Schalal

城市系统的空间交互

推荐阅读:

  1. 复杂系统的奇遇——随椋鸟飞行,乔治·帕里奇,文铮译
  2. The Structure and Dynamics of Cities – Urban Data Anakysis and Theoretical Modeling
  3. Human mobility: Models and Applications, Physics Reports
  4. 人类行为时空特性的统计力学, 电子科技大学学报
  5. 大数据时代的空间交互分析方法和应用再论, 地理学报

1 动物的移动规律 v.s. 人的移动规律

Rhee l, Shin M, Hong S, Lee K,Kim SJ, Chong S. On the levy-walk nature of human mobility. lEEE Trans on Networking. 2011

Human walk patterns closely follow Levy walk patterns commonly observed in animalssuch as monkeys, birds and jackals.

Levy walks are more diffusive than Brownian motion while less diffusive than random way point.

\[p(l)\sim\frac{1}{l^{1+\alpha}}(0<\alpha<2)\]

The scale-free distribution of flight lengths leads to super-diffusion where mean squared displacement(MSD) is proportional to

\[t^{\gamma}(\gamma>1)\]

44个志愿者的1000小时轨迹数据,关心一下四个变量:

其他参考文献:

2 宏观层面空间交互及人类移动性

重力模型,Gravity Model

  1. 前身:The Laws of Migration, E.G.Ravenstein,发现了移民与国家人口规模、距离之间的关系,指出性别在移民中的影响,但未给出形式化的表达
  2. 雏形:The \(\frac{P_1P_2}{D}\) Hypothesis: on the intercity movement of persons, Zipf
  3. 进化:THE USE OF ENTROPY MAXIMISING MODELS in the Theory of Trip Distribution, Mode Split and Route Split, A.G.Wilson,指出当熵最大时,上述一般模型中的\(\alpha=\beta\)
\[T_{ij}\sim\frac{P_i^{\alpha}P_j^{\beta}}{f(r)}\]
  1. 深度重力模型:A Deep Gravity model for mobility flows generation,Filippo Simini,et al,视作流量分配问题,借鉴了重力模型的形式,输入:人口,距离,OSM(路网,土地利用,poi)
  2. 符号回归(Symbolic Regression):定义函数集合\(P\),\(R\)上的距离函数\(L\),函数复杂度的度量\(C\),数据集\((x_i, y_i)\),求函数\(f^*\)使得:
\[f^*\epsilon\argmin_{f\epsilon F}{\frac{1}{n}\sum^n_{i=1}}{L(y_i,f(x_i))}+\lambda C(f)\]

其中\(F\)由\(P\)中的函数复合得到

即对数据求解合适的函数表达,可以通过Python包PySR进行计算。

中介机会模型,Stouffer’s Law of intervening opportunities

Intervening opportunities: a theory relating mobility and distance, Sociological Review, Samuel Stouffer, 1940

辐射模型(Radiation Model):对重力模型的回应,本质上是中介机会模型,Simini F Gonzalez MC, Maritan A, Barabasi AL. A universal model for mobility and migration patterns. Nature. 2012.

\[T_{ij}=T_i\frac{m_in_j}{(m_i+s_{ij})(m_i+n_j+s_{ij})}\]

其中:

在城市内部区域辐射模型效果不佳。

改进:人口权重模型,Universal predictability of mobility patterns in cities [2014] YanX YY et al.

场论(Field Theory)

Field theory for recurrent mobility. Maxime Lenormand, Pere Colet & José J. Ramasco

Motivation:

limitations

3 微观层面交互及人类移动性模型(人的尺度)

4 城市空间结构与交互

Fujita and Ogawa模型及其简化

Fujita and Ogawa模型,居住在\(i\)的人选择\(j\)地作为工作地是为了最大化以下\(Z_0\)进行决策:

\[Z_0=W(j)-C_R(i)-C_T(i,j)\]

其中:

\(W(j)\):工作地\(j\)的平均工资 \(C_R(i)\): 居住地\(i\)的租金 \(C_T(i,j)\): 通勤成本

Modeling the Polycentric Transition of Cities [2013] Rémi Louf,Marc Barthelemy.:对上述模型进行简化:

得到:

\[Z_{ij}=\eta_j-\frac{d_{ij}}{l}(1+(\frac{T_j}{c})^{\mu})\]

假设原有的城市是单中心的,设原有的center为点1,如果城市发展到一定规模会有subcenter出现,则必有:

\[\eta_j-\frac{d_{ij}}{l}>\eta_1-\frac{d_{i1}}{l}(1+(\frac{P}{c})^{\mu})\]

再假设:

从而得到人口规模的临界值:

\[P^*=c(\frac{l}{LN_c})^{\frac{1}{\mu}}\]

则对次中心的数量\(k\)与城市人口规模的关系有:

\[k\sim (\frac{P}{P^*})^{\frac{\mu}{\mu+1}}\]

改进:

Visitiaton Laws

The universal visitation law of human mobility [2021] Markus Schläpfer, Lei Dong, Geoffrey B. West et al.

  1. rf-scaling law
\[\rho_i(r, f)=\frac{\mu_i}{(rf)^{\eta}}\]

其中:

  1. Conserved effective travel distance per visitor: 家到网格\(i\)的距离\(d_i\)不依赖于吸引性\(\mu_i\)
  2. Spatial clusters with an area distribution that follows Zipf’s Law

解释:P-EPR,在探索过程中加入偏好依附机制

Data and Code

Collective mobility models: Emergence of urban growth patterns from human mobility behavior [2021],对各类模型按照socially independent/interactive 和 memory less / memory-aware 进行分类讨论

Limitation

5 新的一些实证观测结果

  1. 流动模式显著而平稳地发展,个人在任何时候访问的熟悉地点的数量都是一个守恒量,典型的大小为~25:Evidence for a conserved quantity in human mobility [2018] Laura Alessandretti, Piotr Sapiezynski, Sune Lehmann & Andrea Baronchelli et al.
  2. Noether’s Theorem(诺特定理),守恒量与对称性
  3. 局部交互和全局交互的综合:Understanding the mesoscopic scaling patterns within cities [2020] Lei Dong, Zhou Huang, Yu Liu et al.
  4. 应用:e.g.疾病传播:The Hidden Geometry of Complex, Network-Driven Contagion Phenomena [2013] Dirk Brockmann, Dirk Helbing,交通流预测/通过交互估计地点的属性/Social Segregation的度量/可达性与空间优化