Quadric Surfaces

Learning Objectives

By the end of this section, you will be able to:

1. Identify and classify the six main types of quadric surfaces from their equations
2. Visualize quadric surfaces in 3D and understand their geometric properties
3. Analyze traces (cross-sections) to understand surface structure
4. Convert between standard form equations and their graphical representations
5. Apply quadric surfaces to real-world problems in physics, engineering, and architecture
6. Connect surface geometry to optimization landscapes in machine learning

The Big Picture: Surfaces in Three Dimensions

"Nature uses only the longest threads to weave her patterns, so each small piece of her fabric reveals the organization of the entire tapestry." — Richard Feynman

In two dimensions, we studied conic sections—ellipses, parabolas, and hyperbolas—which arise from slicing a cone with a plane. In three dimensions, we encounter their natural generalizations: quadric surfaces, which are defined by second-degree polynomial equations in three variables.

These surfaces appear everywhere in the physical world: the shape of planets (ellipsoids), cooling towers (hyperboloids), satellite dishes (paraboloids), and even Pringles chips (hyperbolic paraboloids). They are the simplest curved surfaces after planes, yet rich enough to model countless natural and engineered shapes.

Historical Context

The study of quadric surfaces dates back to the ancient Greeks, who thoroughly investigated conic sections. However, the systematic study of their 3D analogs began with Apollonius of Perga (c. 262 – c. 190 BC), whose work on conics laid the groundwork.

The modern classification came in the 17th century when René Descartes and Pierre de Fermat developed analytic geometry, allowing algebraic equations to represent geometric surfaces. Leonhard Euler in the 18th century provided the complete classification we use today.

Engineering Applications Through History

• Lighthouses (1800s): Parabolic reflectors concentrate light from a source at the focus into a parallel beam
• Radio Telescopes (1930s): Parabolic dishes collect faint radio waves from space
• Cooling Towers (1910s): Hyperboloids of one sheet provide structural strength while using minimal material
• GPS Satellites (1970s): Hyperbolic positioning uses the geometry of hyperboloids

The General Second-Degree Equation

A quadric surface is the set of all points $(x, y, z)$ satisfying a second-degree polynomial equation:

where $A, B, C, D, E, F, G, H, I, J$ are constants and at least one of $A, B, C, D, E, F$ is non-zero (otherwise we'd have a plane).

Classification of Quadric Surfaces

There are exactly six distinct types of quadric surfaces (excluding degenerate cases). Each has a characteristic equation pattern and geometric shape:

Additionally, cylinders are quadric surfaces where one variable is missing entirely from the equation (e.g., $x^2/a^2 + y^2/b^2 = 1$ is an elliptic cylinder extending along the z-axis).

The Ellipsoid

The ellipsoid is the 3D analog of an ellipse—a smooth, closed surface where every point is at a bounded distance from the center:

The constants $a, b, c$ are the semi-axes: the ellipsoid extends from $-a$ to $a$ along the x-axis, and similarly for y and z.

Properties

• All traces are ellipses: Setting any variable to a constant gives an ellipse (or a single point at the extremes)
• When a = b = c: The ellipsoid becomes a sphere of radius a
• Positive Gaussian curvature: The surface curves the same way in all directions (like a ball)
• Volume: $V = \\frac{4}{3}\\pi abc$

Hyperboloids

There are two types of hyperboloids, distinguished by whether they form one connected piece or two separate pieces.

Hyperboloid of One Sheet

This elegant surface looks like a waisted cylinder or hourglass. It has remarkable properties:

• Ruled surface: Can be constructed entirely from straight lines (two families of lines lie on the surface)
• Horizontal traces: Ellipses (smallest at z = 0, the "waist")
• Vertical traces: Hyperbolas
• Negative Gaussian curvature: Curves oppositely in perpendicular directions

Hyperboloid of Two Sheets

This surface consists of two separate bowl-shaped pieces—one opening upward, one downward—with a gap between them.

• No xy-plane trace: Setting z = 0 gives $-x^2/a^2 - y^2/b^2 = 1$, which has no real solutions
• Horizontal traces exist only for |z| ≥ c: Ellipses that grow larger as |z| increases
• Each sheet has positive curvature: Like the inside of a bowl

Paraboloids

Paraboloids are unbounded surfaces that extend infinitely in one direction.

