Galois Theory: Symmetry Behind Solving Polynomials

Build a working mental model of Galois theory by tracking how root symmetries become groups, how groups match intermediate fields, and how that structure predicts when radicals can exist. You will know what to compute and what the computations mean.

Roots are not just answers to equations. Taken together, they come with allowed shuffles that preserve every rational relation among them. Galois theory is the idea that those shuffles, the symmetries of the roots, are the real object controlling what formulas you can write down. If two polynomials have the same degree, their solvability can still differ because their root symmetries differ. Once you start thinking symmetry first, radicals stop feeling like magic and start feeling like a constraint.

Take a look at how different root symmetries can look even in small degrees.

The point is not that roots permute, but that only certain permutations are compatible with the algebra you can do using rational numbers. The rest of the guide makes that compatibility precise.

Splitting fields are where symmetries live

A polynomial over $\mathbf{Q}$ might have one real root, three real roots, or complex roots, but its symmetry story only stabilizes once all roots are present in one field. The splitting field is the smallest field extension of $\mathbf{Q}$ that contains every root of the polynomial. Smallest matters because it prevents irrelevant symmetries from entering the picture.

To get there, you typically build the field in steps by adjoining one root at a time. Each step is an extension where you allow rational expressions in the new element, then simplify using its defining polynomial relation.

Three practical reasons the smallest field with all roots matters:

• It is the natural domain where every automorphism has to send roots to roots.
• Intermediate fields correspond to partial information, like knowing some symmetric expression of the roots.
• Towers of fields translate into stepwise solvability by radicals later.

Explore what a typical tower looks like for a concrete cubic.

When you see $\mathbf{Q}\subset \mathbf{Q}(\sqrt[3]{2})\subset \mathbf{Q}(\sqrt[3]{2},\omega)$, you are seeing dependencies. Some roots force you to introduce new elements, like a primitive cube root of unity $\omega$, because the other roots cannot be expressed using only the first adjoined element.

The Galois group is symmetry with rules

Galois group means group of field automorphisms of the splitting field that fix the base field, usually $\mathbf{Q}$. Concretely, an automorphism $\sigma$ is a bijection from the splitting field to itself that preserves addition and multiplication and satisfies $\sigma(q)=q$ for all $q\in\mathbf{Q}$.

That last condition is the anchor. Fixing $\mathbf{Q}$ forces $\sigma$ to respect every rational coefficient in the original polynomial, so when $f(x)\in\mathbf{Q}[x]$ and $f(\alpha)=0$, applying $\sigma$ gives

$$0=\sigma(f(\alpha))=f(\sigma(\alpha)).$$

So $\sigma(\alpha)$ is another root. Automorphisms permute the roots, but not arbitrarily. They must preserve all algebraic relations generated over $\mathbf{Q}$.

Explore which root maps are actually legal automorphisms in simple examples.

A good sanity check is this. If you propose a map by sending one root somewhere, it must extend to the whole field in a way that keeps products and sums consistent. Many tempting permutations fail because they break a relation like $\alpha\beta=\text{something rational}$ or because they do not respect how $\omega$ behaves under multiplication.

Consistency test
Any candidate symmetry must preserve every polynomial identity with rational coefficients that your roots satisfy.

The correspondence that makes the theory useful

The Fundamental Theorem of Galois Theory says that for a Galois extension $E/F$, intermediate fields $K$ with $F\subseteq K\subseteq E$ correspond to subgroups $H\le \mathrm{Gal}(E/F)$. The matching is concrete.

Featured snippet version: In a Galois extension $E/F$, each subgroup $H$ of $\mathrm{Gal}(E/F)$ fixes a subfield $E^H=\{x\in E:\sigma(x)=x\ \text{for all }\sigma\in H\}$. Each intermediate field $K$ corresponds to the subgroup $\mathrm{Gal}(E/K)$ of automorphisms that fix $K$ elementwise. Larger subgroups fix smaller fields.

Fixing means elementwise, not setwise. If $H$ is a subgroup, $E^H$ consists of elements that every symmetry in $H$ leaves untouched. If you demand more symmetries to fix something, fewer elements survive, so the fixed field shrinks.

See how this looks in the classic cubic case where the group is $S_3$.

The lattice diagram is more than a picture. It is a control panel. Want an intermediate field of degree $2$ over $\mathbf{Q}$ inside the splitting field. Look for a subgroup of index $2$. Want to know whether a certain expression in roots is rational. Ask whether it is fixed by the whole group.

Normal, separable, and Galois extensions

The correspondence behaves cleanly when the extension is Galois, which bundles two conditions that are easy to mix up.

