4
$\begingroup$

(this can be considered as some ad)

Consider the system of equations $F(x,y)=0$. (Here $x$, $y$ are multi-variables. The equations are over a local ring. e.g. polynomial/analytic/formal/$C^\infty$ etc.) The classical Implicit Function Theorem starts form the assumption: the matrix of partial derivatives, $F'_y(0,0)$, is non-degenerate.

Often this does not hold. For example, the classical version fails to ensure the solution of equations as simple as $xy=0$ or $y^2=0$ (say over $k[[x]]$).

A strengthening is known as Tougeron's Implicit Function Theorem. Tougeron replaces the condition "$F'_y(0,0)$ is non-degenerate" by the weaker "the entries of $F(x,0)$ belong to $I_{max}(F'_y(x,0))^2$". Here $I_{max}$ is the ideal of the maximal minors, i.e. the zero's fitting ideal.

Tougeron's condition is still too restrictive, e.g. it cannot ensure the solvability of $H(x,y)+y_1 x^m_1+y_2x^m_2+p(x)=0$. (Here $H(x,y)$ is at least quadratic in $y$ while $p(x)$ is of Taylor order at least $2m+1$.)

There have been several strengthenings (i.e. further weakenings of this condition). All of them being sufficient but not necessary.

In arxiv:1311.0088 we have obtained a necessary and sufficient condition for an equation to possess a "good" solution. (Well, actually an order-by-order solution. Once one has it, one gets a solution in the completion. Then one can use some Artin-type approximation results.)

Besides being iff result, this is also quite general, e.g. it holds for finitely generate modules over a commutative associative ring (which is not necessarily over a field). For example, if in the equation $H(x,y)+y_1 x^m_1+y_2x^m_2+p(x)=0$ both $H(x,y)$ and $p(x)$ are defined over $\mathbb{Z}$ (with $p(x)$ of order at least $2m+1$) then we get a solution over $\mathbb{Z}[[x]]$.

  1. Is anything of this style known? (References?)
  2. What are the potential applications? (Some particular cases where one needs a solution but $F'_y(0,0)$ is degenerate?)

(For the bookkeeping, here are some related questions: 1, 2, 3, 4)

$\endgroup$
2
  • $\begingroup$ What kind of functions are you considering? You label this commutative algebra, so are you solving polynomial equations? That wouldn't strictly be a generalization of the implicit function theorem, if so. There's a lot of basic singularity theory that characterises the singular strata of a smooth map. With some smart choices of strata you can guarantee smooth level-sets. But this might be straying from your purpose. $\endgroup$ Jun 6, 2014 at 23:10
  • $\begingroup$ The initial motivation came from the rings like $k[[x]]$, $k\{x\}$ or $k[x]_{(x)}$. Then we saw that we do not need the ring to be Noetherian. Then we saw that the statement holds over an arbitrary comm.assoc. ring. (Of course, in this case we clarify what do we mean by the "equation".) <br> The geometric view (or from singularities): given a germ of some space, e.g. $Spec(R_X\times R_Y)$ we check whether it contains a smooth component that surjects onto $Spec(R_X)$. $\endgroup$ Jun 7, 2014 at 6:10

0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.