📚 A Numerical Method for Phase Paths | 相路径的数值方法
Phase paths offer a geometric window into the behaviour of dynamical systems without demanding explicit solutions. In IB Mathematics, particularly within the calculus and differential equations options, we often encounter autonomous systems that model phenomena from population ecology to mechanical vibrations. Because analytic solutions are rarely available, numerical methods become essential tools for sketching trajectories in the phase plane. This article walks through the key ideas behind constructing phase paths numerically, from first-order Euler steps to high-precision Runge–Kutta schemes, and illustrates how a few lines of code or careful hand calculation can reveal the long-term fate of a system.
相路径为我们提供了一扇几何之窗，无需显式求解即可观察动力系统的行为。在IB数学中，尤其是在微积分与微分方程选修内容里，我们经常遇到描述从种群生态到机械振动等各类现象的自治系统。由于解析解往往难以获得，数值方法便成为在相平面上勾勒轨迹的重要工具。本文将逐步介绍用数值方法构建相路径的核心思想，从一阶欧拉法到高精度的龙格–库塔格式，并展示如何通过几行代码或细致的手算揭示系统的长期演化趋势。
1. Introduction to Phase Paths | 相路径简介
th, or trajectory, is a curve in the phase plane that represents the evolution of a system' ondition generates a unique phase path. The collection of such paths, together with direction indicators, forms the phase portrait. Phase paths can show equilibrium points, limit cycles, and separatrices — t govern the qualitative dynamics without solving the differential equations analytically.
相路径，或称轨迹，是相平面上表示系统状态(x, y)随时间演化的曲线。对于二维自治系统dx/dt = f(x, y), dy/dt = g(x, y)，每一个初始条件生成一条唯一的相路径。这些路径连同方向指示所构成的整体就是相图。相路径能够展示平衡点、极限环和分界线等特征——这些特征决定了系统的定性行为，而无需解析求解微分方程。
2. Autonomous Systems and Phase Plane | 自治系统与相平面
wherever f(x, y) ≠ 0. This allows us to view trajectories as solutions of a single first-order ODE in the plane. The phase plane thus becomes a state space where every point determines the instantaneous direction of motion, making it natural to approximate curves by stepping along the local vector field.
自治系统不显含自变量t。由于等式的右端只与x和y有关，只要f(x, y) ≠ 0，斜率场dy/dx = g(x, y)/f(x, y)就有明确定义。这使得我们可以将相轨迹视为平面上一阶常微分方程的解。相平面因此成为一个状态空间，其中的每一点都决定了瞬时的运动方向，从而自然地沿着局部向量场逐步推进来逼近曲线。
3. Why Numerical Methods? | 为何需要数值方法？
Analytical solutions of nonlinear systems are rare. Even when a first integral exists, parametric expressions for x(t) and y(t) are often impossible to invert. Numerical methods step through time directly, approximating (x, y) at discrete points. By choosing a small step size h, we can trace a phase path with controlled error. The simplicity of Euler's method and the efficiency of higher-order schemes make them indispensable in mathematical modelling, allowing students to explore dynamics interactively.
非线性系统的解析解十分稀少。即使存在首次积分，往往也无法将x(t)和y(t)的参数表达式反演出来。数值方法直接按时间推进，在离散点上近似得到(x, y)。通过选取较小的步长h，我们就能在可控误差内追踪相路径。欧拉法的简洁与高阶格式的效率使它们在数学建模中不可或缺，也让学生能够交互式地探索动力学行为。
4. Euler's Method for Systems | 欧拉法求解系统
The simplest numerical scheme extends the forward Euler method to a system of two equations. Given a step size h and current point (xₙ, yₙ), we compute the next point as
xₙ₊₁ = xₙ + h · f(xₙ, yₙ)
yₙ₊₁ = yₙ + h · g(xₙ, yₙ)
Graphically, we follow the tangent line for a short interval. Although easy to implement, Euler's method has a local truncation error proportional to h² and accumulates a global error of order h. For phase paths that are sensitive to initial conditions, a very small h may be necessary to avoid spurious behaviour.
