# The twisted cubic is an affine variety.

Let $Y=\{(t,t^2,t^3)\mid t\in k\}$ be the twisted cubic curve. I'm trying to prove this curve is a variety, i.e., it's irreducible and affine algebraic set.

The easier part is to prove the twisted cubic curve is an affine algebraic set $(Y=Z(x^2-y,x^3-z))$.

I don't know how to prove that $Y$ is irreducible, I'm trying to prove that $(x^2-y,x^3-z)$ is prime, I think if I do this I proved what I want, but I found this hard to prove.

I need help to finish this question.

Thanks a lot.

• Please use the word "prove" appropriately. In what you call "the easy part", you didn't prove anything; you applied a trivial definition. Nov 15, 2013 at 20:03
• Show that the quotient is an integral domain by exhibiting an explicit isomorphism with the polynomial ring in one variable. Nov 15, 2013 at 20:35
• @ZhenLin Do you mean $k[x,y,z]/(x^2-y,x^3-z)\cong k[x,x^2,x^3]\cong k[t]$? Nov 15, 2013 at 21:23
• How do we know $Y = Z(x^2-y, x^3-z)$? Aug 24, 2020 at 17:52
• @user46372819 In case you are still wondering three years later: a point in $Y$ obviously kills both $x^2-y$ and $x^3-z$; on the other hand, a point $(a,b,c)$ that kills both equations is constrained to $b=a^2$ and $c=a^3$, so it has the form $(t,t^2,t^3)$. May 17 at 18:02

Let's first prove that ideal $I:=(x^2-y, x^3-z)$ is prime. Suppose $f\cdot g \in I$. Using obvious isomorphisms $k[x,y,z] \cong (k[x,y])[z]$, $k[x,y] \cong (k[x])[y]$ and division algorithm, we have $$f(x,y,z)=(x^3-z)f_1(x,y,z) + (x^2-y)f_2(x,y) + f_3(x),$$ $$g(x,y,z)=(x^3-z)g_1(x,y,z) + (x^2-y)g_2(x,y) + g_3(x).$$ Now we have $f_3(x) \cdot g_3(x) \in I$, therefore $$f_3(x) \cdot g_3(x)=(x^3-z)h_1(x,y,z) + (x^2-y)h_2(x,y,z).$$ Insert $(x,y,z)=(t,t^2,t^3)$ and get $f_3(t) \cdot g_3(t) =0$ for all $t \in k$. If $k$ is algebraically closed (therefore infinite), we have $f_3 \cdot g_3 = 0$, so $f_3 = 0$ or $g_3= 0$. Then $f \in I$ of $g \in I$, so $I$ is prime (and therefore radical). We have $I(Y)=I(V(I)) = \operatorname{Rad}(I) = I$, which is prime. So $Y$ is irreducible.

• There is no need to assume that $k$ is algebraically closed (and the OP didn't do it): in the last polynomial relation plug in $y=x^2$ and $z=x^3$ to find $f_3=0$ or $g_3=0$.
– user89712
Nov 23, 2013 at 9:44
• Element calculations make everything complicated ... $I$ is prime because obviously $k[x,y,z]/(I) = k[x]$. Nov 23, 2013 at 16:40
• Thank you for you answer, I have a question, why can you use division algorithm? Dec 3, 2013 at 11:04
• Not necessarily. It is enough that $k[x,y]$ is a commutative ring, and that leading coefficient of $x^3-z$ is an unit in that ring. The leading coefficient is $-1$ (in variable $z$ of course). Proof is the same as in the case of fields. Dec 4, 2013 at 9:01
• No, no calculation is needed. It is trivial to check that both sides represent the same functor, hence they are isomorphic (Yoneda). Dec 5, 2013 at 0:53

There is an obvious isomorphism $Y \cong \mathbb{A}^1$. This proves everything else.

• How do you know which polynomials to choose to define the variety? Mar 31, 2021 at 22:06

Expand a little on Martin's answer.

Define $$f:X \to Y$$ as $$f(x) = (x, x^2, x^3)$$ Define $$g:Y \to X$$ as $$g(x,y,z) = x$$

Then $$f^{-1} = g$$, $$f$$ is an isomorphism between $$A^1$$ and Y.

Now by p.g.29 of Shafarecich's Basic algebraic Geometry 1, there exists an isomorphism between coordinate rings of $$A^1$$ and $$Y$$, which means $$k[x] \cong k[x,y,z]/((x^2-y),x^3-z)$$

Since k[x] is an integral domain, $$k[x,y,z]/((x^2-y),x^3-z))$$ is an integral domain, which means that $$((x^2-y),x^3-z))$$ is irreducible.

Here is a way to show that $$Y$$ is irreducible with little computation: note that since $$Y$$ is closed (as the set of zeros of two polynomials) we have $$Y=Z(I(Y))$$, where $$I(Y)$$ is the ideal of $$Y$$. The only thing to check is that $$I(Y)$$, is prime. But it is easy to see that $$I(Y)$$ is the kernel of a surjective ring homomorphism $$\phi:k[x,y,z]\to k[t]$$, namely the homomorphism such that $$\phi(x)=t,\phi(y)=t^2,\phi(z)=t^3$$. Let's check this in detail, even though this is a bit tautological: To say that $$p\in k[x,y,z]$$ belongs to $$\ker\phi$$ means that the polynomial $$p(t,t^2,t^3)\in k[t]$$ is the zero polynomial. This implies, obviously, that for all $$t\in k$$ we have $$p(t,t^2,t^3)=0$$, i.e. that $$p\in I(Y)$$. Conversely, if $$p\in I(Y)$$ then the polynomial $$\phi(p)=p(t,t^2,t^3)\in k[t]$$ has any $$t\in k$$ as a root, which means that $$\phi(p)=0$$. Hence $$I(Y)=\ker\phi$$.

It follows that the ring $$A(Y)=k[x,y,z]/I(Y)$$ is isomorphic to $$k[t]$$, which is integral, hence $$I(Y)$$ is prime, and $$Y$$ is an affine variety isomorphic to the affine line. Q.E.D.

Of course the ideal $$I(Y)$$ is the radical of $$(x^2-y,x^3-z)$$, and the above arguments show that in fact $$I(Y)=(x^2-y,x^3-z)$$, but the computation of $$I(Y)$$ is not needed.