Puiseux Series

In the world of mathematics and physics, Taylor series are a cornerstone, allowing us to approximate complex functions with simpler polynomials. However, this powerful tool breaks down at points of singularity—cusps, self-intersections, or branch points—where a function’s behavior becomes unruly and non-analytic. This leaves a critical gap in our ability to analyze some of the most interesting phenomena in nature, from phase transitions to the splitting of quantum energy levels. How can we describe a function's behavior at these difficult points?

This article introduces the Puiseux series, an elegant and powerful extension of the Taylor series that resolves this very problem. By incorporating fractional powers, Puiseux series provide a rigorous framework for understanding the intricate local structure of algebraic functions even at their singularities, revealing a hidden order where traditional methods see only chaos.

We will first delve into the theoretical underpinnings in the ​​Principles and Mechanisms​​ chapter, exploring why Taylor series fail and how the fractional powers of Puiseux series emerge to provide a solution. We will also cover practical methods for constructing these series. Following this, the chapter on ​​Applications and Interdisciplinary Connections​​ will showcase the remarkable utility of Puiseux series, from explaining degeneracy in quantum physics and describing geometric curves to modeling real-world electronic devices. By the end, you will understand not just the mechanics of Puiseux series but also their profound role in uncovering the fundamental structure of the physical and mathematical world.

Imagine you have a powerful microscope. You point it at a smooth, continuous line drawn on a piece of paper. As you zoom in, it looks straighter and straighter. This is the world of well-behaved functions, the kind we first meet in calculus. Their local behavior is beautifully captured by ​​Taylor series​​—an infinite sum of simple integer powers like $(x-a)$, $(x-a)^2$, $(x-a)^3$, and so on. For a physicist or an engineer, this is a marvelous tool. It tells us that, if we look closely enough, nearly everything complicated can be approximated by something simple.

But what happens when our line isn't so well-behaved? What if it crosses itself, or comes to a sharp cusp, or abruptly turns back on itself? If we zoom in on such a point—a ​​singularity​​—the Taylor series machinery breaks down spectacularly. The approximations fail, the derivatives we need might blow up, and our neat, orderly world dissolves. Must we simply give up and declare these points "un-analyzable"? Nature, after all, is full of such interesting behavior—phase transitions, shock waves, the focusing of light into a caustic. To give up would be to ignore some of the most exciting phenomena.

The answer, it turns out, is a resounding "No!" We don't need to abandon the idea of a series expansion; we just need to be more creative. This is the genius of the ​​Puiseux series​​, a concept developed in part by Isaac Newton and later placed on a rigorous footing by Victor Puiseux.

Let's look at a concrete example. An algebraic function is any function $y(z)$ that is the root of a polynomial equation in two variables, say $P(z, y) = 0$. The simplest one you can think of that isn't just a polynomial in $z$ is $y^2 - z = 0$, or $y = \sqrt{z}$.

Try to write a Taylor series for $y(z)=\sqrt{z}$ around the point $z=0$. A Taylor series must look like $y(z) = c_0 + c_1 z + c_2 z^2 + \dots$. The first term, $c_0$, would be $y(0)=0$. The next term, $c_1$, would be the derivative $y'(0)$. But the derivative of $\sqrt{z}$ is $\frac{1}{2\sqrt{z}}$, which goes to infinity as $z$ approaches zero! The entire foundation of the Taylor expansion crumbles. The point $z=0$ is a special kind of singularity known as a ​​branch point​​. If you imagine walking a small circle in the complex plane around $z=0$, when you get back to your starting point (say, the positive real number 4), the value of the function has changed from $\sqrt{4}=2$ to $\sqrt{4}=-2$. You've ended up on a different "branch" of the function.