最简单的数值格式是将显式欧拉法推广到两个方程的系统中。给定步长 h 和当前点 (xₙ, yₙ)，我们用上述公式计算下一个点。从几何上看，我们沿着切线方向前进一小段距离。欧拉法虽然易于实现，但其局部截断误差与 h² 成正比，全局误差为 h 的一阶精度。对于对初始条件敏感的相路径，可能需要极小的 h 才能避免虚假行为。
5. Improved Euler Method (Heun's Method) | 改进欧拉法（休恩法）
Heun's method, also known as the improved Euler or trapezoidal method, uses an average of slopes to reduce error. For the system we compute two sets of slopes:
k₁₁ = f(xₙ, yₙ), k₁₂ = g(xₙ, yₙ)
k₂₁ = f(xₙ + h, yₙ + h·k₁₂), k₂₂ = g(xₙ + h, yₙ + h·k₁₂)
The new point is then obtained by averaging:
xₙ₊₁ = xₙ + ½h (k₁₁ + k₂₁)
yₙ₊₁ = yₙ + ½h (k₁₂ + k₂₂)
This method is second-order accurate, meaning the global error scales with h². It captures curvature much better than Euler and is often a good balance between simplicity and accuracy when tracing phase paths by hand or with a spreadsheet.
休恩法，亦称改进欧拉法或梯形法，利用斜率的平均值来减小误差。对于系统，我们先计算两组斜率，然后通过平均得到新点。该方法具有二阶精度，即全局误差与 h² 成正比。它比欧拉法能更好地捕捉曲率，而且在手算或使用电子表格追踪相路径时，往往在简单性与准确性之间取得良好平衡。
6. RK4 Method for Phase Paths | 龙格–库塔四阶法
The classical fourth-order Runge–Kutta method (RK4) is a workhorse for numerical integration. For the system it uses four intermediate slope evaluations per step:
k₁ = f(xₙ, yₙ), l₁ = g(xₙ, yₙ)
k₂ = f(xₙ + ½h, yₙ + ½h l₁), l₂ = g(xₙ + ½h, yₙ + ½h l₁)
k₃ = f(xₙ + ½h, yₙ + ½h l₂), l₃ = g(xₙ + ½h, yₙ + ½h l₂)
k₄ = f(xₙ + h, yₙ + h l₃), l₄ = g(xₙ + h, yₙ + h l₃)
The weighted average then updates the state:
xₙ₊₁ = xₙ + (h/6)(k₁ + 2k₂ + 2k₃ + k₄)
yₙ₊₁ = yₙ + (h/6)(l₁ + 2l₂ + 2l₃ + l₄)
With global error of order h⁴, RK4 allows relatively large step sizes while preserving fine details of the phase path. It is the default choice in many computational tools for producing accurate phase portraits.
经典的四阶龙格–库塔方法（RK4）是数值积分的主力。对于系统，每步需要计算四个中间斜率，然后通过加权平均更新状态。该方法的全局误差为 h⁴ 阶，允许使用相对较大的步长，同时能够保持相路径的精细结构。它是许多计算工具中生成精确相图的默认选择。
7. Step-by-Step Algorithm | 逐步算法
To produce a phase path numerically, follow these steps:
1. Write the system in autonomous form: dx/dt = f(x,y), dy/dt = g(x,y).
2. Choose an initial condition (x₀, y₀), a step size h, and a number of steps N.
3. Select the numerical scheme (Euler, Heun, or RK4).
4. For n = 0 to N-1, compute (xₙ₊₁, yₙ₊₁) using the scheme's update formulas.
5. Store each point and, if possible, draw arrows to indicate the direction of increasing t.
6. Optionally, repeat for several initial conditions to sketch the phase portrait.
dle singularities where f(x,y) = 0 by either pausing or changing the integration variable.
要通过数值方法生成相路径，请遵循以下步骤：
1. 将系统写成自治形式：dx/dt = f(x,y), dy/dt = g(x,y)。
2. 选定初始条件 (x₀, y₀)、步长 h 和步数 N。
3. 选择数值格式（欧拉法、休恩法或 RK4）。
4. 对 n = 0 到 N-1，利用所选格式的更新公式计算 (xₙ₊₁, yₙ₊₁)。
5. 储存每一点，如可行，绘制箭头以表示 t 增大的方向。
6. 可选择对多个初始条件重复操作，以勾画完整的相图。
需要注意处理 f(x,y) = 0 的奇点，可通过暂停积分或变换积分变量来解决。
8. Accuracy and Stability | 准确度与稳定性
The accuracy of a phase path depends on both the order of the method and the step size. Higher-order methods like RK4 achieve a given tolerance with fewer steps. Stability is equally important: the numerical method must not amplify small perturbations. For stiff systems, where widely differing time scales exist, explicit methods may require impractically small h. In such cases, implicit methods or specialised solvers are used, but for the typical models encountered in IB (e.g., Lotka–Volterra, pendulum), RK4 or Heun's method with moderate h give excellent results.
