From Surface Parameters to a Tangent Plane
A parametrized surface describes points in space using two parameters. Near a point on a smooth surface, changing one parameter while holding the other fixed traces a curve on the surface; changing the other parameter traces a second curve. The velocities of those curves give two directions lying in the tangent plane.
When the two directions are independent, they determine a plane. A cross product packages the same information in a different form: it gives a vector perpendicular to both directions, which can be used as a normal vector in the plane equation.
At a regular point, the tangent plane is the plane through \(P\) in the two directions \(\mathbf r_u(u_0,v_0)\) and \(\mathbf r_v(u_0,v_0)\). If \(\mathbf n\) is any nonzero normal vector to that plane, its equation is \(\mathbf n\cdot(\mathbf x-P)=0\), where \(\mathbf x=(x,y,z)\) is a general point in space.
The Tangent-Plane Formulas
Write the point and parameter derivatives as vectors in \(\mathbb R^3\). The two derivative vectors themselves give a parametric description of the plane. Their cross product gives a normal-vector description.
The order of the cross product matters for the direction of the normal: \(\mathbf r_v\times\mathbf r_u=-(\mathbf r_u\times\mathbf r_v)\). But either nonzero normal gives the same plane equation, because multiplying the entire equation by \(-1\) does not change its solution set.
Why the Parameter Derivatives Give the Tangent Plane
The connection follows from differentiating curves on the surface. Holding \(v=v_0\) fixed and varying \(u\) traces a parameter curve through \(P\); its velocity at \(u_0\) is \(\mathbf r_u(u_0,v_0)\). Holding \(u=u_0\) fixed and varying \(v\) gives a curve with velocity \(\mathbf r_v(u_0,v_0)\). Thus both vectors are tangent directions.
More generally, suppose a differentiable curve on the patch is described by parameter functions \(u=u(t)\) and \(v=v(t)\), with \(u(t_0)=u_0\) and \(v(t_0)=v_0\). The chain rule gives the curve's velocity at \(t_0\) as a linear combination of those two directions. This shows that every such velocity lies in their span.
Write \(\boldsymbol\gamma(t)=\mathbf r(u(t),v(t))\), where \(\boldsymbol\gamma(t_0)=P\).
At \(t_0\), \(\boldsymbol\gamma'(t_0)=\mathbf r_u(u_0,v_0)u'(t_0)+\mathbf r_v(u_0,v_0)v'(t_0)\).
Every surface-curve velocity lies in the span of the two parameter derivatives, and the coordinate curves supply those two directions. At a regular point they are independent, so their span is a plane.
This argument explains why the construction works, rather than merely providing a calculation rule. Regularity is essential to this parametrization-based method: if the derivatives are dependent, their span is not a two-dimensional plane, and their cross product is zero. The surface itself may still be smooth there, but a different regular parametrization may be needed to find its tangent plane.
Worked Example: Tangent Plane to a Graph Patch
Worked Example: Tangent Plane to a Graph Patch
Find the tangent plane to \(\mathbf r(u,v)=(u,v,u^2+2v^2)\) at the parameter values \((u_0,v_0)=(1,-1)\).
First evaluate the parametrization:
Differentiate with respect to each parameter:
At the specified parameter values, the tangent directions are \(\mathbf r_u=(1,0,2)\) and \(\mathbf r_v=(0,1,-4)\). Their cross product is
Since this vector is nonzero, the point is regular for this parametrization. Use it as the plane normal:
Expanding gives \(-2(x-1)+4(y+1)+(z-3)=0\), or \(-2x+4y+z+3=0\).
Check the result at the point: \(-2(1)+4(-1)+3+3=0\). Also, the normal is perpendicular to each tangent direction: \((-2,4,1)\cdot(1,0,2)=-2+2=0\), and \((-2,4,1)\cdot(0,1,-4)=4-4=0\). Both checks agree with the plane equation.
Worked Example: A Tangent Plane to a Sphere
Worked Example: A Tangent Plane to a Sphere
Use the spherical parametrization \(\mathbf r(\phi,\theta)=(2\sin\phi\cos\theta,2\sin\phi\sin\theta,2\cos\phi)\) to find the tangent plane at \((\phi_0,\theta_0)=(\pi/3,\pi/2)\). Here \(\phi\) is the angle measured down from the positive \(z\)-axis, and \(\theta\) is the angle in the \(xy\)-plane.
