Let $X$ and $Y$ be two complex reduced affine algebraic or analytic varieties, possibly singular. Take a regular proper function $$f\colon X \to Y $$ and assume that it is bijective at the level of $\mathbb{C}$-points. Moreover, assume that for every point $x$ of $X$ the differential $$df(x)\colon T_xX \to T_{f(y)}Y $$ is an isomorphism, where the tangent spaces are Zariski tangent spaces.

Can we conclude that $X$ and Y are isomorphic? Do we need any assumption on the singularities? In particular, should $Y$ be normal?

In other words, I am asking under which conditions on the singularities the implicit function theorem holds. I remember some notes by Kollar, where this issue was related with Canonical singularities, but I could not find them anymore.

Any reference or example is welcome.


  • 1
    $\begingroup$ Simple counterexample --- $X = Spec (k[t]/t^2)$, $Y = Spec (k[t]/t^3)$, the map is natural. $\endgroup$ – Sasha Feb 25 '15 at 20:07
  • 1
    $\begingroup$ So I need at least $X$ and $Y$ reduced $\endgroup$ – Giulio Feb 25 '15 at 20:11
  • $\begingroup$ thanks Sasha, I edited my question and added reduced in the hypothesis. $\endgroup$ – Giulio Feb 25 '15 at 21:03
  • $\begingroup$ W. S. Loud, Some Singular Cases of the Implicit Function Theorem The American Mathematical Monthly Vol. 68, No. 10 (Dec., 1961), pp. 965-977 $\endgroup$ – user21574 Jul 24 '17 at 22:39
  • 1
    $\begingroup$ There is a chaper about singular version of implicit function theorem link.springer.com/book/10.1007%2F978-1-4614-5981-1 $\endgroup$ – user21574 Jul 24 '17 at 22:43

In Joe Harris' book, Algebraic Geometry, this is theorem 14.9.


At least you should assume that $f$ is a homeomorphism (equivalently: proper, or finite).

Otherwise here is a counterexample: let $X=X_1\coprod X_2$ where $X_1$ is the union of the coordinate axes in $\mathbb{C}^2$, and $X_2=\mathbb{C}^*$. Let $Y\subset \mathbb{C}^2$ be the union of $X_1$ and the diagonal, and $f$ be the obvious map (identity on $X_1$, ant $t\mapsto (t,t)$ on $X_2$). It is clearly bijective, and an isomorphism if you remove the origin from $X_1$ and $Y$. At the origin, the Zariski tangent spaces of $X_1$ and $Y$ are the same, namely $\mathbb{C}^2$.

  • $\begingroup$ $f$ is bijective, so finite $\endgroup$ – Giulio Feb 25 '15 at 21:03
  • 4
    $\begingroup$ Dear Giulio, in Algebraic Geometry the property of finiteness has a specific meaning which implies properness. The example of Laurent is bijective but not proper since the image of the closed subset $X_2$ is not closed in $Y$. $\endgroup$ – Matthieu Romagny Feb 25 '15 at 21:20
  • $\begingroup$ Ok, I will edit the answer, and assume that $f$ is proper. Thanks $\endgroup$ – Giulio Feb 26 '15 at 22:01

Your Answer

By clicking "Post Your Answer", you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.