🔢 mathematics

Confocal families of plane algebraic curves

This paper investigates families of plane algebraic curves sharing the same foci by reformulating confocality through a focal map on equiclassical families and analyzing its fibers using deformation theory.

Original authors: Ragni Piene, Boris Shapiro

Published 2026-04-29

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Ragni Piene, Boris Shapiro

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are an architect trying to design a family of unique, curved shapes (like the outline of a leaf or a cloud). In the world of geometry, every shape has a secret "skeleton" or a set of hidden anchor points called foci. You might know these from school as the two special points inside an ellipse that define its shape.

This paper asks a very specific question: If I give you a set of these anchor points (foci), can you build a whole family of different curves that all share those exact same points?

Here is the breakdown of what the authors, Ragni Piene and Boris Shapiro, discovered, explained in simple terms.

1. The "Focal Map" Idea

Think of the authors as cartographers. They created a special map called a "Focal Map."

On one side of the map, you have all the possible curves you could draw.
On the other side, you have the specific sets of anchor points (foci).
The map connects a curve to its anchor points.

The authors wanted to see what happens if you stand on a specific spot on the "anchor point" side of the map and look back. Do you see just one curve? A few curves? Or a whole endless family of curves? In math terms, they are studying the "fibers" of this map—essentially, the group of curves that all point to the same set of foci.

2. The Big Surprise: Most Curves Are "Lonely"

The authors found that for most complex shapes (curves with high degrees, like intricate flowers or spirals), the answer is no.

If you pick a random, complicated curve and look at its foci, you generally cannot wiggle that curve into a different shape while keeping those foci the same.
It's like trying to change the shape of a rubber band while keeping two specific pins stuck in it; for most complex shapes, the rubber band is too stiff. It's stuck.
The only time you get a "positive-dimensional family" (a whole range of shapes you can slide through) is for the simplest shapes: conics (like circles, ellipses, and parabolas).

3. The "Class" vs. "Degree" Distinction

The paper makes a crucial distinction between how "wiggly" a curve is (its degree) and how many tangent lines you can draw to it (its class).

Think of degree as the complexity of the recipe.
Think of class as the number of "handles" or anchor points the shape has.
The authors realized that to have a family of curves sharing foci, the shape needs to be very special: it must have a surprisingly low number of "handles" (low class) compared to its complexity.
The Rule: If a curve is too "complex" relative to its number of handles, it cannot belong to a confocal family. It's too rigid.

4. Special Cases: When You Can Find Families

The paper explores specific scenarios where you can find these families:

Rational Curves (The Simple Ones): If you stick to the simplest types of curves (rational curves), you generally only find confocal families for degree 2 (conics). Once you get to degree 3 or higher, the "room" for variation disappears, and you usually get just a finite number of solutions or none at all.
The "Minimal Class" Construction: The authors show that if you take a set of anchor points and ask for the simplest possible curve that fits them, you can actually build a whole family of them.
- Analogy: Imagine you have 3 pins on a board. If you ask for the simplest string shape that touches all 3, there is actually a whole 3-dimensional "cloud" of different string shapes you can make that all share those 3 pins.
Siebeck and Poncelet Curves: The paper looks at famous historical examples (like Siebeck curves and Poncelet polygons).
- The Twist: In these famous examples, the curve is often unique (there is only one). Why? Because the mathematicians who discovered them added extra rules (like "the curve must touch the sides of a triangle").
- The authors clarify: If you only give them the foci (the pins), there is a huge family of curves. But if you add the extra "touching" rules, the family shrinks down to just one single, unique curve.

Summary

The paper is essentially a study of flexibility.

Question: If I lock the anchor points of a curve, how much can I still change the curve's shape?
Answer: For almost all complex curves, you can't change them at all; they are rigid. For simple curves (conics), you have a lot of freedom. For some specific, highly constrained shapes, you might have a little bit of freedom, but often the extra rules used to define them in history actually force them to be unique.

The authors didn't invent a new way to build bridges or model fluids; they simply mapped out the mathematical rules of which shapes can "stretch" while keeping their anchor points fixed, and which ones are stuck in place.

1. Problem Statement

The paper addresses the deformation-theoretic problem of identifying which families of real plane algebraic curves admit nontrivial confocal deformations. Specifically, the authors investigate:

Confocal Families: Positive-dimensional families of curves that share the exact same set of real foci.
The Core Question: Given a set of prescribed real foci, does there exist a continuous family of curves (of a specific degree and class) sharing these foci, or do the foci uniquely determine the curve (or a finite set of curves)?
Historical Context: While foci were classically defined for conics (ellipses, parabolas, hyperbolas) and generalized to higher-degree curves by Plücker in 1832, the systematic study of families of curves sharing foci via deformation theory had not been previously established.

2. Methodology and Framework

The authors employ a combination of classical algebraic geometry, duality theory, and modern deformation theory.

A. Definition of Foci via Duality

The paper utilizes the classical definition of foci based on isotropic lines (lines passing through the circular points at infinity $p_\pm = (1 : \pm i : 0)$ ).

For a curve $C$ of class $c$ , there are $c$ tangents from $p_+$ and $c$ tangents from $p_-$ .
The intersections of these tangents define the foci.
Dual Formulation: Let $G(u, v, w) = 0$ be the equation of the dual curve. The foci are encoded by the polynomials obtained by restricting $G$ to the isotropic lines:
$G_+(w) = G(-1, -i, w) \quad \text{and} \quad G_-(w) = G(-1, i, w)$
For a real curve, $G_-$ is the complex conjugate of $G_+$ . A real focus corresponds to a pair of complex conjugate roots of $G_+$ .