Elliptic Paraboloid

This is a bowl-shaped surface opening upward (or downward if z is negated).

• Horizontal traces: Ellipses (circles when a = b)
• Vertical traces: Parabolas opening upward
• Vertex at origin: The lowest point is (0, 0, 0)
• Positive curvature: Like a satellite dish

Hyperbolic Paraboloid (Saddle Surface)

The saddle surface or "Pringles chip" is one of the most interesting quadric surfaces:

• Traces parallel to xz-plane: Parabolas opening upward
• Traces parallel to yz-plane: Parabolas opening downward
• Horizontal traces: Hyperbolas (or crossing lines at z = 0)
• Negative curvature everywhere: A saddle point at every point!
• Ruled surface: Like the hyperboloid, straight lines lie on it

Elliptic Cones

The elliptic cone consists of two nappes meeting at a single point (the vertex at the origin).

• Horizontal traces: Ellipses (point at z = 0)
• Vertical traces through origin: Two intersecting lines
• Zero Gaussian curvature: The cone can be "unrolled" flat (developable surface)

Cylinders

A cylinder in the context of quadric surfaces is formed when one variable is missing from the equation entirely. The surface extends infinitely along the axis of the missing variable.

In each case, the 2D curve in the xy-plane is simply extruded along the z-axis. The z variable is completely free to take any value.

Traces: Understanding Through Cross-Sections

A trace is the intersection of a surface with a coordinate plane or a plane parallel to one. Traces provide a systematic way to understand 3D surfaces by reducing them to familiar 2D curves.

The Three Types of Traces

• xy-trace: Set z = k (constant). The curve in the plane z = k.
• xz-trace: Set y = k. The curve in the plane y = k.
• yz-trace: Set x = k. The curve in the plane x = k.

Using Traces for Identification

To identify a quadric surface from its equation, analyze its traces:

1. Set each variable to 0 and sketch the resulting 2D curve (if it exists)
2. Set each variable to a few non-zero constants to see how traces change
3. Match the pattern of traces to the classification table

Identifying Quadric Surfaces

Here's a systematic approach to identifying a quadric surface from its equation:

Step-by-Step Process

1. Put in standard form: Complete squares if needed, divide to get 1 on the right side (for ellipsoids and hyperboloids)
2. Count the signs: How many positive, negative, and zero squared terms?
3. Check for missing variables: A missing variable means cylinder
4. Look for equals-zero: Equation = 0 indicates a cone
5. Check for linear z term: z = ... indicates a paraboloid

Quick Reference: Signs Pattern

Real-World Applications

Architecture and Engineering

Physics and Optics

• Ellipsoidal mirrors: Reflect light from one focus to the other focus
• Parabolic reflectors: Focus parallel rays to a point (telescopes, headlights)
• Hyperbolic lenses: Correct spherical aberration in optical systems
• Ellipsoidal cavities: Create uniform electromagnetic fields for scientific instruments

Machine Learning Connections

The geometry of quadric surfaces appears prominently in machine learning, particularly in understanding loss landscapes and optimization dynamics.

Saddle Points in High Dimensions

In the loss landscape of a neural network, the Hessian matrix (second derivatives) at a critical point determines the local geometry:

• All positive eigenvalues: Local minimum (like an elliptic paraboloid bowl)
• All negative eigenvalues: Local maximum (inverted bowl)
• Mixed signs: Saddle point (like a hyperbolic paraboloid)

Gaussian Curvature and Surface Geometry

The Gaussian curvature $K = \\kappa_1 \\cdot \\kappa_2$ (product of principal curvatures) classifies points on any surface:

Python Implementation

Plotting Quadric Surfaces

When a=b=c=1, this produces a unit sphere.