These branch points are where the multiple solutions of an algebraic equation meet and tangle. Consider the equation $w^3 - 3w + 2z = 0$. For a typical value of $z$, this cubic equation has three distinct solutions for $w$. But at the special point $z=1$, the equation becomes $w^3 - 3w + 2 = (w-1)^2(w+2)=0$. Suddenly, two of the solutions have merged to become $w=1$. This point $z=1$ is a branch point where two function sheets are joined. Any attempt at a standard Taylor series for $w(z)$ around $z=1$ will fail for the same reason it did for $\sqrt{z}$.

Here is the revolutionary idea: What if we amend our series-building toolkit to include ​​fractional powers​​?

Look at $y=\sqrt{z}$ again. The "series" is just $y = z^{1/2}$. It's a series with only one term, and the power is not an integer. And it works perfectly! It exactly describes the function, even at the troublesome origin.

This is the essence of Puiseux's theorem. It tells us that for any algebraic function $y(z)$ defined by $P(z,y)=0$, its behavior near any point $z_0$ can be described by a series of the form: $y(z) = \sum_{k=k_0}^{\infty} c_k (z-z_0)^{k/n}$ This is a Puiseux series. It's just like a Taylor series, but we're allowed to use a "local clock" that ticks not in steps of $(z-z_0)$, but in smaller, fractional steps of $(z-z_0)^{1/n}$ for some integer $n$. The integer $n$ corresponds to the number of branches that get tangled up at the point $z_0$. For $\sqrt{z}$ at $z=0$, two branches are involved (the plus and minus roots), so we use powers of $z^{1/2}$. For the function from problem, two branches also merge at $z=1$, so the theory predicts that the solution should be expressible in powers of $(z-1)^{1/2}$. And indeed it is: the two merging branches can be written as $w(z) = 1 \pm \frac{i\sqrt{6}}{3}(z-1)^{1/2} + \dots$.

This is an incredibly powerful statement. It guarantees that the seemingly chaotic behavior of functions near their singularities has an underlying, beautifully simple structure. No matter how complicated the polynomial $P(z,y)$, the solutions can always be untangled locally using these fractional power series.

This is all very nice, but how do we actually find these series? How do we determine the right fractional powers and the right coefficients? There are two main strategies, much like a detective solving a case.

First, there is the elegant path of deduction and insight. Sometimes, the algebraic structure of the defining equation itself almost shouts the answer at you. Consider the equation from problem: $y^4 - 2x^3y^2 + x^6 - x^7 = 0$. This looks like a complete mess. But a keen eye might notice that the first three terms are a perfect square: $(y^2 - x^3)^2$. So the equation is simply $(y^2 - x^3)^2 = x^7$. Now we can just solve it! $y^2 - x^3 = \pm \sqrt{x^7} = \pm x^{7/2}$ $y^2 = x^3 \pm x^{7/2} = x^3(1 \pm x^{1/2})$ $y = \pm \sqrt{x^3(1 \pm x^{1/2})} = \pm x^{3/2} (1 \pm x^{1/2})^{1/2}$ The fractional powers $x^{3/2}$ and $x^{1/2}$ have appeared naturally! Now we just need to use the old, familiar binomial series for $(1+u)^{1/2}$ where $u = \pm x^{1/2}$ to get all the terms of the Puiseux series. The algebra itself hands us the solution on a silver platter. Another simple example can be seen in problem, where $f(z)^2 = z^2+z$ becomes $f(z) = z^{1/2}(1+z)^{1/2}$, again inviting a binomial expansion.

The second strategy is more of a brute-force approach, but it's universally applicable. We assume a solution of the form $y(t) = \sum c_k t^{k/n}$, plug it into the polynomial equation $P(t,y)=0$, and see what happens. The principle is that the resulting expression must be zero for all values of $t$. This can only happen if the coefficients of each power of $t$ sum to zero. This gives us a system of equations to solve for the unknown coefficients $c_k$.

