跳转到内容

A2.2 Solving Equations as Breaking Symmetry

此内容尚不支持你的语言。

A2.2 Solving Equations as Breaking Symmetry

Section titled “A2.2 Solving Equations as Breaking Symmetry”

1. The STEP Problem We Are Trying To Understand

Section titled “1. The STEP Problem We Are Trying To Understand”

STEP III 2017 Q3 begins with a quartic

x4+px3+qx2+rx+s=0x^4+px^3+qx^2+rx+s=0

with roots

α,β,γ,δ.\alpha,\beta,\gamma,\delta.

It then introduces three expressions:

αβ+γδ,\alpha\beta+\gamma\delta, αγ+βδ,\alpha\gamma+\beta\delta, αδ+βγ.\alpha\delta+\beta\gamma.

The problem says these three expressions satisfy a cubic equation of the form

y3+Ay2+(pr4s)y+(4qsp2sr2)=0.y^3+Ay^2+(pr-4s)y+(4qs-p^2s-r^2)=0.

At first, this looks unexpected. Why should a quartic problem produce a cubic? Why these three expressions?

The answer is not a trick specific to this question. It is part of a larger story:

The coefficients see the roots symmetrically. Solving means gradually breaking that symmetry until the individual roots become visible.

Before the quartic, look at degrees 2 and 3.


2. Degree 2: The Discriminant Separates Two Roots

Section titled “2. Degree 2: The Discriminant Separates Two Roots”

Let a quadratic have roots α\alpha and β\beta:

x2sx+p=0.x^2-sx+p=0.

Vieta gives

α+β=s,αβ=p.\alpha+\beta=s,\qquad \alpha\beta=p.

This is symmetric information. If we swap α\alpha and β\beta, the sum and product do not change.

But if we want the roots themselves, we need to separate them.

The difference

αβ\alpha-\beta

is not symmetric. It changes sign when α\alpha and β\beta are swapped.

However, its square is symmetric:

(αβ)2.(\alpha-\beta)^2.

Compute it using Vieta:

(αβ)2=(α+β)24αβ=s24p.(\alpha-\beta)^2=(\alpha+\beta)^2-4\alpha\beta=s^2-4p.

So the discriminant is

Δ=s24p.\Delta=s^2-4p.

The discriminant itself is still symmetric. It has not chosen which root is which. But taking a square root gives

αβ=±Δ.\alpha-\beta=\pm\sqrt{\Delta}.

Now

α=s+Δ2,β=sΔ2,\alpha=\frac{s+\sqrt{\Delta}}2,\qquad \beta=\frac{s-\sqrt{\Delta}}2,

with the two signs swapped if the two roots are renamed.

This is the first model:

The discriminant is a symmetric square. Its square root introduces a sign choice, and that sign choice separates the roots.

That is what it means to say the discriminant breaks the symmetry. The discriminant itself is symmetric; the square root gives access to a non-symmetric difference.


It is too big a jump to go straight from quadratics to quartics. Cubics show a different kind of symmetry-breaking.

Let the roots of a cubic be α,β,γ\alpha,\beta,\gamma. The cubic discriminant has the structural form

Δ3=(αβ)2(βγ)2(γα)2.\Delta_3=(\alpha-\beta)^2(\beta-\gamma)^2(\gamma-\alpha)^2.

This is symmetric in the three roots. If we rename the roots, the product may be rearranged, but its value does not change.

For the depressed cubic

x3+px+q=0,x^3+px+q=0,

a common coefficient form is

Δ3=4p327q2.\Delta_3=-4p^3-27q^2.

The usual cubic formula often contains the related quantity

(q2)2+(p3)3=Δ3108.\left(\frac q2\right)^2+\left(\frac p3\right)^3 =-\frac{\Delta_3}{108}.

Do not worry about using this formula today. The important point is what happens when we take a square root:

Δ3=±(αβ)(βγ)(γα).\sqrt{\Delta_3} =\pm(\alpha-\beta)(\beta-\gamma)(\gamma-\alpha).

This new expression is no longer fully symmetric. If we swap two roots, its sign changes. So the square root of the cubic discriminant distinguishes the two possible orientations of the three roots:

αβγαorαγβα.\alpha\to\beta\to\gamma\to\alpha \qquad\text{or}\qquad \alpha\to\gamma\to\beta\to\alpha.

