Regular Values

In the quest to understand complex systems, from the shape of a molecule to the dynamics of the cosmos, scientists and mathematicians seek principles that reveal underlying simplicity and order. One of the most elegant and far-reaching of these principles is the distinction between regular and critical values. This concept provides a powerful lens for dissecting intricate structures into manageable pieces, much like a CT scanner builds a 3D image from simple 2D slices. It addresses the fundamental problem of how to describe the geometry of complex shapes and the stability of physical systems in a rigorous yet intuitive way.

This article will guide you through this foundational idea. We will first explore the mathematical heart of the topic in "Principles and Mechanisms," defining regular and critical points and unveiling the two cornerstone results that give them their power: the Preimage Theorem and Sard's Theorem. Then, in "Applications and Interdisciplinary Connections," we will embark on a tour of the scientific world to witness how this abstract mathematical concept provides a unifying framework for understanding everything from visual perception and chemical reactions to the clockwork of the solar system.

Imagine you are an ant exploring a vast, rolling landscape. Your world is a surface, and your altitude at any given point is determined by a simple rule. As you walk around, you notice that in most places, the ground is tilted. There is a clear "uphill" and "downhill." But occasionally, you find special spots: the very top of a hill, the bottom of a basin, or a perfectly level pass between two mountains. At these locations, the ground is momentarily flat in all directions.

This simple picture holds the key to understanding one of the most elegant and powerful ideas in modern geometry: the distinction between ​​regular values​​ and ​​critical values​​. These concepts allow us to understand the intricate structure of complex shapes by slicing them into simpler pieces, much like a CT scanner reconstructs a 3D image from 2D slices.

Let's make our ant's landscape more precise. Imagine a smooth function, let's call it $f$, that takes a point in a space (a ​​manifold​​, $M$) and assigns a number to it. For instance, our space could be a 2-dimensional sphere $S^2$ floating in 3D space, and the function $f(x,y,z) = z$ could simply be its height. Or the space could be our familiar 3D world $\mathbb{R}^3$, and the function could be the squared distance from the origin, $f(x,y,z) = x^2+y^2+z^2$.

The "tilt" of the landscape at a point $p$ is captured by the function's derivative, or ​​differential​​, $df_p$. This is a linear map that tells us how the output value of $f$ changes as we move infinitesimally away from $p$.

A point $p$ is called a ​​critical point​​ if the derivative "vanishes" there—that is, if the landscape is perfectly flat at $p$. For a function mapping to a single number (like our height or distance examples), this means its gradient is the zero vector. For a more general map $f: M^m \to N^n$ from an $m$-dimensional manifold to an $n$-dimensional one, a point $p$ is critical if its differential $df_p$ is not ​​surjective​​. This means the derivative map fails to "cover" all possible directions in the target space's tangent space. A point that is not critical is, naturally, a ​​regular point​​.

Let's look at our examples:

• For the height function $f(x,y,z)=z$ on a sphere, the only points where the "uphill" direction isn't well-defined are the very top (the North Pole) and the very bottom (the South Pole). These are the two critical points.
• For the distance-squared function $f(x,y,z) = x^2+y^2+z^2$ in $\mathbb{R}^3$, the gradient is $(2x, 2y, 2z)$. This is zero only at the origin $(0,0,0)$. The origin is the only critical point—the bottom of a four-dimensional "parabolic bowl".
• On a torus (a donut shape), a typical height function will have four critical points: a maximum (the highest point), a minimum (the lowest), and two saddle points on the inner and outer equators.

The value that the function takes at a critical point is called a ​​critical value​​. For the height function on the sphere, the critical values are $1$ (at the North Pole) and $-1$ (at the South Pole). For the distance-squared function, the only critical value is $f(0,0,0) = 0$. Any number that is not a critical value is called a ​​regular value​​.

So, we've divided all the possible output values of our function into two camps: the rare, special critical values and the vast majority of regular values. What's the payoff for this classification? The answer is a beautiful piece of mathematical machinery called the ​​Preimage Theorem​​ (or Regular Value Theorem).

In essence, the theorem states: ​​The set of all input points that map to a regular value is always a nice, smooth submanifold.​​

What does this mean? It means if you "slice" your original space $M$ by taking all the points $p$ where $f(p)=c$, the resulting shape (called the ​​preimage​​ or ​​level set​​) will be a perfectly smooth, lower-dimensional manifold itself, as long as the value $c$ you chose is a regular one. The theorem even tells us its dimension: $\dim(M) - \dim(N)$.