A beautiful illustration of this is the "method of dominant balance," often used in physics. Let's look at the equation from problem: $y^3 - t^{-2} y^2 + t = 0$. We want to find a solution for $y$ as a series in $t$ when $t$ is very small. We assume the leading term is $y \approx A t^\alpha$. Plugging this in gives three terms with different powers of $t$: $A^3 t^{3\alpha}$, $-A^2 t^{2\alpha-2}$, and $t^1$. For very small $t$, the term with the most negative exponent will dominate. For the equation to hold, at least two of these "most dominant" terms must cancel each other out. This means their exponents must be equal.

• If $3\alpha = 1$, then $\alpha=1/3$. The other exponent is $2\alpha-2 = -4/3$. This is the most negative, so it can't be balanced.
• If $2\alpha-2 = 1$, then $\alpha=3/2$. The other exponent is $3\alpha=9/2$. The $t^1$ term is dominant and unbalanced.
• If $3\alpha = 2\alpha-2$, then ​​$\alpha = -2$​​. The third exponent is $1$. The two dominant terms are those with exponent $-6$. Success! The leading behavior must be governed by the balance between the first two terms. This tells us the leading exponent is $\alpha=-2$. Furthermore, for these terms to cancel, their coefficients must sum to zero: $A^3 t^{-6} - A^2 t^{-6} = 0$, which implies $A^3 - A^2 = 0$. Since we are looking for a non-zero solution, we find $A=1$. The leading term is $y \approx t^{-2}$. We can repeat this process to find subsequent terms, always balancing the next most dominant contribution. This powerful idea lets us systematically build the solution, piece by piece.

This method can handle all sorts of situations, from functions that blow up with negative fractional powers (like in to the asymptotic behavior of a function at infinity, where we use descending powers of $z$ (like $z^{1/3}, z^{-1/3}, \dots$).

A series representation is only useful if we know where it is valid. What is the ​​radius of convergence​​ of a Puiseux series? The answer is beautifully geometric. A Taylor series for a function converges in a disk that extends from its center out to the nearest singularity of the function. The same is true for Puiseux series! The series expansion around a point $z_0$ will converge in a punctured disk whose outer edge is determined by the next closest branch point.

Think of the branch points as poles holding up a circus tent (the function). If you are standing at one pole and describing the shape of the canvas, your description will be accurate until you hit the next pole, where the shape changes dramatically. To find these critical locations, we look for points $z$ where the equation $P(z,y)=0$ has multiple roots for $y$. This happens precisely when the system of two equations, $P(z,y)=0$ and $\frac{\partial P}{\partial y}(z,y)=0$, has a simultaneous solution. By finding these branch points, we can map out the domains where our Puiseux series descriptions are valid.

We have seen that fractional powers work, but the question still nags: why them? What is their fundamental origin? A profound insight comes from thinking about inverse functions.

Let's say we have a normal, well-behaved analytic function, $w = f(z)$. We can ask for its inverse, $z = f^{-1}(w)$. Usually, this is also a well-behaved function. But there's a catch. What if $f(z)$ has a ​​critical point​​ at $z_0$? That is, what if its derivative $f'(z_0)$ is zero? This means that near $z_0$, the function is "flat." For example, if the first non-zero derivative is the $k$-th one, the function behaves like $w - w_0 \approx C(z-z_0)^k$, where $w_0 = f(z_0)$.

Now, try to find the inverse. We need to solve for $z$: $(z-z_0)^k \approx \frac{1}{C}(w-w_0)$ $z-z_0 \approx \left(\frac{1}{C}\right)^{1/k} (w-w_0)^{1/k}$ Look! The fractional power $1/k$ has appeared, not from some complicated algebraic equation, but from the simple act of inverting a function at a point where it was momentarily flat. A critical point of order $k$ for a function $f$ becomes a branch point of order $k$ for its inverse $f^{-1}$. The "flattening" of the forward map requires a "multi-sheeted" inverse map to cover the territory, and the mathematical description of that multi-sheeted structure is the Puiseux series with its fractional powers.