In plain symmetry language:

six ways to name the three rootstwo possible cyclic orientations.\text{six ways to name the three roots} \quad\longrightarrow\quad \text{two possible cyclic orientations}.

This is already more subtle than the quadratic case. In a quadratic, the square root leaves only a sign choice. In a cubic, after the square-root step there is still a threefold cyclic ambiguity: the roots can still be cycled.

That is why cubic formulas involve cube roots. Cube roots naturally come with three compatible choices, and those choices deal with the remaining cyclic symmetry. Students do not need the full complex-number machinery here. The picture to keep is:

square root of discriminantchooses an orientation,\text{square root of discriminant} \quad\text{chooses an orientation,} cube rootsseparate the remaining threefold cycle.\text{cube roots} \quad\text{separate the remaining threefold cycle.}

A complete cubic formula packages this process into a calculation, but that calculation is not the object of this lesson. The object is the symmetry story:

The cubic discriminant is symmetric; its square root separates the two cyclic orientations of the three roots; cube-root choices handle the remaining threefold ambiguity.

This is enough bridge for today. We are not trying to teach a complete cubic formula. We only need to see why the move from degree 2 to degree 4 cannot be explained by “just take another square root.”


Now return to the quartic in STEP III 2017 Q3.

For four roots, Vieta gives symmetric information:

α+β+γ+δ,\alpha+\beta+\gamma+\delta, αβ+αγ+αδ+βγ+βδ+γδ,\alpha\beta+\alpha\gamma+\alpha\delta+\beta\gamma+\beta\delta+\gamma\delta, αβγ+αβδ+αγδ+βγδ,\alpha\beta\gamma+\alpha\beta\delta+\alpha\gamma\delta+\beta\gamma\delta, αβγδ.\alpha\beta\gamma\delta.

This is not enough to name the four roots.

The next controlled move is to split the four roots into two pairs. There are exactly three ways:

(α,β)(γ,δ),(\alpha,\beta)(\gamma,\delta), (α,γ)(β,δ),(\alpha,\gamma)(\beta,\delta), (α,δ)(β,γ).(\alpha,\delta)(\beta,\gamma).

For these pairings, form

y1=αβ+γδ,y_1=\alpha\beta+\gamma\delta, y2=αγ+βδ,y_2=\alpha\gamma+\beta\delta, y3=αδ+βγ.y_3=\alpha\delta+\beta\gamma.

None of y1,y2,y3y_1,y_2,y_3 is fully symmetric on its own. If the roots are renamed, the three values may swap places.

But the set

{y1,y2,y3}\{y_1,y_2,y_3\}

is stable. Renaming the four roots just rearranges the three possible pairings.

That is why a resolvent cubic appears: it can have these three pairing quantities as its roots. Solving this cubic does not yet give the four roots. It first chooses a pairing structure.


Let

x4+px3+qx2+rx+s=0x^4+px^3+qx^2+rx+s=0

have roots α,β,γ,δ\alpha,\beta,\gamma,\delta. By Vieta,

αβ+αγ+αδ+βγ+βδ+γδ=q.\alpha\beta+\alpha\gamma+\alpha\delta+\beta\gamma+\beta\delta+\gamma\delta=q.

Using the definitions

y1=αβ+γδ,y_1=\alpha\beta+\gamma\delta, y2=αγ+βδ,y_2=\alpha\gamma+\beta\delta, y3=αδ+βγ,y_3=\alpha\delta+\beta\gamma,

we get

y1+y2+y3=q.y_1+y_2+y_3=q.

The resolvent cubic has roots y1,y2,y3y_1,y_2,y_3. Therefore it has the form

(yy1)(yy2)(yy3)=0.(y-y_1)(y-y_2)(y-y_3)=0.

The coefficient of y2y^2 is

(y1+y2+y3)=q.-(y_1+y_2+y_3)=-q.

So in

y3+Ay2+(pr4s)y+(4qsp2sr2)=0,y^3+Ay^2+(pr-4s)y+(4qs-p^2s-r^2)=0,

we must have

A=q.\boxed{A=-q.}

This is Vieta applied one level higher: not to the four original roots, but to the three pairing quantities.


