Normal Lines to Implicit Surfaces

From an Implicit Surface to a Normal Line

An implicit equation describes a surface without necessarily solving for one coordinate in terms of the others. For example, \(x^2+y^2+z^2=25\) describes a sphere. At a suitable point on such a surface, the gradient of its defining function gives a direction perpendicular to the surface. A point and a perpendicular direction determine a line: the normal line.

This construction is useful whenever a surface is given as a level set, such as \(F(x,y,z)=c\). The essential condition is that the gradient at the point is nonzero. We will see why this condition matters, how it leads to a parametric equation, and how to check the result.

Definition: Let \(F:\mathbb{R}^3\to\mathbb{R}\) be differentiable near a point \(P=(x_0,y_0,z_0)\), and let the surface be the level set \(F(x,y,z)=c\). If \(F(P)=c\) and \(\nabla F(P)\ne\mathbf{0}\), then \(P\) is a regular point of this level surface. The normal line at \(P\) is the line through \(P\) whose direction is \(\nabla F(P)\).

Writing \(\nabla F(P)=\langle F_x(P),F_y(P),F_z(P)\rangle\), a parametric equation of the normal line is

\mathbf{r}(t) = \langle x_0,y_0,z_0\rangle + t\langle F_x(P),F_y(P),F_z(P)\rangle, \qquad t\in\mathbb{R}.

In coordinate form, this is

x=x_0+tF_x(P),\qquad y=y_0+tF_y(P),\qquad z=z_0+tF_z(P).

The gradient is evaluated at the specified point, so its components are constants in these line equations. The parameter \(t\) varies along the line. Any nonzero scalar multiple of the gradient gives the same geometric line: changing the direction vector's length changes only the parameterization, not the line itself.

Why the Gradient Gives a Normal Direction

The reason comes from differentiating the level-surface equation along a curve on the surface. A tangent direction at \(P\) can be represented by the velocity of a differentiable curve that stays on the surface and passes through \(P\). Along such a curve, \(F\) remains equal to \(c\), so its rate of change is zero.

Justification: Suppose \(\mathbf{r}(s)\) is a differentiable curve on \(F(x,y,z)=c\), with \(\mathbf{r}(0)=P\). Since \(F(\mathbf{r}(s))=c\), the chain rule gives

0 = \left.\frac{d}{ds}F(\mathbf{r}(s))\right|_{s=0} = \nabla F(P)\cdot\mathbf{r}'(0).

Thus the gradient is perpendicular to every tangent direction obtained from a curve on the surface through \(P\). At a regular point, \(\nabla F(P)\ne\mathbf{0}\), so this nonzero perpendicular vector gives the normal direction. The line through \(P\) in that direction is therefore the normal line.

This argument also explains the two requirements in the definition. The point must actually lie on the level surface, and the gradient must be nonzero there. If the gradient is zero, this particular construction provides no direction; that alone does not establish whether some other geometric description of a normal is possible.

Worked Examples

Worked Example: Normal Line to a Sphere

Find the normal line to the sphere \(x^2+y^2+z^2=25\) at \(P=(3,4,0)\).

Define \(F(x,y,z)=x^2+y^2+z^2\), so the sphere is the level surface \(F=25\). First check that \(P\) is on it:

$$ F(3,4,0)=3^2+4^2+0^2=9+16+0=25. $$

The gradient is

\nabla F(x,y,z)=\langle 2x,2y,2z\rangle, \qquad \nabla F(3,4,0)=\langle 6,8,0\rangle.

Because \(\langle6,8,0\rangle\ne\mathbf{0}\), the point is regular. Use this gradient as the line's direction vector:

\mathbf{r}(t)=\langle3,4,0\rangle+t\langle6,8,0\rangle,

or, in coordinates,

x=3+6t,\qquad y=4+8t,\qquad z=0.

Check the result: at \(t=0\) the line gives \((3,4,0)\), and this point satisfies the sphere equation, as computed above. Its direction vector is exactly \(\nabla F(3,4,0)\). For an additional perpendicularity check, the tangent plane has normal \(\langle6,8,0\rangle\); the line's direction is that same normal vector. The displayed line is therefore the normal line at \(P\).

Worked Example: A Surface with a Linear Term

Find the normal line to the surface \(x^2+2y^2-z=2\) at \(P=(1,-1,1)\).

Set \(G(x,y,z)=x^2+2y^2-z\). Verify membership in the level surface:

$$ G(1,-1,1)=1^2+2(-1)^2-1=1+2-1=2. $$

Differentiate to find the gradient, then evaluate it at \(P\):

\nabla G(x,y,z)=\langle2x,4y,-1\rangle, \qquad \nabla G(1,-1,1)=\langle2,-4,-1\rangle.

This vector is nonzero. The normal line is

\mathbf{r}(t)=\langle1,-1,1\rangle+t\langle2,-4,-1\rangle,

equivalently,

x=1+2t,\qquad y=-1-4t,\qquad z=1-t.

