📚 Gradient Vectors in 3D Space | 三维空间中的梯度向量
In multivariable calculus, the gradient vector is one of the most powerful tools for understanding how a scalar function changes in three-dimensional space. It packages all first-order partial derivatives into a single vector that points in the direction of greatest increase.
在多变量微积分中，梯度向量是理解三维空间中数量函数变化规律的最有力工具之一。它将所有一阶偏导数整合为一个向量，该向量指向函数增长最快的方向。
1. Scalar Fields and Level Surfaces | 数量场与等值面
A scalar field in 3D is a function f(x, y, z) that assigns a single real number to every point in space. Common examples include temperature, pressure, or electric potential. Even though we cannot "see" the function directly, we can visualize it using level surfaces, where the function is constant.
三维空间中的数量场是指对空间中每一点赋予一个实数值的函数 f(x, y, z) 。常见的例子包括温度、气压或电势。虽然我们无法直接“看见”这个函数，但可以通过等值面来可视化它——在等值面上，函数值保持不变。
For a given constant c , the level surface is defined by f(x, y, z) = c . In two dimensions, this would be a level curve; in three dimensions, it is generally a surface.
对于给定常数 c ，等值面由 f(x, y, z) = c 定义。在二维情况下，这对应等值线；在三维空间中，它通常是一个曲面。
2. Partial Derivatives in 3D | 三维中的偏导数
Before defining the gradient, we need the partial derivatives of f . A partial derivative measures the rate of change of f as only one variable changes, while the other two are held fixed.
在定义梯度之前，我们需要先了解 f 的偏导数。偏导数衡量的是当仅有一个变量变化、另外两个变量保持固定时，函数 f 的变化率。
The three first-order partial derivatives are written as:
fx = ∂ f∂ x, fy = ∂ f∂ y, fz = ∂ f∂ z
Each expression is computed using ordinary differentiation rules, treating the other variables as constants.
每个表达式均使用普通求导法则计算，只需将其他变量视为常数即可。
3. Definition of the Gradient Vector | 梯度向量的定义
The gradient of f , denoted ∇ f or grad f , is a vector formed by combining these partial derivatives.
函数 f 的梯度记作 ∇ f 或 grad f ，它是由这些偏导数组合而成的向量。
∇ f = ( ∂ f∂ x, ∂ f∂ y, ∂ f∂ z )
Equivalently, in terms of the standard unit vectors i, j, k :
∇ f = ∂ f∂ x i + ∂ f∂ y j + ∂ f∂ z k
Thus ∇ f is a vector field: at every point in space, it assigns a vector whose components are the three partial derivatives.
因此 ∇ f 是一个向量场：在空间中的每一点，它都赋予一个向量，其分量就是三个偏导数。
4. Constructing the Gradient: A Simple Example | 构造梯度：一个简单例子
Consider the function f(x, y, z) = x2 y + y z3 . We compute each partial derivative separately.
考虑函数 f(x, y, z) = x2 y + y z3 。我们分别计算每个偏导数。
- With respect to x : treat y and z as constants, giving fx = 2xy .
- With respect to y : differentiate x2 y to get x2 , and y z3 to get z3 , so fy = x2 + z3 .
- With respect to z : y z3 gives 3 y z2 , so fz = 3 y z2 .
- 对 x 求偏导：将 y 和 z 视为常数，得到 fx = 2xy 。
- 对 y 求偏导： x2 y 的导数为 x2 ， y z3 的导数为 z3 ，因此 fy = x2 + z3 。
- 对 z 求偏导： y z3 的导数为 3 y z2 ，因此 fz = 3 y z2 。
∇ f = (2xy, x2 + z3, 3 y z2)
At a specific point, such as (1, 2, -1) , the gradient becomes (4, 2, 6) .
在特定点如 (1, 2, -1) 处，梯度变为 (4, 2, 6) 。
5. Directional Derivatives and the Gradient | 方向导数与梯度
The directional derivative measures the rate of change of f in the direction of a unit vector u . It is given by the dot product of the gradient and u .
方向导数衡量的是函数 f 沿单位向量 u 方向的变化率。它由梯度与 u 的点积给出。
Du f = ∇ f · u
Because u is a unit vector, this dot product equals |∇ f| cos θ , where θ is the angle between ∇ f and u .
由于 u 是单位向量，该点积等于 |∇ f| cos θ ，其中 θ 是 ∇ f 与 u 之间的夹角。
This formula shows that the gradient completely determines the directional derivative in every direction.
这一公式表明，梯度完全决定了任意方向上的方向导数。
6. Steepest Ascent and Descent | 最陡上升与最陡下降
Since Du f = |∇ f| cos θ , the largest possible value occurs when cos θ = 1 , that is, when u points in the same direction as ∇ f .
因为 Du f = |∇ f| cos θ ，最大值出现在 cos θ = 1 时，即 u 与 ∇ f 方向相同时。
Therefore, the gradient vector points in the direction of steepest ascent, and its magnitude |∇ f| equals the maximum rate of increase per unit distance. Conversely, -∇ f points in the direction of steepest descent.
因此，梯度向量指向最陡上升方向，其大小 |∇ f| 等于单位距离上的最大增长率。相反， -∇ f 指向最陡下降方向。
This idea is fundamental in optimization algorithms such as gradient descent, where one moves opposite to the gradient to minimize a function.
这一思想是梯度下降等优化算法的基础，即沿梯度的反方向移动以最小化函数。
7. Gradient and Level Surfaces | 梯度与等值面
A crucial geometric property is that ∇ f is always normal (perpendicular) to the level surface f(x, y, z) = c .