A field extension $E/F$ is separable if every element of $E$ is a root of a separable polynomial over $F$, meaning its minimal polynomial has no repeated roots. Over characteristic $0$ fields like $\mathbf{Q}$, every algebraic extension is separable, so separability is not the bottleneck in classical polynomial solving.

An extension is normal if every irreducible polynomial in $F[x]$ that has one root in $E$ splits completely in $E$. Splitting fields are normal by design. Normality ensures that automorphisms cannot get stuck. If one root is in the field, all conjugate roots are too, so symmetries can permute them internally.

An extension is Galois if it is both normal and separable. Then $\lvert \mathrm{Gal}(E/F)\rvert=[E:F]$, and the subgroup to subfield matching is as tight as it looks in the theorem.

Memory hook
Separable is about repeated roots. Normal is about having all conjugates. Galois is the combination that turns symmetries into a perfect bookkeeping tool.

Solvable groups and solvable by radicals

Radical formulas correspond to building a splitting field by repeatedly adjoining $n$th roots. The hidden group theory is that each radical step forces the Galois group to have a certain kind of chain of subgroups.

A group is solvable if it has a chain of subgroups where each quotient is abelian. In field terms, solvable by radicals means you can climb from the base field to the splitting field through extensions whose Galois groups are abelian, like cyclic groups coming from adjoining an $n$th root.

This is why the quintic story is not about degree $5$ itself. It is about which groups appear as Galois groups of quintic polynomials. The symmetric group $S_5$ is not solvable, so a generic quintic with Galois group $S_5$ cannot have a general radical formula. Special quintics can, because their Galois groups can be smaller and solvable.

Compare the common small groups that show up and what they imply about radicals.

The key interpretation is not memorizing that $A_5$ blocks radicals. It is learning to read group structure as a proof of impossibility. Once you know the Galois group is $S_5$ or contains $A_5$ in the right way, the radical route is closed, no matter how clever the algebraic manipulation.

Inferring the Galois group from computations

You rarely compute the whole Galois group by explicitly listing automorphisms. Instead, you use invariants and reductions that constrain what the group can be.

Three common tools work together:

• Discriminant: If the discriminant is a square in $\mathbf{Q}$, the Galois group sits inside the alternating group $A_n$. If it is not a square, odd permutations occur, so the group is not contained in $A_n$.
• Resolvents: Auxiliary polynomials whose factorization patterns reveal subgroup structure, often used for cubics and quartics.
• Modulo $p$ factorization: Reduce the polynomial mod a prime $p$ where it has distinct roots. The degrees of factors correspond to cycle types of Frobenius elements, which rules groups in or out.

Explore how changing coefficients changes these signals.

A productive workflow is to treat each computation as a constraint. Discriminant tells you parity information. Mod $p$ patterns tell you which cycle types must exist. A few primes often narrow the possibilities to one group. Then the group answers structural questions, like whether a radical tower is possible, without ever writing a formula for the roots.

Where symmetry pays off beyond radicals

Galois theory keeps showing up because it is a template. Identify the right field, identify its symmetries, then read off what is possible.

Constructible numbers are the cleanest example. A complex number is constructible with straightedge and compass exactly when it lies in a tower of quadratic extensions of $\mathbf{Q}$. Translating that into groups, you are asking for a Galois group that is a $2$-group, built from repeated index $2$ steps. This is why angle trisection fails in general and why certain regular polygons exist and others do not.

Cyclotomic fields give a second laboratory. The field obtained by adjoining a primitive $n$th root of unity has a Galois group isomorphic to $(\mathbf{Z}/n\mathbf{Z})^\times$, turning questions about roots of unity into arithmetic of units mod $n$.

Finite fields show the same idea in a different costume. Extensions of $\mathbf{F}_p$ have cyclic Galois groups generated by the Frobenius map $x\mapsto x^p$, making the symmetry structure extremely explicit.

Next steps that actually stick

Start with examples where you can see every moving part. Quadratics let you match the lone nontrivial automorphism to $\sqrt{d}\mapsto-\sqrt{d}$. Cubics like $x^3-2$ force you to confront why adjoining one root is not enough to get all roots, and why $\omega$ enters.

Then do group theory with a purpose. Learn actions, subgroups, and quotient groups as tools for reading field towers, not as a separate subject. When you return to the correspondence theorem, prove it with a single extension you understand well and use the lattice as your guide. The proof is easier to hold in your head when every symbol has a concrete job.

When you want a concrete project, pick one polynomial family, like $x^4+ax^2+b$, and practice the loop. Compute discriminants, test a few primes, guess the group, then translate that guess into a statement about radicals or intermediate fields. That loop is where the theory becomes skill.