1import numpy as np
2import matplotlib.pyplot as plt
3from mpl_toolkits.mplot3d import Axes3D
4
5def plot_ellipsoid(a=1, b=1, c=1, ax=None):
6    """
7    Plot an ellipsoid: x²/a² + y²/b² + z²/c² = 1
8
9    When a = b = c, this is a sphere.
10    The ellipsoid is the 3D generalization of an ellipse.
11    """
12    u = np.linspace(0, 2 * np.pi, 50)
13    v = np.linspace(0, np.pi, 50)
14
15    # Parametric equations for ellipsoid
16    x = a * np.outer(np.cos(u), np.sin(v))
17    y = b * np.outer(np.sin(u), np.sin(v))
18    z = c * np.outer(np.ones_like(u), np.cos(v))
19
20    if ax is None:
21        fig = plt.figure(figsize=(8, 8))
22        ax = fig.add_subplot(111, projection='3d')
23
24    ax.plot_surface(x, y, z, alpha=0.7, cmap='viridis')
25    ax.set_xlabel('X')
26    ax.set_ylabel('Y')
27    ax.set_zlabel('Z')
28    ax.set_title(f'Ellipsoid: x²/{a}² + y²/{b}² + z²/{c}² = 1')
29    return ax
30
31def plot_hyperboloid_one_sheet(a=1, b=1, c=1, ax=None):
32    """
33    Plot hyperboloid of one sheet: x²/a² + y²/b² - z²/c² = 1
34
35    This is the surface of cooling towers - strong and elegant.
36    It's a "ruled surface" - can be constructed from straight lines.
37    """
38    u = np.linspace(0, 2 * np.pi, 50)
39    v = np.linspace(-2, 2, 50)
40
41    # Parametric equations
42    x = a * np.outer(np.cosh(v), np.cos(u))
43    y = b * np.outer(np.cosh(v), np.sin(u))
44    z = c * np.outer(np.sinh(v), np.ones_like(u))
45
46    if ax is None:
47        fig = plt.figure(figsize=(8, 8))
48        ax = fig.add_subplot(111, projection='3d')
49
50    ax.plot_surface(x, y, z, alpha=0.7, cmap='plasma')
51    ax.set_xlabel('X')
52    ax.set_ylabel('Y')
53    ax.set_zlabel('Z')
54    ax.set_title('Hyperboloid of One Sheet')
55    return ax
56
57def plot_saddle(a=1, b=1, ax=None):
58    """
59    Plot hyperbolic paraboloid: z = x²/a² - y²/b²
60
61    This is the famous "saddle" or "Pringles chip" shape.
62    It has negative Gaussian curvature at every point.
63    Used in architecture (hyperbolic paraboloid roofs).
64    """
65    x = np.linspace(-2, 2, 50)
66    y = np.linspace(-2, 2, 50)
67    X, Y = np.meshgrid(x, y)
68
69    # Explicit equation for saddle
70    Z = X**2 / a**2 - Y**2 / b**2
71
72    if ax is None:
73        fig = plt.figure(figsize=(8, 8))
74        ax = fig.add_subplot(111, projection='3d')
75
76    ax.plot_surface(X, Y, Z, alpha=0.7, cmap='coolwarm')
77    ax.set_xlabel('X')
78    ax.set_ylabel('Y')
79    ax.set_zlabel('Z')
80    ax.set_title(f'Hyperbolic Paraboloid (Saddle): z = x²/{a}² - y²/{b}²')
81    return ax
82
83# Create a comparison figure
84fig = plt.figure(figsize=(15, 5))
85
86# Ellipsoid
87ax1 = fig.add_subplot(131, projection='3d')
88plot_ellipsoid(a=1.5, b=1, c=0.8, ax=ax1)
89
90# Hyperboloid
91ax2 = fig.add_subplot(132, projection='3d')
92plot_hyperboloid_one_sheet(a=1, b=1, c=1, ax=ax2)
93
94# Saddle
95ax3 = fig.add_subplot(133, projection='3d')
96plot_saddle(a=1, b=1, ax=ax3)
97
98plt.tight_layout()
99plt.show()
100
101# Print classification guide
102print("\n=== QUADRIC SURFACE CLASSIFICATION GUIDE ===")
103print("\nGeneral form: Ax² + By² + Cz² + Dxy + Exz + Fyz + Gx + Hy + Iz + J = 0")
104print("\nFor diagonal forms (D=E=F=0):")
105print("• All same signs (+,+,+) or (-,-,-) → Ellipsoid (or sphere)")
106print("• Two same, one opposite (+,+,-) → Hyperboloid of one sheet")
107print("• One positive, two negative (+,-,-) → Hyperboloid of two sheets")
108print("• Two positive, one zero (+,+,0) → Elliptic paraboloid (if linear z term)")
109print("• One positive, one negative, one zero (+,-,0) → Hyperbolic paraboloid")
110print("• Missing variable entirely → Cylinder of that type")

