📚 Applications of Differentiation in Mathematical Modeling | 微分在数学模型构建中的应用
Differentiation is one of the most powerful tools in mathematical modeling. It allows us to describe how quantities change, to find optimal values, and to understand the sensitivity of a system to small perturbations. In this article, we explore how derivatives are used to construct, refine, and interpret mathematical models across science, engineering, and economics.
微分是数学建模中最强大的工具之一。它使我们能够描述量如何变化、寻找最优值，并理解系统对微小扰动的敏感性。在本文中，我们探讨导数如何用于构建、改进和解释跨科学、工程和经济学的数学模型。
1. The Role of Derivatives in Modeling | 导数在建模中的作用
Every mathematical model aims to capture the essential behaviour of a real-world system. Derivatives provide the language of change: the rate at which a population grows, the speed of a falling object, or the marginal cost of producing one extra unit. When we write a model as a function f(x) , its derivative f'(x) describes how the output responds to a small change in the input. This local information is then used to predict global behaviour through integration and differential equations.
每个数学模型都旨在捕捉现实世界系统的本质行为。导数提供了变化的语言：种群增长的速度、物体下落的速度，或多生产一单位产品的边际成本。当我们把模型写成一个函数 f(x) 时，其导数 f'(x) 描述了输出对输入微小变化的响应。这种局部信息随后通过积分和微分方程用于预测全局行为。
2. Constructing a Model from a Rate of Change | 从变化率构建模型
In many situations, we know how a quantity changes relative to another variable, and we use this information to build the model. For example, suppose a bacteria population grows at a rate proportional to its current size. If P(t) is the population at time t , this statement translates into the differential equation
在许多情况下，我们知道一个量相对于另一个变量如何变化，并利用这一信息来构建模型。例如，假设细菌种群以与其当前大小成正比的速度增长。如果 P(t) 是时间 t 时的种群数量，这一陈述可转化为微分方程
dPdt = kP
where k is the growth constant. Solving this equation gives P(t)=P0ekt , the exponential growth model. Here the derivative is not just an accessory; it is the very foundation of the model's construction.
其中 k 是增长常数。求解该方程得到 P(t)=P0ekt ，即指数增长模型。这里导数不仅仅是附属品，而是模型构建的根本基础。
3. Using the Derivative to Identify Equilibrium Points | 用导数识别平衡点
For a model described by y' = f(y) , equilibrium points occur where f(y)=0 . Once found, the sign of f'(y) at those points determines stability. If f'(y)<0 , small perturbations decay and the equilibrium is stable; if f'(y)>0 , it is unstable. This analysis is crucial in population dynamics, where equilibria represent sustainable population levels, and in economics, where they represent market-clearing prices.
对于由 y' = f(y) 描述的模型，平衡点出现在 f(y)=0 处。找到后， f'(y) 在这些点的符号决定其稳定性。若 f'(y)<0 ，微小扰动会衰减，平衡是稳定的；若 f'(y)>0 ，则是不稳定的。这一分析在种群动态中至关重要——平衡代表可持续种群水平，在经济学中则代表市场出清价格。
4. Optimization in Engineering Design | 工程设计中的优化
One of the most common uses of differentiation is finding maximum or minimum values of a function. In engineering design, we often need to minimize cost, maximize efficiency, or find the strongest shape under constraints. The classic procedure is to set f'(x)=0 and solve for critical points, then use the second derivative f''(x) to distinguish maxima from minima.
微分最常见的用途之一是求函数的最大值或最小值。在工程设计中，我们常常需要在约束下最小化成本、最大化效率，或寻找最强形状。经典步骤是令 f'(x)=0 解出临界点，然后用二阶导数 f''(x) 区分极大值与极小值。
Example: A cylindrical can with a fixed volume V must be designed to minimize surface area. If the radius is r and height is h , the volume constraint gives h = V/(π r2) . The surface area S = 2π r2 + 2π r h = 2π r2 + 2V/r . Differentiating:
示例：一个固定体积 V 的圆柱罐需要设计以最小化表面积。若半径为 r ，高为 h ，体积约束给出 h = V/(π r2) 。表面积 S = 2π r2 + 2π r h = 2π r2 + 2V/r 。求导：
S'(r) = 4π r - 2V⁄r2 = 0 ⇒ r = (V2π)1/3
This optimal radius yields the classic result h = 2r , a shape that minimizes material cost.
这个最优半径给出了经典结果 h = 2r ，即最小化材料成本的形状。
5. Marginal Analysis in Economics | 经济学中的边际分析
In economics, the derivative of a cost function C(x) is called the marginal cost, denoted MC = C'(x) . It approximates the cost of producing one additional unit. Similarly, marginal revenue MR = R'(x) is the additional revenue from one more unit sold. Profit maximization occurs where MR = MC , provided the second-order condition holds. This is a direct application of differentiation to model producer behaviour.
在经济学中，成本函数 C(x) 的导数称为边际成本，记作 MC = C'(x) 。它近似于多生产一单位产品的成本。类似地，边际收益 MR = R'(x) 是再多销售一单位产品所增加的收入。利润最大化发生在 MR = MC 处，前提是二阶条件成立。这是微分直接应用于生产者行为建模的实例。
6. Sensitivity Analysis and Error Propagation | 敏感度分析与误差传播
Models often depend on parameters that are measured with uncertainty. Differentiation gives us a way to estimate how errors in inputs translate into errors in outputs. If y = f(x) , then a small error Δ x in x produces an approximate error
模型通常依赖于带有不确定性的测量参数。微分为我们提供了一种估计输入误差如何转化为输出误差的方法。若 y = f(x) ，则 x 中的微小误差 Δ x 会产生近似误差
Δ y ≈ f'(x) Δ x
This is the basis of error propagation formulas used in experimental physics and engineering. For example, if T = 2π √L/g is the period of a pendulum, a small error in the length L causes a period error Δ T ≈ (π/√gL)Δ L.