The journey from the failure of Taylor series to the triumph of Puiseux series is a perfect example of mathematical progress. We start with a problem, a place where our tools don't work. By refusing to give up and asking "what if?", we are led to a new, more powerful tool. The Puiseux series is this tool. It tells us that the world of algebraic functions, for all its tangles and apparent complexities, is governed by a remarkable and unified order.

This is more than just a mathematical curiosity. As we saw, the methods for finding Puiseux series by balancing dominant terms are the very heart of ​​perturbation theory​​, one of the most essential tools in all of theoretical physics. From calculating the energy levels of atoms in electric fields to finding the orbits of planets under the influence of other bodies, this idea of starting with a simple solution and systematically adding corrections is fundamental.

Ultimately, the theory gives us a glimpse of a hidden, perfect structure. The collection of all Puiseux series over the complex numbers forms an ​​algebraically closed field​​. This is a fancy way of saying that any polynomial equation whose coefficients are themselves Puiseux series will have all of its roots within the world of Puiseux series. The system is complete; it contains all of its own answers. It's a self-contained universe of functions, capable of describing every possible twist and turn that an algebraic function can take, revealing an inherent beauty and unity in what first appeared to be chaos.

In our journey so far, we have explored the inner workings of Puiseux series, seeing how they generalize the familiar concept of Taylor series to handle the unruly world of algebraic functions near their singular points. We have seen the "what" and the "how." Now, we ask the most exciting questions: "Why?" and "Where?" Why is this mathematical machinery so important, and where does it appear in the grand tapestry of science and engineering?

You see, the universe is full of systems that are poised on a knife's edge. Think of a perfectly symmetrical structure, a set of identical bells, or a quantum system with several states sharing the same energy. These are systems with degeneracy. They possess a special, fragile balance. What happens when you give them a tiny nudge—a small perturbation? The comfortable, simple description often shatters. The familiar language of integer-power Taylor series fails us precisely at these most interesting junctures. It is here, in the landscape of the "almost broken," that Puiseux series emerge not as a mathematical curiosity, but as the indispensable language of physics. They allow us to see the hidden order within the apparent chaos of a singularity.

Perhaps the most profound application of Puiseux series lies in perturbation theory. This is the art of understanding how a system responds to small changes. When a system is degenerate, its response is often dramatic and, crucially, non-analytic.

Imagine a simple physical system whose properties, like energy levels or vibrational frequencies, are given by the eigenvalues of a matrix. What happens if two of these eigenvalues are identical? The system is degenerate. Now, let's introduce a small perturbation, a parameter $t$ that nudges the system away from its perfect state. Consider a matrix like $A(t) = \begin{pmatrix} t^2 t \\ 1 0 \end{pmatrix}$. For $t=0$, its eigenvalues are both zero—a degenerate state. But as soon as we turn on a tiny $t$, the degeneracy is lifted. The two eigenvalues split apart, but not in a way that is proportional to $t$. Instead, they fly apart with a speed proportional to $t^{1/2}$. This characteristic square-root dependence is the fingerprint of a simple degeneracy being broken. The Puiseux series, with its fractional powers, is the natural tool to describe this splitting.

This phenomenon is not just a matrix curiosity; it is central to quantum mechanics. In non-Hermitian quantum systems, which describe open systems that exchange energy with their environment, these points of degeneracy are called ​​Exceptional Points (EPs)​​. At an EP, something even more dramatic than eigenvalue degeneracy occurs: the corresponding eigenvectors also coalesce and become identical. A system at an EP is exquisitely sensitive to perturbations. If we perturb a system near an EP by a small amount $\epsilon$, the eigenvalues split apart as $\epsilon^{1/2}$. But the consequences are even more profound for the system's states. Physical quantities, such as the norm of the spectral projector which describes the "identity" of a state, can diverge as we approach the EP, scaling as $1/|\epsilon|$. This extreme sensitivity, predicted precisely by Puiseux series, is now being harnessed in technologies like ultra-sensitive sensors.