To check the line, set \(t=0\): it passes through \(P\), and \(G(P)=2\), so \(P\) is on the specified surface. Its direction is \(\nabla G(P)\), as required. For a plane-based check, the tangent plane at \(P\) has equation

$$ 2(x-1)-4(y+1)-(z-1)=0. $$

Substituting \(P\) gives \(0\). The plane's normal vector is \(\langle2,-4,-1\rangle\), which agrees with the line's direction vector.

Worked Example: A Product and a Square

Find the normal line to the surface \(xy+z^2=3\) at \(P=(2,1,1)\).

Let \(H(x,y,z)=xy+z^2\). The point lies on the level surface because

$$ H(2,1,1)=(2)(1)+1^2=2+1=3. $$

Compute the gradient component by component:

H_x=y,\qquad H_y=x,\qquad H_z=2z, \qquad \nabla H(2,1,1)=\langle1,2,2\rangle.

The gradient is nonzero, so a parametric equation is

\mathbf{r}(t)=\langle2,1,1\rangle+t\langle1,2,2\rangle,

x=2+t,\qquad y=1+2t,\qquad z=1+2t.

At \(t=0\), the line returns \((2,1,1)\), and substitution into the surface equation gives \(2+1=3\). Its direction vector \(\langle1,2,2\rangle\) equals the gradient at this point. The tangent plane is \(x+2y+2z=6\); substituting \(P\) gives \(2+2+2=6\), and its normal vector agrees with the line direction. These checks confirm the line's point and orientation.

What the Normal Line Does—and Does Not—Describe

A normal line gives a direction perpendicular to the surface at one point. It does not generally describe a path that remains on the surface. For instance, in the second example, taking \(t=1\) in the normal line gives \((3,-5,0)\). At that point, \(x^2+2y^2-z=9+50-0=59\), not \(2\). The normal line passes through the surface point, but its other points need not lie on the surface.

It is also important not to confuse the line's direction with its starting point. The gradient supplies the direction; the given point supplies the anchor. Both are needed to specify the line. For a regular point \(P\) on \(F=c\), a reliable procedure is:

Check the point.
Substitute its coordinates into \(F\) and confirm that the result is \(c\).

Evaluate the gradient.
Find \(\nabla F\), then substitute the coordinates of \(P\).

Check regularity.
Confirm that the resulting gradient vector is not zero.

Write and check the line.
Use \(P+t\nabla F(P)\); verify that \(t=0\) gives \(P\) and that the direction is the evaluated gradient.

A Common Mistake: Using the Gradient at the Wrong Point

A frequent error is to write \(\nabla F(x,y,z)\) as though it were already a fixed direction vector. A line needs a constant direction. The gradient must first be evaluated at the specified point.

For example, for \(F(x,y,z)=x^2+y^2+z^2\) at \(P=(3,4,0)\), an incorrect calculation might write

\mathbf{r}(t)=\langle3,4,0\rangle+t\langle2x,2y,2z\rangle.

This is not the requested parametric line: the proposed direction still depends on \(x,y,z\), which are the coordinates the line is meant to specify. The correction is to evaluate first:

\nabla F(3,4,0)=\langle6,8,0\rangle, \qquad \mathbf{r}(t)=\langle3,4,0\rangle+t\langle6,8,0\rangle.

Now the direction is constant, the line passes through \(P\) when \(t=0\), and the direction is the gradient at \(P\). Always distinguish the gradient field, which varies from point to point, from its value at the point where the normal line is required.

When This Method Applies

The gradient method applies when the surface is expressed as a differentiable level set and the specified point lies on that level set with a nonzero gradient. The level-surface equation alone does not guarantee regularity at every point: the gradient must be checked where the line is requested. If it is zero, do not insert the zero vector as a line direction; the formula would give only the point \(P\), not a line. Also, a nonzero gradient at the specified point does not excuse a membership error: if \(F(P)\ne c\), then \(P\) is not a point on the surface \(F=c\).

The main calculation is compact, but its geometry is precise. The point identifies where the line touches the surface, and the evaluated gradient identifies the perpendicular direction there. Keeping those roles separate makes the parametric equation straightforward to construct and verify.

Check Your Understanding

For each question, identify the point, the level surface, and the gradient information needed before writing a normal line.

For \(F(x,y,z)=x^2+y^2+z^2\), find the normal line at \((0,0,5)\) to the level surface through that point.
What two conditions must be checked at \(P\) before using \(\mathbf{r}(t)=P+t\nabla F(P)\)?
For \(G(x,y,z)=xz+y^2\) at \(P=(1,2,3)\), compute \(\nabla G(P)\) and write the normal line to the level surface through \(P\).
Why does the chain rule imply that the gradient is perpendicular to tangent directions on a level surface?
Does a normal line generally remain on the surface? Explain what the line describes instead.