这是实验物理和工程中误差传播公式的基础。例如，若 T = 2π √L/g 是单摆周期，则长度 L 的微小误差引起周期误差 Δ T ≈ (π/√gL)Δ L。
7. Related Rates in Dynamic Models | 动态模型中的相关变化率
When two variables in a model are both functions of time, differentiation with respect to time links their rates of change. For instance, in a gas expanding in a cylinder, the pressure P and volume V satisfy PV = nRT . Differentiating with respect to t gives
当模型中的两个变量都是时间的函数时，对时间求导将它们的变化率联系起来。例如，气缸中膨胀的气体，压力 P 与体积 V 满足 PV = nRT 。对时间求导得到
PdVdt + VdPdt = 0
which allows us to determine one rate from the other. Related-rates problems are essential in monitoring systems where direct measurement of a quantity is difficult.
这使我们能从其中一个变化率确定另一个。相关变化率问题在直接测量某量困难的监测系统中至关重要。
8. Linearization and Local Approximations | 线性化与局部近似
Real-world models are often nonlinear, but near a point of interest we can replace them with a linear approximation using the tangent line: f(x) ≈ f(a) + f'(a)(x-a) . This technique, called linearization, is used in control theory, calibration of instruments, and numerical methods. The error of this approximation is measured by the second derivative via Taylor's theorem.
现实世界的模型通常是非线性的，但在感兴趣的点附近，我们可以用切线将其替换为线性近似： f(x) ≈ f(a) + f'(a)(x-a) 。这种称为线性化的技术用于控制理论、仪器校准和数值方法。该近似的误差通过泰勒定理由二阶导数衡量。
For example, the pendulum equation θ'' + (g/L)sinθ = 0 is nonlinear. For small angles, sinθ≈θ, yielding the linear model θ'' + (g/L)θ=0. Differentiation guides this simplification because the derivative of sinθ at 0 is 1.
例如，单摆方程 θ'' + (g/L)sinθ = 0 是非线性的。对于小角度，sinθ≈θ，得到线性模型 θ'' + (g/L)θ=0。微分指导这种简化，因为 sinθ 在 0 处的导数为 1。
9. Interpreting Graphs of Model Outputs | 解释模型输出图像
Differentiation also helps us visualize and interpret the behaviour of a model. The first derivative f'(x) tells us where a function is increasing or decreasing; the second derivative f''(x) reveals concavity and points of inflection. These features are essential when comparing model predictions with observed data, identifying transitions, or detecting thresholds in a system.
微分还帮助我们可视化和解释模型的行为。一阶导数 f'(x) 告诉我们函数在哪里递增或递减；二阶导数 f''(x) 揭示凹凸性和拐点。这些特征在将模型预测与观测数据比较、识别转变或检测系统阈值时至关重要。
10. Differential Equations as Models of Change | 微分方程作为变化模型
Ultimately, differentiation leads to differential equations, the foundation of most continuous-time models. Newton's law of cooling, dT/dt = -k(T-Ta) , models the temperature of a cooling object. The logistic equation, dP/dt = rP(1-P/K) , models population growth with carrying capacity. In each case, the derivative represents the underlying law of change, and solving the equation yields the model's predictions.
最终，微分引向微分方程——大多数连续时间模型的基础。牛顿冷却定律 dT/dt = -k(T-Ta) 模拟物体的冷却温度。逻辑斯蒂方程 dP/dt = rP(1-P/K) 模拟具有环境承载力的种群增长。在每种情况下，导数代表基本的变化规律，求解方程即得模型的预测。
11. Parameter Estimation Using Derivatives | 利用导数进行参数估计
When fitting a model to data, we often use optimization of a loss function, such as least squares. Differentiation is central to finding the parameter values that minimize the sum of squared errors. For a linear model y = mx + b , the normal equations are obtained by setting the partial derivatives of the error function to zero. For nonlinear models, gradient descent uses derivatives to iteratively update parameters.
在将模型拟合到数据时，我们通常优化损失函数，如最小二乘法。微分是寻找使误差平方和最小的参数值的核心。对于线性模型 y = mx + b ，通过将误差函数的偏导数设为零获得正规方程。对于非线性模型，梯度下降使用导数迭代更新参数。
12. Limitations and Higher-Order Effects | 局限性与高阶效应
While differentiation is extremely powerful, it has limitations. Derivatives only capture local behaviour; models with abrupt changes or discontinuities require careful handling. Furthermore, using only the first derivative neglects curvature and higher-order effects. In such cases, we extend the model using Taylor series, which includes second and higher derivatives to improve accuracy. A good modeler always checks whether the assumptions behind differentiation remain valid.
虽然微分非常强大，但也有局限性。导数只能捕捉局部行为；具有突变或间断的模型需要谨慎处理。此外，仅使用一阶导数忽略曲率和高阶效应。在这种情况下，我们通过泰勒级数扩展模型，包含二阶及更高阶导数以提高精度。一个好的建模者总是检查微分背后的假设是否仍然成立。
Published by TutorHao | Mathematics Revision Series | aleveler.com
更多咨询请联系16621398022（同微信）
或截图后用「扫一扫 → 相册」选择本图