# Isomorphisms of Affine Varieties (Tuesday, October 13) Last time: proved that if $X, Y$ are affine varieties then there is a bijection \[ \correspond{\text{Morphisms} \\ f:X\to Y} &\iff \correspond{\text{$k\dash$algebra morphisms}\\ A(Y) \to A(X)} \\ f & \mapsto f^*: \OO_Y(Y) \to \OO_X(X) .\] :::{.remark} A morphism $f:X\to Y$ is by definition a morphism of ringed spaces where $\OO_X, \OO_Y$ are the sheaves of regular functions. This shows $X\cong Y$ as ringed spaces iff $A(X) \cong A(Y)$ as $k\dash$algebras. ::: ## Counterexample: Isomorphisms Are Not Just Bijective Morphisms :::{.warnings title="Isomorphisms are not necessarily bijective morphisms"} Let $X = V(y^2 - x^3) \subset \AA^2$ and define a map \[ f: \AA^1 &\to X \\ t &\mapsto \tv{t^2, t^3} ,\] This is a morphism by proposition 4.7 in [@AndreasGathmann515], since the coordinates $t^2, t^3$ are regular functions. Then $f$ is a bijection, since we can define a piecewise inverse \[ f^{-1}: X &\to \AA^1 \\ \tv{x, y} &\mapsto \begin{cases} y/x & x\neq 0 \\ 0 & \text{else}. \end{cases} \] However, $f^{-1}$ is not a morphism, since it is not in $A(X)$ and thus not a regular function on $X$. For instance, pulling back the function $g(t) = t$ yields \[ \qty{ (f\inv)^* g} \qty{ \tv{x, y} } = \begin{cases} y/x & x\neq 0\\ 0 & x=y=0 \end{cases} \quad \not \in A(X) .\] Since $f$ is a morphism, however, we can still consider the corresponding map of $k\dash$algebras: \[ f^*: A(X) = \frac{k[x, y]}{\gens{y^2 - x^3}} &\to A(\AA^1) = k[t] \\ x & \mapsto t^2 \\ y & \mapsto t^3 ,\] but even though $f$ is a bijective morphism, it's not an isomorphism of rings: this can be seen from the fact that $t\not \in \im f^*$. ::: ## Categorical Products > Review of introductory category theory. We'll define a category $\mathrm{AffVar}_k$ whose objects are affine varieties over $k$ and morphisms in $\hom(X, Y)$ will be morphisms of ringed spaces. There is a contravariant functor $A$ into reduced[^def_reduced] finitely generated $k\dash$algebras which sends $X$ to $A(X)$ and sends morphisms $f:X\to Y$ to their pullbacks $f^*:A(Y) \to A(X)$. > Review of the universal property of the product. :::{.remark} If we have $X,Y$ affine varieties, we take $X\cross Y$ to be the categorical product instead of the underlying product of topological spaces. We have \[ A(X\cross Y) \cong A(X) \tensor_k A(Y) \cong \frac{ k[x_1, \cdots, x_n, y_1, \cdots, y_m]} { I(X) \tensor 1 + 1 \tensor I(Y) } .\] This recovers the product, since we have \begin{tikzcd} Z \ar[dr, dotted, "{\exists! H = (f, g)}"]\ar[rrd, bend left, "f"]\ar[rdd, bend right, "g"] & & \\ & X\cross Y\ar[r]\ar[d] & X \\ & Y & \\ \end{tikzcd} ::: :::{.remark} Products of spaces are sent to the tensor product of $k\dash$algebras, i.e. pullbacks are sent to pushouts. ::: :::{.remark} Note that the groupoid associated to a group does not have products: there can only be one element, but the outer triangles will not necessarily simultaneously commute. :::