Let's see this magic at work:

• Consider $f(x,y,z) = x^2+y^2+z^2$. We found that every value except $0$ is regular. What is the preimage of a regular value, say $c=4$? It's the set of points where $x^2+y^2+z^2=4$, which is a sphere of radius $2$. A sphere is a perfect 2-dimensional submanifold of $\mathbb{R}^3$. The theorem predicts the dimension correctly: $3-1=2$.

• Take the height function $f(x,y,z)=z$ on the unit sphere. The regular values are all numbers in $(-1, 1)$. What is the preimage of a regular value like $c=1/3$? It's the set of all points on the sphere with height $z=1/3$. This is a circle of latitude. A circle is a perfect 1-dimensional submanifold. Again, the dimension checks out: $2-1=1$.

The theorem also implicitly warns us that things can get strange at critical values. Consider the surface defined by $F(x, y, z) = (x^2 + y^2 - 5)^2 + z^2 = c$. The critical values turn out to be $c=0$ and $c=25$. For a regular value like $c=1$, we get a smooth torus. But for the critical value $c=25$, the surface is a shape that pinches itself at the origin, which is not a smooth manifold at that point. The Preimage Theorem's guarantee breaks down, just as expected.

This leads to a crucial question. Is this Preimage Theorem a widely applicable tool, or are regular values so rare that we can hardly ever use it? Are we surrounded by critical values, or are they the exceptions?

The astonishing answer comes from ​​Sard's Theorem​​. It states that for any reasonably smooth function, the set of all its critical values has ​​measure zero​​. In simple terms, this means the set of critical values is negligibly small compared to the set of all possible values. If you were to pick a value $c$ from the target space at random, the probability of you hitting a critical value is literally zero.

This is a profoundly deep and useful result. It tells us that "almost every" level set is a beautiful, well-behaved submanifold. The strange, singular shapes that can occur at critical values are infinitely rare. This philosophical assurance becomes a practical tool. For instance, suppose you want to calculate the area of the set of regular values for the map that projects a torus onto a flat plane. You could try to find the critical values and subtract their (potentially complicated) image. Or, you could use Sard's Theorem, which tells you the set of critical values has area zero. Therefore, the area of the regular values is simply the area of the entire projection!.

This idea is also at the heart of more advanced analytical tools like the ​​coarea formula​​. This formula relates an integral over a space to an integral of integrals over its level sets. The formula naturally includes a weighting factor, a Jacobian, which happens to be exactly zero at critical points. In a way, the very mathematics of integration has learned to ignore the negligible set of critical points.

The story doesn't end with slicing by numbers. We can generalize this entire framework to understand how different shapes intersect. Instead of looking at the preimage of a single point (a value $c$), we can ask about the preimage of an entire submanifold $S$ living inside our target space $N$.

This leads to the concept of ​​transversality​​. A map $f: M \to N$ is ​​transverse​​ to a submanifold $S \subset N$ if, at every point of intersection, the tangent spaces of the colliding objects "cooperate" to span the entire ambient tangent space. This is the direct generalization of the derivative being surjective.

When this condition holds, a generalized Preimage Theorem kicks in: the preimage $f^{-1}(S)$ is a nice, smooth submanifold of $M$. And here we see the beautiful unity of mathematics. Our original Preimage Theorem is just a special case! A regular value $c$ is simply a 0-dimensional submanifold. The transversality condition for a map $f$ to a point $c$ reduces exactly to the definition of $c$ being a regular value.

And the grand finale? Sard's Theorem also has a powerful cousin, ​​Thom's Transversality Theorem​​. It ensures that any smooth map can be tweaked ever so slightly—"jiggled" into general position—to become transverse to any given submanifold. This means that "nice" intersections are the norm, and tangled, degenerate intersections are infinitely rare. This principle is a cornerstone of differential topology, the field that studies the properties of shapes that are preserved under smooth deformations. It allows mathematicians to prove profound theorems by first assuming that all intersections are well-behaved, a luxury granted by the remarkable consequences of Sard's theorem.