The point is
since \(\sin(\pi/3)=\sqrt{3}/2\), \(\cos(\pi/3)=1/2\), \(\cos(\pi/2)=0\), and \(\sin(\pi/2)=1\).
Differentiate in the stated parameter order, first \(\phi\), then \(\theta\):
At the selected parameters these become
Taking the cross product in that same order gives
This is nonzero, so the parametrization is regular at this point. The normal is \((0,3,\sqrt{3})\), and the plane equation is
Check by substitution: \(\sqrt{3}(\sqrt{3})+1-4=3+1-4=0\). The normal also points in the radial direction: \((0,3,\sqrt{3})=\sqrt{3}(0,\sqrt{3},1)\). Reversing the cross-product order would give the negative normal, but the same tangent plane.
Worked Example: A Tangent Plane to a Cylinder
Worked Example: A Tangent Plane to a Cylinder
Consider the radius-\(3\) cylinder parametrized by \(\mathbf r(\theta,z)=(3\cos\theta,3\sin\theta,z)\). Find the tangent plane at \((\theta_0,z_0)=(\pi/3,2)\).
The point on the cylinder is
The parameter derivatives are \(\mathbf r_\theta=(-3\sin\theta,3\cos\theta,0)\) and \(\mathbf r_z=(0,0,1)\). At the specified parameters,
Their cross product is
which is nonzero. Dividing the normal by \(3/2\) gives the simpler normal \((1,\sqrt{3},0)\). Therefore the plane equation is
Substitute the point: \(3/2+\sqrt{3}(3\sqrt{3}/2)=3/2+9/2=6\), as required. The normal is perpendicular to the parameter directions: \((1,\sqrt{3},0)\cdot(-3\sqrt{3}/2,3/2,0)=-3\sqrt{3}/2+3\sqrt{3}/2=0\), and \((1,\sqrt{3},0)\cdot(0,0,1)=0\).
Regularity and a Common Cross-Product Mistake
For a cylinder of radius \(a>0\), the usual parametrization \(\mathbf r(\theta,z)=(a\cos\theta,a\sin\theta,z)\) has derivatives \(\mathbf r_\theta=(-a\sin\theta,a\cos\theta,0)\) and \(\mathbf r_z=(0,0,1)\). Their cross product is \((a\cos\theta,a\sin\theta,0)\), which is nonzero because \(a>0\). By contrast, standard spherical coordinates have \(\mathbf r_\theta=\mathbf 0\) at the poles, where \(\sin\phi=0\). The zero cross product there signals that this particular parametrization is singular; it does not by itself show that the sphere lacks a tangent plane.
A frequent error is to compute the cross product in one order and then use the result for the other order. For example, with \(\mathbf a=(1,0,0)\) and \(\mathbf b=(0,1,0)\), the correct calculation is \(\mathbf a\times\mathbf b=(0,0,1)\), while \(\mathbf b\times\mathbf a=(0,0,-1)\). These are opposite vectors. They produce the same plane normal direction, but only if the chosen vector is used consistently in the plane equation.
Another common error is to omit the point \(P\) when writing the plane. A normal vector alone specifies a direction, not the plane's location. The equation must include the displacement from the point of tangency: \(\mathbf n\cdot(\mathbf x-P)=0\).
Check Your Understanding
Use the parameter derivatives and tangent-plane formulas to answer these questions.
- What condition on \(\mathbf r_u(u_0,v_0)\) and \(\mathbf r_v(u_0,v_0)\) makes a point regular for a parametrization?
- Why does differentiating a curve on a parametrized surface produce a linear combination of the two parameter derivatives?
- If a cross product used as a plane normal is replaced by its negative, does the tangent plane change? Explain.
- For \(\mathbf r(\theta,z)=(a\cos\theta,a\sin\theta,z)\), why is the parametrization regular when \(a>0\)?
- What does a zero cross product tell you, and what does it not necessarily tell you, about the surface?