Gaussian Curvature and ML Applications

1import numpy as np
2
3def classify_point_curvature(K):
4    """
5    Classify surface geometry by Gaussian curvature K.
6
7    Gaussian curvature K = κ₁ × κ₂ (product of principal curvatures)
8
9    This connects calculus to differential geometry and tells us
10    about the intrinsic shape of a surface.
11    """
12    if K > 0:
13        return "Elliptic (bowl-like): curves same direction"
14    elif K < 0:
15        return "Hyperbolic (saddle-like): curves opposite directions"
16    else:
17        return "Parabolic (cylinder-like): flat in one direction"
18
19# Quadric surfaces and their Gaussian curvature
20surfaces = {
21    "Ellipsoid": {
22        "equation": "x²/a² + y²/b² + z²/c² = 1",
23        "K_sign": "positive everywhere",
24        "geometry": "Elliptic (like a ball)"
25    },
26    "Hyperboloid (One Sheet)": {
27        "equation": "x²/a² + y²/b² - z²/c² = 1",
28        "K_sign": "negative everywhere",
29        "geometry": "Hyperbolic (saddle at every point)"
30    },
31    "Hyperboloid (Two Sheets)": {
32        "equation": "-x²/a² - y²/b² + z²/c² = 1",
33        "K_sign": "positive everywhere",
34        "geometry": "Elliptic (bowl-shaped pieces)"
35    },
36    "Hyperbolic Paraboloid": {
37        "equation": "z = x²/a² - y²/b²",
38        "K_sign": "negative everywhere",
39        "geometry": "Hyperbolic (saddle)"
40    },
41    "Elliptic Paraboloid": {
42        "equation": "z = x²/a² + y²/b²",
43        "K_sign": "positive everywhere",
44        "geometry": "Elliptic (bowl)"
45    },
46    "Cone": {
47        "equation": "x²/a² + y²/b² - z²/c² = 0",
48        "K_sign": "zero everywhere (except vertex)",
49        "geometry": "Developable (can flatten)"
50    },
51    "Cylinder": {
52        "equation": "x²/a² + y²/b² = 1",
53        "K_sign": "zero everywhere",
54        "geometry": "Developable (can unroll)"
55    }
56}
57
58print("=== GAUSSIAN CURVATURE OF QUADRIC SURFACES ===\n")
59for name, info in surfaces.items():
60    print(f"📐 {name}")
61    print(f"   Equation: {info['equation']}")
62    print(f"   Curvature K: {info['K_sign']}")
63    print(f"   Geometry: {info['geometry']}")
64    print()
65
66# ML Application: Saddle points in optimization
67print("=== SADDLE POINTS IN MACHINE LEARNING ===\n")
68print("The hyperbolic paraboloid is crucial for understanding optimization:")
69print("• Neural network loss landscapes have many saddle points")
70print("• At a saddle: gradient = 0, but NOT a minimum or maximum")
71print("• Hessian has both positive and negative eigenvalues")
72print("• Gradient descent can get stuck near saddle points")
73print("• Modern optimizers (Adam, RMSprop) help escape saddles")

Test Your Understanding

Summary

Quadric surfaces are the natural 3D generalizations of conic sections, defined by second-degree polynomial equations in $x, y, z$. They provide essential models for understanding curved surfaces in calculus and have widespread applications in physics, engineering, and computer science.

The Six Quadric Surfaces

Key Takeaways

1. Classification by signs: The pattern of positive and negative squared terms determines the surface type
2. Traces reveal structure: Cross-sections with coordinate planes show familiar 2D curves
3. Curvature matters: Positive (K > 0), negative (K < 0), or zero curvature gives geometric insight
4. Engineering applications: From cooling towers to satellite dishes to GPS navigation
5. ML connection: Saddle points in loss landscapes are hyperbolic paraboloids locally
6. Missing variable = cylinder: The surface extends infinitely along that axis

Coming Next: In the next section, we'll explore Cylindrical and Spherical Coordinates. These alternative coordinate systems are perfectly suited for describing surfaces with rotational symmetry—making many quadric surfaces easier to express and integrate over.