This principle of degeneracy-lifting is remarkably general. It applies not just to matrix eigenvalues but to the roots of any algebraic equation. Suppose you have a triple root, for instance, in the equation $z^3=0$. Now, let's slightly perturb it to $z^3 - \epsilon(z+1) = 0$ for a tiny $\epsilon > 0$. The single root at the origin blossoms into three distinct roots. Where do they go? They don't just shift a bit. They fly out to form a perfect equilateral triangle around the origin, with their distance from the center scaling as $\epsilon^{1/3}$. The exponent $1/3$ is no coincidence; it is the fingerprint of a third-order degeneracy. A $k$-th order degeneracy, when perturbed, will typically split into roots whose distance from the singular point scales as $\epsilon^{1/k}$. The Puiseux series foretells this beautiful, symmetric unfolding.

Puiseux series are also the geometer's secret weapon for describing the behavior of curves at points where they cease to be "nice." This can happen at infinity, or it can happen at a sharp, singular point.

Consider an algebraic curve defined by an equation like $y^3 - 3x^2y + 2x^3 - xy = 0$. We might ask: what does this curve look like for very large values of $x$? One of its "branches" heads off in a direction where $y$ is approximately equal to $x$. But a simple linear asymptote, $y = x+c$, is not a good enough description. The curve bends away from this line. How can we describe this bending trajectory? A Puiseux series in powers of $1/x$ provides the answer. It reveals that the deviation from the line $y=x$ is not constant, but behaves like $C_1 x^{1/2} + C_0 + \dots$. This "parabolic asymptote" gives us a far more accurate roadmap of the curve's journey towards infinity.

Now let's zoom in instead of out. What happens at a sharp corner or a "cusp" on a curve, like the point of a cardioid?. A function that describes the geometry of a smooth curve is analytic. At a cusp, that smoothness is lost, and so is the analyticity. We can use a special tool called the Schwarz function, which encodes the curve's geometry. Near the cusp at $z=0$, this function is no longer a simple power series in $z$. Instead, it reveals its singular nature through a Puiseux series containing terms like $z^{3/2}$. The very presence of these half-integer exponents is the mathematical signature of the geometric singularity, telling us precisely how the curve fails to be smooth at that point.

Finally, let us connect this abstract mathematics to the tangible world of electronics and physical modeling. Many physical systems are governed by differential equations. Sometimes, these equations become singular at a crucial point of operation, making standard analysis impossible.

Consider a simplified model for a resonant tunneling diode, a tiny electronic component whose current-voltage ($I$-$V$) relationship is governed by a differential equation like $\frac{dI}{dV} = \frac{\alpha + \beta V}{I}$. We are most interested in its behavior near zero voltage, where the initial condition is $I(0)=0$. But at this very point $(V,I)=(0,0)$, the equation's right-hand side blows up! We cannot find a standard Taylor series solution. However, if we solve this equation, we discover that the current does not turn on linearly with voltage. Instead, the solution is naturally a Puiseux series, with the leading term being $I(V) \approx \sqrt{2\alpha} V^{1/2}$. This isn't just a mathematical fix; it is a physical prediction. The square-root dependence tells us something fundamental about the device's turn-on characteristics, a non-analytic behavior that is a direct consequence of the physics at low bias.

In a fascinating parallel, the mathematical structure of Puiseux series has also proven to be the natural form for solutions to a modern class of equations known as fractional differential equations (FDEs). These equations involve derivatives of non-integer order, like a "half-derivative." It turns out that when seeking series solutions to FDEs, a fractional power series is often the perfect ansatz, as the fractional derivative operators act very naturally on fractional powers of the variable,.

From the quantum splitting of energy levels to the geometry of a curve's cusp, and from the turn-on voltage of a diode to the solutions of fractional differential equations, Puiseux series emerge as a unifying theme. They are the language of nature at its most critical and interesting points—the singularities. They reveal a hidden, fractional-order world that governs the behavior of systems at the very boundary where our simpler, integer-power descriptions break down, showing us that even at these points of apparent chaos, there is a deep and beautiful mathematical order.