Substance-Connection Model (SCM) Overview

• SCM is a mathematically rigorous framework that models substance flow across linear networks with mechanisms for transfer, leakage, and environmental exchange.
• The model yields a generalized family of discrete distributions, integrating classical cases like Poisson, Binomial, and Negative Binomial through specific parameter choices.
• SCM is applied to migration modeling, material transport, and queueing systems, offering practical insights into network connectivity and local transfer dynamics.

The Substance–Connection Model (SCM) is a mathematically rigorous framework for describing the motion, redistribution, and statistical equilibrium of a “substance” traversing a linear chain of nodes (urns) within a network, where each node can exchange substance with adjacent nodes, the broader network, and the environment. SCM yields a generalized family of discrete distributions that incorporates classical distributions as special cases and provides interpretable parameters linking network topology, local transfer processes, and external interactions. Originally developed in the context of migration modeling, SCM has broad applicability in systems characterized by discrete, channel-based flow with local leakage and environmental coupling (Vitanov et al., 2019, Vitanov et al., 2018).

1. Mathematical Structure of the SCM

In the canonical SCM, consider a finite, directed linear chain of $N+1$ urns (nodes), indexed by $i=0,1,\ldots,N$. The substance quantity in node $i$ at discrete time $t_k$ is $x_i(t_k)$. Transfer dynamics are governed by the following components:

• Forward transfer: Substance moves from urn $i$ to $i+1$ at rate $f_i$.
• Backward transfer: Motion from $i$ to $i-1$ allowed at rate $\delta_i$ (often neglected for unidirectional flow, i.e., $\delta_i = 0$).
• Network interaction: Each urn may pump in ($\epsilon_i$) or leak ($\gamma_i$) the substance to/from external network nodes.
• Environmental exchange: Each urn interacts with the environment via supply ($\sigma_i$) and loss ($\mu_i$) rates.

The model’s full master equation at node $i$ is

$$x_i(t_{k+1}) = x_i(t_k) + i^e_i - o^e_i + i^n_i - o^n_i + i^c_i - o^c_i$$

with terms explicitly

$$\begin{aligned} i^e_i &= \sigma_i x_i,\quad o^e_i = \mu_i x_i, \ i^n_i &= \epsilon_i x_i,\quad o^n_i = \gamma_i x_i, \ i^c_i &= \delta_{i+1} x_{i+1},\quad o^c_i = f_i x_i. \end{aligned}$$

Boundary nodes modify these expressions according to their index (e.g., $f_N = 0$, $\delta_0 = 0$) (Vitanov et al., 2019).

2. Stationarity and Recursion Relations

In the stationary regime ($x_i(t_{k+1}) = x_i(t_k) = x_i^{*}$), with all rates independent of time, the balance equations collapse to algebraic relations. With no backward flow ($\delta_i=0$), these simplify to:

For $i=0$: $$(\sigma_0 - \mu_0 - f_0 - \gamma_0 + \epsilon_0) x_0 = 0 \implies f_0 = \sigma_0 - \mu_0 - \gamma_0 + \epsilon_0.$$ For $i=1,\ldots,N$: $$(\mu_i + f_i + \gamma_i - \sigma_i - \epsilon_i)x_i = f_{i-1} x_{i-1}.$$ The general solution is recursive: $$x_i = x_0 \prod_{k=1}^i \frac{f_{k-1}}{\mu_k + f_k + \gamma_k - \sigma_k - \epsilon_k},\quad (i=1,\ldots,N).$$ Normalization over all nodes yields the stationary fraction in node $i$: $y_i = \frac{x_i}{x},\qquad x = \sum_{i=0}^N x_i.$ With the fraction update written as

$$y_0 = \frac{1}{1 + \sum_{l=1}^N\prod_{k=1}^l F_k},\qquad y_i = \frac{\prod_{k=1}^i F_k}{1 + \sum_{l=1}^N\prod_{k=1}^l F_k},\qquad F_k = \frac{f_{k-1}}{\mu_k + f_k + \gamma_k - \sigma_k - \epsilon_k},$$

the entire stationary profile is determined by the sequence $\{F_k\}$ (Vitanov et al., 2019).

3. Emergence of Classical Distributions

By particular choices of $F_k$, SCM synthesizes a broad array of named statistical distributions as degenerate or truncated cases. Some notable forms include:

Family	Fraction Update $F_k$	Distribution Instance (truncated for finite $N$)
Katz	$F_k = \frac{\alpha + \beta(k-1)}{k}$	Poisson (β=0), Binomial, Negative Binomial
Extended-Katz	$F_k = \frac{\alpha + \beta(k-1)}{\gamma+k-1}$	Hyper-Poisson, Hyper-negative-binomial
Sundt–Jewell	$F_k = a + \frac{b}{k}$	—
Ord	Complex polynomial ratio in $k$ (see text)	Ord family
Kemp	Hypergeometric form $\frac{\lambda(a_1+k)\cdots(a_p+k)}{(b_1+k)\cdots}$	Generalized hypergeometric

For finite channels, these yield truncated distributions; for $N\to\infty$, SCM includes the classical Poisson, binomial, negative-binomial, hypergeometric, Waring, Yule, and Hermite families as exact stationary profiles (Vitanov et al., 2019).