B. The Focal Map

The central conceptual contribution is the introduction of the Focal Map $\Phi$ on equiclassical families.

Let $W_{c,g,d}$ be the equiclassical family of irreducible plane curves of degree $d$ , genus $g$ , and class $c$ .
The map $\Phi$ sends a curve (represented by its dual polynomial $G$ ) to the coefficients of the polynomial $G_+(w)$ .
Key Insight: Two curves are confocal if and only if they map to the same point under $\Phi$ (i.e., they share the same $G_+$ ). Thus, confocal families are the fibers of this map.

C. Dimensional Analysis

The authors analyze the dimension of these fibers using the Fiber-Dimension Formula:
$\dim(\text{Confocal Locus}) = \dim(W) - \text{rank}(d\Phi)$

The expected dimension of the equiclassical family is $d - g + c + 1$ .
Fixing $c$ foci imposes $2c$ real conditions (since each focus fixes a root pair).
The expected dimension of the confocal locus is therefore $d - g - c + 1$ .
Maximal Rank Condition: A positive-dimensional confocal family exists if and only if the differential of the focal map $d\Phi$ does not have maximal rank ( $2c$ ) at a general point.

3. Key Contributions and Results

A. General Existence Criteria

The paper establishes that for general smooth curves of degree $d \geq 3$ , the expected dimension of the confocal locus is negative.

Result: General smooth curves of degree $d \geq 3$ do not admit positive-dimensional confocal families.
Exception: Only conics ( $d=2$ ) among smooth curves admit such families.
Implication: Curves admitting confocal families must be singular and possess an unusually small class relative to their degree.

B. The Rational Case ( $g=0$ )

The authors focus on rational curves, where the inequality $c \leq d$ allows for the possibility of confocal families.

For a rational nodal-cuspidal curve with $\delta$ nodes and $\kappa$ cusps, the class is $c = 2(d-1) - \kappa$ .
The expected dimension of the confocal locus is $3 - d + \kappa$ .
Condition: A positive-dimensional confocal family exists only if $\kappa \geq d - 2$ $κ \geq d - 2$ .
- $d=2$ (Conics): Always admits a 1D family.
- $d=3$ (Cubics): Requires $\kappa \geq 1$ (cuspidal).
- $d \geq 4$ : Requires high numbers of cusps.

C. The Focal Map and Cohomology (Theorems 3.4 & 3.6)

The paper proves that the rank of the focal map is controlled by the cohomology of the normalization of the dual curve.

Theorem 3.4: For rational curves ( $g=0$ ), the focal map has maximal rank ( $2c$ ) at a general smooth point. This implies that for general rational curves, the confocal locus has the expected dimension (often negative or zero).
Conjecture 3.6: The differential of the focal map has maximal rank if and only if the class line bundle $A$ on the normalization of the dual curve is nonspecial (i.e., $H^1(A) = 0$ ).
Proposition 3.7: This conjecture is proven for degrees $d=2, 3$ and singular quartics ( $d=4$ ).

D. Classification of Examples

The paper provides explicit constructions and dimension counts for specific cases:

Rational Curves:
- $d=2$ : 1D confocal family (Conics).
- $d=3$ : Finite set of curves (0D) for general foci.
- $d \geq 4$ : No such curves exist for general foci.
Curves of Minimal Class ( $c$ ):
- Proposition 4.1: For any $c$ distinct real points, there exists a real curve of class $c$ having these points as foci.
- The space of such curves forms an affine space of dimension $\binom{c}{2}$ .
- Conclusion: Prescribed foci never determine a unique curve of minimal class (unless $c=1$ , which is trivial). The foci define a large confocal family.
Specific Examples:
- Conics ( $c=2$ ): 2 foci $\implies$ 1D family.
- Class 3 ( $c=3$ ): 3 foci $\implies$ 3D family of curves.
- Siebeck Curves: These are unique curves determined by a set of points $z_1, \dots, z_n$ (via tangency conditions). However, the paper clarifies that Siebeck curves are not uniquely determined by their foci alone. The foci define a large confocal family, and the Siebeck construction selects a specific member (dimension 0) within that family by imposing extra tangency constraints.
- Poncelet Curves: Complete Poncelet curves of minimal class are uniquely determined by their foci (dimension 0), contrasting with the general minimal-class case where the family is positive-dimensional.

4. Significance

Theoretical Unification: The paper unifies the study of foci with deformation theory, providing a rigorous framework (the focal map) to analyze when foci constrain a curve uniquely versus when they allow for continuous deformation.
Resolution of Uniqueness: It clarifies a common misconception in the literature (e.g., regarding Siebeck curves). While certain geometric constructions yield unique curves for a given set of points, the foci alone generally do not determine a unique curve; they define a high-dimensional family. Uniqueness only arises when additional constraints (like specific tangency conditions or Poncelet properties) are imposed.
Cohomological Insight: By linking the rank of the focal map to the vanishing of $H^1$ of the class line bundle, the paper connects the geometry of foci to deep results in the Brill-Noether theory of algebraic curves.
Classification: It provides a complete dimension count for rational curves and establishes that non-trivial confocal families are rare among smooth curves but abundant among specific singular curves of minimal class.

In summary, Piene and Shapiro demonstrate that confocality is a property of the class of the curve rather than just its degree, and that for most curves, foci are insufficient to fix the curve uniquely; rather, they define a fiber in a high-dimensional moduli space, the dimension of which is governed by the curve's singularities and the cohomology of its dual.