Now take

p=0,q=3,r=6,s=10.p=0,\qquad q=3,\qquad r=-6,\qquad s=10.

The quartic is

x4+3x26x+10=0.x^4+3x^2-6x+10=0.

The resolvent cubic is

y33y240y+84=0.y^3-3y^2-40y+84=0.

Factor:

y33y240y+84=(y7)(y2)(y+6).y^3-3y^2-40y+84=(y-7)(y-2)(y+6).

The largest root is 77. Hence

αβ+γδ=7.\alpha\beta+\gamma\delta=7.

Using q=3q=3,

(α+β)(γ+δ)=αγ+αδ+βγ+βδ=q(αβ+γδ).(\alpha+\beta)(\gamma+\delta) =\alpha\gamma+\alpha\delta+\beta\gamma+\beta\delta =q-(\alpha\beta+\gamma\delta).

So

(α+β)(γ+δ)=37=4.(\alpha+\beta)(\gamma+\delta)=3-7=-4.

Using s=10s=10,

(αβ)(γδ)=10.(\alpha\beta)(\gamma\delta)=10.

Also

αβ+γδ=7.\alpha\beta+\gamma\delta=7.

Therefore αβ\alpha\beta and γδ\gamma\delta are the roots of

z27z+10=0.z^2-7z+10=0.

Since αβ>γδ\alpha\beta>\gamma\delta,

αβ=5,γδ=2.\alpha\beta=5,\qquad \gamma\delta=2.

Now let

U=α+β,V=γ+δ.U=\alpha+\beta,\qquad V=\gamma+\delta.

Since p=0p=0,

U+V=0.U+V=0.

Since r=6r=-6,

αβγ+αβδ+αγδ+βγδ=6.\alpha\beta\gamma+\alpha\beta\delta+\alpha\gamma\delta+\beta\gamma\delta=6.

Rewrite the left side as

αβ(γ+δ)+γδ(α+β)=5V+2U.\alpha\beta(\gamma+\delta)+\gamma\delta(\alpha+\beta)=5V+2U.

Thus

5V+2U=6.5V+2U=6.

Together with U+V=0U+V=0, this gives

U=2,V=2.U=-2,\qquad V=2.

Now split each pair:

α+β=2,αβ=5,\alpha+\beta=-2,\qquad \alpha\beta=5,

so α,β\alpha,\beta are roots of

X2+2X+5=0.X^2+2X+5=0.

Thus

X=1±2i.X=-1\pm2i.

Similarly,

γ+δ=2,γδ=2,\gamma+\delta=2,\qquad \gamma\delta=2,

so γ,δ\gamma,\delta are roots of

X22X+2=0.X^2-2X+2=0.

Thus

X=1±i.X=1\pm i.

Therefore the quartic roots are

1±2i,1±i.\boxed{-1\pm2i,\qquad 1\pm i.}

Writing check:

Solving the resolvent cubic does not give the four roots directly. It chooses a pairing quantity. The remaining work is to use square roots to split each pair into two individual roots.


The three cases now line up.

For degree 2:

symmetric sum/product(αβ)2two roots.\text{symmetric sum/product} \longrightarrow \sqrt{(\alpha-\beta)^2} \longrightarrow \text{two roots}.

For degree 3:

symmetric cubic dataΔ3orientation of the three rootscube-root choice for the remaining cyclethree roots.\text{symmetric cubic data} \longrightarrow \sqrt{\Delta_3} \longrightarrow \text{orientation of the three roots} \longrightarrow \text{cube-root choice for the remaining cycle} \longrightarrow \text{three roots}.

For degree 4:

symmetric quartic dataresolvent cubic for pairingschoose a pairingsquare roots split the pairs.\text{symmetric quartic data} \longrightarrow \text{resolvent cubic for pairings} \longrightarrow \text{choose a pairing} \longrightarrow \text{square roots split the pairs}.

The phrase “solving breaks symmetry” means this:

The coefficients treat the roots as an unordered set. A solution method adds controlled choices that gradually separate that unordered set into individual roots.

This also explains why formulas become more complicated as the degree rises. The issue is not just algebraic length. The issue is how much symmetry must be broken, and what intermediate objects make that possible.