From the simple picture of an ant on a hill, we have journeyed to the heart of modern geometry, discovering a universal principle: nature, in its smoothest manifestations, is almost always "regular." The singular, "critical" behavior we sometimes observe, while interesting, is the exception, not the rule.

Now that we have tinkered with the machinery of regular and critical values, you might be asking the most important question a scientist can ask: So what? Is this just a clever piece of mathematical classification, a way for mathematicians to neatly sort points into two different boxes? Or does it tell us something profound about the world?

The answer, and I hope this excites you as much as it does me, is that this simple-seeming distinction is one of the most powerful and unifying ideas in modern science. It is a lens that brings into focus the structure of everything from the shadows on a cave wall to the clockwork of the cosmos and the very foundations of our most advanced physical theories. The world, it turns out, is almost everywhere regular, and all the interesting action—the creation, the change, the collapse—happens at the critical points. Let's go on a tour and see for ourselves.

Let's begin with the most immediate experience we have: vision. When you look at an object, say a smooth, polished stone, what is the "outline" or "silhouette" that your brain so effortlessly perceives? You can think of your eye as performing a mathematical operation: a projection map. It takes the three-dimensional surface of the stone and maps it onto the two-dimensional plane of your retina.

A point on the surface is a critical point of this projection if the surface is "edge-on" from your point of view—that is, if your line of sight is perfectly tangent to the surface at that point. The collection of all such points forms a curve on the stone's surface. The image of this curve on your retina is precisely the silhouette! All the other points on the surface are regular points; their images fill in the interior of the silhouette. The silhouette itself is the set of critical values of the projection map.

This perspective immediately explains something remarkable. Sard's Theorem, the central pillar of our discussion, tells us that the set of critical values has "measure zero." In this geometric context, it means that for almost any angle you choose to view the stone from, the silhouette will be a nice, smooth, one-dimensional curve, not a messy, space-filling smudge. The singular views, where the silhouette might cross itself in complex ways, are the rare exception, not the rule. This is the principle behind the crisp outlines we see in the world and the clean "line art" generated by computer graphics programs, which essentially compute these critical sets.

A closely related idea is the "height function," which assigns to each point on a surface its height $z$ above a table. The critical points are the places where the surface is perfectly flat: the local peaks, the bottoms of the valleys, and, most interestingly, the saddle-shaped mountain passes. The heights of these special points are the critical values. All other heights are regular values. If you were to slice the surface horizontally at a regular height, you would get a collection of smooth curves—the familiar contour lines on a topographic map. But if you slice exactly at a critical height, something dramatic happens: a new island appears (a minimum), an island vanishes (a maximum), or two islands merge into one (a saddle). This is a deep insight: critical values are where the topology of the level sets changes. This very idea is the seed of a magnificent field called Morse Theory.

Let's take the idea of a height function and give it a physical meaning. Imagine a molecule, a complex arrangement of atoms held together by chemical bonds. The potential energy of this molecule depends on the relative positions of its atoms. This relationship defines a "potential energy surface," a landscape in a high-dimensional space. The "height" at any point in this landscape is the molecule's potential energy.

Where are the critical points of this energy landscape? They are the points where the net force on every atom is zero—the equilibrium configurations of the molecule. A local minimum is a stable configuration, a state the molecule is happy to rest in. Local maxima and saddle points are unstable equilibria. A saddle point is particularly important; it represents a transition state, the top of an energy hill that the molecule must climb to get from one stable configuration to another. The energy values at these equilibria are the critical values.

A chemist trying to understand a chemical reaction is, in essence, trying to map out this landscape. They want to know the stable states (the minima) and the energy barriers between them (the heights of the saddles). The regular values of the energy function correspond to energies where the molecule is in motion, vibrating and rotating, but not settled in an equilibrium. The study of critical values on these landscapes is fundamental to understanding reaction rates, protein folding, and the very stability of matter. Morse theory, born from the simple height function, becomes a powerful tool for cataloging the states of a physical system and the transitions between them.

The power of our concept truly explodes when we venture into the realm of classical mechanics. The state of any mechanical system—from a pendulum to a planet—can be described by a point in a high-dimensional "phase space." As the system evolves in time, this point traces a path through phase space.

For many idealized systems, like a single planet orbiting a star, there are special quantities that are conserved along any path: energy, angular momentum, and so on. These conserved quantities, or "integrals of motion," can be combined into a single map, $F$, from the phase space to a simpler space like $\mathbb{R}^N$.

