系统的组件形成一组节点，并且有连接（边）将它们连接起来。例如，家谱是一个网络，其中节点是人（有名字），边是关系，如“XX的母亲”或“XX的配偶”。网络中的节点可以采取物理形式，例如原子力将原子结合在一起，或者它们可以是动态状态或条件，例如棋盘上的位置，棋手的移动定义了边。

This dual phase process in the landscape explains the consist appearance of pollen zones in the postglacial forest history of North America, Europe, as well as the suppression of widespread taxa, such as beech and hemlock, followed by huge population explosions. Similar patterns, pollen zones truncated by fire-induced boundaries, have been recorded in most parts of the world

景观中的这种双相过程解释了北美、欧洲后冰川时代森林史中花粉带的出现，以及广泛分布的分类群（如山毛榉和铁杉）受到抑制，随后出现巨大的种群爆炸。类似的模式，如花粉带被火引起的边界截断，在世界大部分地区都有记录。

In mathematical terms (graph theory), a graph [math]\displaystyle{ \textstyle G = \langle N,E\rangle }[/math] is a set of nodes [math]\displaystyle{ \textstyle N }[/math] and a set of edges [math]\displaystyle{ \textstyle E \subset \{ (x,y) \mid x,y \in N \} }[/math]. Each edge [math]\displaystyle{ \textstyle (x,y ) }[/math] provides a link between a pair of nodes [math]\displaystyle{ \textstyle x }[/math] and [math]\displaystyle{ \textstyle y }[/math]. A network is a graph in which values are assigned to the nodes and/or edges.

在数学术语（图论）中，图textstyle G=\langle N，E\rangle是一组节点textstyle N和一组边textstyle E\subset\{（x，y）\mid x，y\In N\}。每条边提供一对节点textstyle x和textstyle y之间的链接。网络是一种图形，其中的值分配给节点或边。

Phase shifts相变

Dual phase evolution is a family of search algorithms that exploit phase changes in the search space to mediate between local and global search. In this way they control the way algorithms explore a search space, so they can be regarded as a family of metaheuristic methods.

双相演化算法是利用搜索空间中的相位变化，在局部搜索和全局搜索之间进行调节的一类搜索算法。通过这种方式，它们控制算法探索搜索空间的方式，因此它们可以被看作是一族元启发式方法。

Graphs and networks have two phases: disconnected (fragmented) and connected. In the connected phase every node is connected by an edge to at least one other node and for any pair of nodes, there is at least one path (sequence of edges) joining them. 图和网络有两个阶段：断开（碎片化）和连接。在连接阶段，每个节点通过一条边连接到至少一个其他节点，对于任何一对节点，至少有一条路径（边序列）连接它们

Problems such as optimization can typically be interpreted as finding the tallest peak (optimum) within a search space of possibilities. The task can be approached in two ways: local search (e.g. hill climbing) involves tracing a path from point to point, and always moving "uphill". Global search involves sampling at wide-ranging points in the search space to find high points.

诸如优化之类的问题通常可以解释为在可能的搜索空间内找到最高的峰值（最优）。这个任务可以通过两种方式来完成：局部搜索（例如爬山）包括从一个点到另一个点追踪一条路径，并且总是“上坡”移动。全局搜索包括在搜索空间中的大范围点进行采样，以找到高点。

The Erdős–Rényi model shows that random graphs undergo a connectivity avalanche as the density of edges in a graph increases.<refname="Erdos1960"> Erdős–Rényi模型表明，随着图中边密度的增加，随机图会经历连接性雪崩。

Many search algorithms involve a transition between phases of global search and local search. A simple example is the Great Deluge algorithm in which the searcher can move at random across the landscape, but cannot enter low-lying areas that are flooded. At first the searcher can wander freely, but rising water levels eventually confine the search to a local area. Many other nature-inspired algorithms adopt similar approaches. Simulated annealing achieves a transition between phases via its cooling schedule. The cellular genetic algorithm places solutions in a pseudo landscape in which they breed only with local neighbours. Intermittent disasters clear patches, flipping the system into a global phase until gaps are filled again.

许多搜索算法涉及全局搜索和局部搜索阶段之间的转换。一个简单的例子是大洪水算法，在该算法中，搜索者可以在整个地形上随意移动，但不能进入被洪水淹没的低洼地区。起初，搜索者可以自由漫步，但不断上升的水位最终将搜索限制在局部地区。许多其他受自然启发的算法采用类似的方法。模拟退火通过其冷却制度实现了相变。细胞遗传算法将解决方案放置在一个只与本地邻居繁殖的伪环境中。断断续续的灾难清除了补丁，使系统进入一个全局阶段，直到缺口再次被填补。

{{cite journal

| author = Erdős, P.

Some variations on the memetic algorithm involve alternating between selection at different levels. These are related to the Baldwin effect, which arises when processes acting on phenotypes (e.g. learning) influence selection at the level of genotypes. In this sense, the Baldwin effect alternates between global search (genotypes) and local search (phenotypes).

模因算法的一些变种涉及到在不同层次的选择之间的交替。这些都与鲍德温效应有关，当作用于表型（如学习）的过程影响基因型水平上的选择时，鲍德温效应就会产生。在这个意义上，鲍德温效应在全局搜索（基因型）和局部搜索（表型）之间交替。

| author2 = Rényi, A.

| name-list-style = amp

| year = 1960

| title = On the evolution of random graphs

Dual phase evolution is related to the well-known phenomenon of self-organized criticality (SOC). Both concern processes in which critical phase changes promote adaptation and organization within a system. However, SOC differs from DPE in several fundamental ways. Under SOC, a system's natural condition is to be in a critical state; in DPE a system's natural condition is a non-critical state. In SOC the size of disturbances follows a power law; in DPE disturbances are not necessarily distributed the same way. In SOC a system is not necessarily subject to other processes; in DPE different processes (e.g. selection and variation) operate in the two phases.

双相演化与著名的自组织临界现象有关。两者都涉及关键相变促进系统内适应和组织的过程。然而，SOC与DPE在几个基本方面不同。在SOC下，系统的自然状态是处于临界状态；在DPE中，系统的自然状态是非临界状态。在SOC中，扰动的大小遵循幂律；在DPE中，扰动的分布不一定相同。在SOC中，一个系统不一定要经历其他过程；在DPE中，不同的过程（如选择和变化）在两个阶段中运行。

| journal = Publications of the Mathematical Institute of the Hungarian Academy of Sciences

| volume =5

| pages = 17–61

| url = http://www.renyi.hu/~p_erdos/1960-10.pdf

 | author2-link = Alfréd Rényi

Category:Nature-inspired metaheuristics

类别: 自然启发的启发式元推理

This page was moved from wikipedia:en:Dual-phase evolution. Its edit history can be viewed at 双相演化/edithistory