4. Extensions: Leakage, Network Interaction, and Two-Substance Systems

SCM’s flexibility encompasses several structural generalizations:

• Network and Environment Coupling: Each node’s substance can leak to or be pumped in from both the broader network ($\epsilon_i$, $\gamma_i$) and environment ($\sigma_i$, $\mu_i$). These parameters directly prescribe nodal inflow and outflow intensities, offering an interpretable control on the equilibrium shape.
• Discrete-Time Formulation: An alternative formulation considers only forward rates ($\alpha_i$) and leakage ($\beta_i$), supplying the entry node with a constant rate $\sigma$. The stationary solution recurses as

$$x_i^* = x_0^* \prod_{j=1}^{i} \frac{\alpha_{j-1}}{\alpha_j + \beta_j},$$

with explicit normalization (Vitanov et al., 2018).

Parameter and equilibrium dependencies include: - Higher forward transfer $\alpha_i$ relative to leakage $\beta_{i+1}$ shifts the stationary mass distribution toward later nodes; - Significant $\beta_i$ at a node concentrates substance there through dominant leakage.

• Two-Substance Systems: SCM further incorporates the interaction between two populations or substances at a node, for example, “migrants” ($x$) and “natives” ($y$). The coupled discrete-time dynamics are:

$$\begin{aligned} y(t+1) &= y(t) + p\,y(t) + q\,y(t)\,x(t), \ x(t+1) &= x(t) + c + r\,x(t) - q\,y(t)\,x(t), \end{aligned}$$

where $q$ is an integration/assimilation rate, and $c$ reflects inflow from the channel. The nontrivial fixed point is

$$x^* = -\frac{p}{q},\qquad y^* = \frac{r\,p-c\,q}{p\,q}.$$

Stability analysis proceeds through linearization and Jacobian spectral radius assessment. Scenario analysis produces regimes of conversion-limited decay, cyclic behavior, or dominance by exogenous inflow (Vitanov et al., 2018).

5. Parameter Interpretation and Empirical Inference

The parameters in SCM directly encode real-world processes:

• Forward rates ($f_i$, $\alpha_i$): Represent the connectivity or permeability of the channel between nodes; higher rates accelerate downstream substance transfer.
• Leakage rates ($\beta_i$, $\gamma_i$, $\mu_i$): Quantify retention or absorption into the background at specific nodes; high values correspond to bottlenecks or significant “escape.”
• External pumping/supply ($\sigma_i$, $\epsilon_i$): Route by which exogenous flux is injected into the channel or the reservoir.

By fitting stationary fractions $y_i$ or probabilities $P(i)$ to observed empirical profiles (e.g., migrant counts across transit states), one infers effective network connection strengths and leakage/exchange parameters. This direct mapping provides a robust framework for parameter identification from data, especially for flows in serial or sequential networks (Vitanov et al., 2019, Vitanov et al., 2018).

6. Applications and Generalizations

SCM has primary applications in modeling networked flows undergoing serial progression with local exchange and environmental interaction. Salient examples:

• Migration Channels: SCM models the spatial distribution of migrants across a chain of countries, with forward transfer, leakage into local society, and external inflows.
• Chemical or Material Transport: Substance transmission along a series of reactors or environmental compartments subject to loss and gain at each stage.
• Queueing and Traffic Models: Translation to occupancy or passage statistics in tandem systems or networks with seepage and exogenous arrivals.

The mathematical framework also admits infinite-chain generalizations, producing entire probability families as equilibrium solutions. Boundary effects, non-trivial topologies, or time-dependent parameters represent natural arenas for further study.

7. Theoretical Significance and Limitations

SCM provides a unifying urn-network formalism whose steady-state solutions include and generalize nearly all classical discrete distributions used in applied probability. All stationary profiles are interpretable as products of local transfer and leakage rates, embedding empirical content in the probabilistic structure.

A plausible implication is that virtually any observed, unimodal distribution across a finite chain subject to serial flow and leakage can be replicated by proper tuning of SCM parameters, with physical interpretation. Limitations include the reliance on time-independent, homogeneous parameters for analytical tractability; highly non-stationary, nonlinear, or feedback-rich environments may require further extension or numerical techniques.

The theoretical integration of parameterized network flows with discrete probabilistic families distinguishes SCM as a comprehensive statistical–mechanistic platform for channel-based systems (Vitanov et al., 2019, Vitanov et al., 2018).