一个关键的几何性质是： ∇ f 始终垂直于等值面 f(x, y, z) = c 。
To see why, consider a smooth curve r(t) = (x(t), y(t), z(t)) lying entirely on the level surface. Then f(r(t)) = c , and differentiating with respect to t gives:
为理解这一点，考虑一条完全位于等值面上的光滑曲线 r(t) = (x(t), y(t), z(t)) 。那么 f(r(t)) = c ，对 t 求导得到：
∇ f · r'(t) = 0
Since r'(t) is tangent to the surface, and its dot product with ∇ f is zero, ∇ f must be perpendicular to every tangent direction. Hence ∇ f is a normal vector to the level surface.
由于 r'(t) 与曲面相切，且它与 ∇ f 的点积为零，因此 ∇ f 必须垂直于所有切线方向。所以 ∇ f 是等值面的法向量。
8. Tangent Planes to Level Surfaces | 等值面的切平面
Using the fact that ∇ f is normal to the level surface, we can write the equation of the tangent plane at a point (x0, y0, z0) .
利用 ∇ f 是等值面法向量这一事实，我们可以在点 (x0, y0, z0) 处写出切平面方程。
Let n = ∇ f(x0, y0, z0) . The tangent plane is the set of points (x, y, z) satisfying:
设 n = ∇ f(x0, y0, z0) 。切平面是满足以下条件的所有点 (x, y, z) 的集合：
n · (x - x0, y - y0, z - z0) = 0
Expanded, this gives:
fx (x - x0) + fy (y - y0) + fz (z - z0) = 0
This equation is used extensively in geometry and optimization problems.
该方程在几何与优化问题中被广泛使用。
9. Surfaces of the Form z = f(x, y) | 形如 z = f(x, y) 的曲面
Many surfaces in IB Mathematics are given explicitly as z = f(x, y) . To use the gradient method, rewrite this as a level surface by defining F(x, y, z) = f(x, y) - z = 0 .
IB 数学中的许多曲面以显式形式 z = f(x, y) 给出。为了使用梯度方法，我们定义 F(x, y, z) = f(x, y) - z = 0 ，将其改写为等值面。
Then the normal vector is:
∇ F = ( ∂ f∂ x, ∂ f∂ y, -1 )
Thus the tangent plane at (x0, y0, z0) is:
∂ f∂ x(x0, y0) (x - x0) + ∂ f∂ y(x0, y0) (y - y0) - (z - z0) = 0
This is a very useful formula for quick exam calculations.
这是在考试中非常实用的快速计算公式。
10. Algebraic Properties of the Gradient | 梯度的代数性质
The gradient operator is linear and satisfies several useful rules. Let f and g be scalar functions and c a constant.
梯度算子具有线性性，并满足若干有用法则。设 f 与 g 为数量函数， c 为常数。
| Rule | Formula |
| Linearity | ∇ (f + g) = ∇ f + ∇ g |
| Constant multiple | ∇ (c f) = c ∇ f |
| Product rule | ∇ (f g) = f ∇ g + g ∇ f |
| Chain rule | ∇ f(u) = f'(u) ∇ u |
These properties allow us to compute gradients of complicated expressions without expanding them fully.
这些性质使我们无需完全展开复杂表达式即可计算其梯度。
11. Worked Exam-Style Example | 考试风格例题
Let f(x, y, z) = x2 + y2 + z2 . Find the tangent plane to the level surface f(x, y, z) = 9 at the point (2, 1, 2) .
设 f(x, y, z) = x2 + y2 + z2 。求等值面 f(x, y, z) = 9 在点 (2, 1, 2) 处的切平面。
Step 1: Compute the gradient.
第一步：计算梯度。
∇ f = (2x, 2y, 2z)
Step 2: Evaluate at (2, 1, 2) .
第二步：在 (2, 1, 2) 处取值。
∇ f(2, 1, 2) = (4, 2, 4)
Step 3: Write the tangent plane equation.
第三步：写出切平面方程。
4(x - 2) + 2(y - 1) + 4(z - 2) = 0
Simplifying gives 4x + 2y + 4z = 18 , or equivalently 2x + y + 2z = 9 .
化简得 4x + 2y + 4z = 18 ，即 2x + y + 2z = 9 。
Notice that the normal vector (2, 1, 2) is exactly half of ∇ f , confirming that the gradient is normal to the surface.
注意法向量 (2, 1, 2) 恰好是 ∇ f 的一半，这验证了梯度垂直于曲面。
12. Common Mistakes and Final Advice | 常见错误与最终建议
Mistake 1: Forgetting that the gradient is a vector, not a scalar. Always write all three components, even if one is zero.
错误一：忘记梯度是向量而不是标量。始终写出全部三个分量，即使其中某个分量为零。
Mistake 2: Using the gradient of z = f(x, y) as (fx, fy) instead of (fx, fy, -1) when finding tangent planes in 3D.
错误二：在求三维切平面时，将 z = f(x, y) 的梯度误写为 (fx, fy) ，而正确形式应为 (fx, fy, -1) 。
Mistake 3: Confusing the direction of steepest ascent with steepest descent. The gradient points uphill, not downhill.
错误三：混淆最陡上升与最陡下降方向。梯度指向上升方向，而非下降方向。
Mistake 4: Evaluating the gradient at the wrong point when a question asks for the value along a given direction.
错误四：当题目要求沿给定方向计算时，在错误的点处求梯度。
Mastering the gradient vector is essential for IB Higher Level Mathematics, particularly for topics in vector calculus and optimisation. Practice by rewriting surfaces as level functions and always checking that your normal vector is perpendicular to the tangent plane.
掌握梯度向量对 IB 高级水平数学至关重要，尤其是在向量微积分与优化问题中。练习时将曲面改写为等值函数，并始终检查法向量是否垂直于切平面。
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