相路径的准确性取决于方法的阶数和步长。像 RK4 这样的高阶方法能以更少的步数达到给定容差。稳定性同样重要：数值方法不得放大小的扰动。对于具有差异悬殊的时间尺度的刚性系统，显式方法可能需要小得不切实际的 h。此时会用到隐式方法或专门的求解器，但对于 IB 常遇到的典型模型（如洛特卡–沃尔泰拉模型、单摆），使用适中步长的 RK4 或休恩法都能给出优异的结果。
9. Worked Example: Predator-Prey Model | 实例：捕食者–猎物模型
Consider the classical Lotka–Volterra equations with parameters α = 1.0, β = 0.1, γ = 0.5, δ = 0.02:
dx/dt = x (α − β y), dy/dt = −y (γ − δ x)
Choose initial condition (x₀, y₀) = (40, 20) and step size h = 0.5. Using Heun's method, we obtain the following first few points:
| n | xₙ | yₙ |
|---|---|---|
| 0 | 40.000 | 20.000 |
| 1 | 32.501 | 23.912 |
| 2 | 26.084 | 26.951 |
| 3 | 21.350 | 28.517 |
| 4 | 18.392 | 28.430 |
ize to h = 0.1 gives a smoother closed loop, confirming the validity of the approximation.
考虑经典的洛特卡–沃尔泰拉方程组，参数取 α = 1.0, β = 0.1, γ = 0.5, δ = 0.02。选取初始条件 (x₀, y₀) = (40, 20)，步长 h = 0.5。使用休恩法得到的若干初始点如上表。将 (x, y) 描点可见一条闭合轨道，这正是该保守系统中中性中心的典型特征。将步长改为 h = 0.1 会得到更光滑的闭环，从而验证近似是可靠的。
10. Drawing Phase Portraits with Software | 使用软件绘制相图
While hand calculations build intuition, digital tools allow quick exploration. Software such as GeoGebra, Desmos, or Python with libraries like NumPy and Matplotlib can iterate numerical schemes and plot the paths. In Python, a simple loop implementing RK4 with very small h can generate smooth trajectories that reveal attractors, repellors, and limit cycles. Many IB investigations ask students to compare analytical predictions with numerical phase portraits, reinforcing the importance of understanding the underlying algorithms.
尽管手算能培养直觉，但数字工具能够快速进行探索。GeoGebra、Desmos 或带有 NumPy 和 Matplotlib 等库的 Python 软件可以迭代数值格式并绘制路径。在 Python 中，用一个简单的循环实现 RK4 并以极小的 h 推进，即可生成光滑的轨迹，揭示吸引子、排斥子和极限环。许多 IB 探究课题要求学生将解析预测与数值相图进行比较，从而加深对底层算法的理解。
11. Common Pitfalls | 常见误区
Students often confuse the order of the numerical method with the number of slope evaluations. While RK4 evaluates four slopes per step, it is not simply 'four times Euler'; the weighting greatly improves accuracy. Another pitfall is ignoring the stiffness of the system: using an explicit method with a large h on a stiff system can produce oscillations that do not represent the true dynamics. Always test different step sizes and, if possible, check whether the path closes or approaches an equilibrium consistently. Finally, be mindful of the direction of integration: integrating backward in time can be equally valuable for a full phase portrait.
学生常常将数值方法的阶数与斜率求值次数相混淆。虽然 RK4 每步计算四个斜率，但它并非简单地将欧拉法叠加四次；加权的方式大幅提升了精度。另一个误区是忽略系统的刚性：在刚性系统上以大步长使用显式方法可能产生并不代表真实动力学的振荡。务必尝试不同的步长，并在可能时检查轨道是否一致闭合或趋近平衡点。最后，要注意积分的方向：向后积分对完整的相图同样具有重要价值。
12. Summary | 总结
Numerical methods for phase paths turn an intractable system of differential equations into a sequence of simple arithmetic steps. From Euler's method to RK4, each scheme offers a different compromise between effort and accuracy. By mastering these techniques, IB students can explore fascinating nonlinear dynamics, validate analytic results, and develop a deeper appreciation for the interplay between discrete mathematics and continuous modelling. Whether you are plotting predator-prey cycles or the swing of a pendulum, numerical phase paths are a window into the heart of the system.
用于相路径的数值方法将棘手的微分方程组转化为一系列简单的算术步骤。从欧拉法到 RK4，每种格式都在运算量与精度之间提供了不同的折中。掌握这些技术后，IB 学生可以探索引人入胜的非线性动力学，验证解析结果，并更深刻地领悟离散数学与连续建模之间的相互作用。无论你绘制的是捕食者–猎物循环还是单摆的摆动，数值相路径都是洞察系统核心的一扇窗口。
Published by TutorHao | Mathematics Revision Series | aleveler.com