Here, the Liouville-Arnold theorem makes a spectacular entrance, and it's all about regular values. It states that if a system has the maximum possible number of independent and compatible integrals of motion (a so-called integrable system), then for any regular value of the map $F$, the portion of phase space corresponding to that set of conserved quantities is not a chaotic mess, but a beautifully simple object: an $N$-dimensional torus. The motion of the system is confined to this torus, where it flows along in a perfectly orderly, quasi-periodic fashion. This is the hidden geometric reason for the predictable, clockwork-like nature of planetary orbits (in simplified models).

And what about the critical values? These are the values of the conserved quantities for which the beautiful torus structure breaks down. They are points of bifurcation, where the qualitative nature of the motion can change dramatically. It is near these critical values that the orderly integrable motion can be destroyed, giving way to the wild unpredictability of chaos. In a deep sense, the distinction between regular and critical values for the integrals of motion is the mathematical embodiment of the distinction between order and chaos in the universe. This same framework helps us understand the behavior of systems from the motion of particles in an accelerator to the dynamics of rotating bodies in space.

So far, we have seen how regular values describe the "typical" behavior of systems. Sard's Theorem gives this idea a rigorous footing: it says the set of critical values is "small" or "thin" (of measure zero). This implies that the set of regular values is "large" and dense. A property that holds for all regular values is said to hold generically. This principle of genericity is not just a philosophical comfort; it is a working tool of immense power.

Imagine you are a physicist or an engineer trying to solve an optimization problem, like minimizing the energy of a system subject to some constraint, say $h(x)=c$. To use the powerful tools of calculus, you need the set of points satisfying the constraint to be a nice, smooth manifold. But what if you were unlucky and chose a $c$ that is a critical value of $h$? Then your constraint set might have nasty singularities—corners, cusps, or self-intersections—where calculus breaks down.

What do you do? You give the constraint a tiny nudge! Sard's Theorem guarantees that you can find a value $c'$ arbitrarily close to $c$ that is a regular value. By the Regular Value Theorem, the new constraint set $h(x)=c'$ is guaranteed to be a smooth manifold, and you can get back to work! This "perturbation" technique, underwritten by the theory of regular values, is a fundamental strategy in modern analysis and is used to lay rigorous foundations for fields from economics to control theory.

This idea reaches its zenith in the most abstract corners of mathematics and theoretical physics. In fields like symplectic geometry, physicists and mathematicians study equations that arise from string theory. To build a coherent theory (like Floer homology), they need to count the solutions to these equations. But this only makes sense if the "space of solutions" is itself a well-behaved manifold. In general, it isn't. The solution? They show that the equations depend on some background "auxiliary data" (a parameter $J_t$). Using an infinite-dimensional version of Sard's Theorem (the Sard-Smale theorem), they prove that for a generic choice of this data, the space of solutions becomes a smooth manifold, and the theory can be built. The existence of regular values, or generic parameters, is the linchpin that holds the entire structure together. Without it, some of the most profound mathematical theories of the last 40 years would not exist.

The influence of this idea extends even further. In linear algebra, one can classify matrices by their invariants, like the trace and determinant. The Regular Value Theorem tells us that the set of all matrices sharing a "generic" set of invariants forms a smooth manifold whose dimension we can compute. This is the starting point for the modern theory of moduli spaces, which are geometric objects that classify other objects.

In complex analysis, the theory that underpins everything from fluid dynamics to quantum mechanics, the derivative of a non-constant analytic function can only be zero at a set of isolated points. This means the set of critical values is at most countable—an even stronger condition than just having measure zero! This implies that the set of regular values is not only open and dense but also connected. You can always draw a path between any two regular values without hitting a critical one. This incredible robustness of "regularity" is a reflection of the deep rigidity and structure of the complex numbers.

From a simple observation about outlines and shadows, we have journeyed to the stability of molecules, the order of the cosmos, the foundations of optimization, and the cutting edge of string theory. The distinction between the common, well-behaved regular values and the rare, transformative critical values is a deep organizing principle of mathematics and nature. It is a beautiful example of how a single, clear idea can illuminate a vast landscape of knowledge, revealing the hidden unity of the scientific world.