Let $k$ be a field and $R = k[x,y]$ and $\mathscr{m} = (x, y)$ and $n\in \mathbb{N}$.
$(a)$ Show that $\mathscr{m}^n$ cannot be generated by less than $n + 1$ elements.
$(b)$ Find a generating set of $\mathscr{m}^n$ containing $n + 1$ elements.

My Attempt:
$(a)$: (Proof by Induction:)
For $n=1$, It holds because $\mathscr{m}$ cannot be generated by less than two elements.

Induction Assumption:
Assume that the statement holds true for $n-1$. $\hspace{50pt}(*)$

Induction Step:
To Prove that the statement holds for $n$.

On Contrary, assume that $\mathscr{m}^n$ is generated by less than $n + 1$ elements.$\hspace{20pt}(**)$

Combining $(*)$ and $(**)$, we get that $\mathscr{m}^n$ is generated by $n$ elements.

In particular, this implies that $\mathscr{m}^2$ is generated by two $1$ element.
It is contradiction to the fact that $\mathscr{m}^2$ cannot be generated by less than three elements.

Hence, the statement holds for $n$, and the proof follows.

For $(b)$, I have no clue, how to solve.

Please check my $(a)$, and provide me a solution if it wrong, else please provide me another solution. Also, Please provide me hints for $(b)$.

Also, I am wondering what is the use of $k$ to be field?

I am referring to Dummit and Foote$(3^{ed})$.

The italicized statements have been edited and added after the answer of ThorWittch

But I want a more elaborate answer for $(a)$ sticking to the reference, I am using.


  • 2
    $\begingroup$ Hint for (a): it's probably easiest to use the fact that $\mathfrak{m}^n / \mathfrak{m}^{n+1}$ is a vector field over $R / \mathfrak{m} \simeq k$. $\endgroup$ – Daniel Schepler Jun 27 at 17:26
  • $\begingroup$ @DanielSchepler I didn't get you. How does $\mathscr{m}^n/\mathscr{m}^{n+1}$ is a vector space? are we saying that $\mathscr{m}^{i+1} \subset \mathscr{m}^{i}$ $\forall 1\leq i < n$ ? $\endgroup$ – Kumar Jun 27 at 17:37
  • 1
    $\begingroup$ Correct, @Kumar. That is, indeed, what Daniel Schepler says. For example $\mathfrak{m}^2$ is generated by $x^2,xy,y^2$ (it consists of all the polynomials of degree two and higher). You should figure out the dimension of $\mathfrak{m}^n/\mathfrak{m}^{n+1}$ over $k$ to get started. $\endgroup$ – Jyrki Lahtonen Jun 27 at 17:43
  • 2
    $\begingroup$ Yes @Kumar. For $n = 1$ that is for example used to define the cotangent space in algebraic and differential geometry at some point of a scheme or smooth manifold. $\endgroup$ – ThorWittich Jun 27 at 17:46
  • 1
    $\begingroup$ Correct, again. Have you covered Nakayama's lemma? Or, may be you can show directly that a generating set of $\mathfrak{m}^n$ (as an $R$-module) spans $\mathfrak{m}^n/\mathfrak{m}^{n+1}$ over $k$? $\endgroup$ – Jyrki Lahtonen Jun 27 at 19:56

Your (a) is not clear at all to me.

Even for $n = 1$. The definition of $\mathcal{m}$ does not imply that the ideal is not principal (for example the ideal $(2,3) \subset \mathbb{Z}$ is principal, as it is generated by $1$), but you can get that with an easy argument.

Assume that $(f) = (x,y)$ for some polynomial $f \in k[x,y]$. Then we have $af = x$ and $bf = y$ for some $a,b \in k[x,y]$. This yields $\text{deg}(a) + \text{deg}(f) = 1$ and $\text{deg}(b) + \text{deg}(f) = 1$. Therefore either $a$ or $f$ has degree $0$. If $a$ has degree $0$, then $a \in k$ is just some non-zero scalar and we get $f = a^{-1}x$ and $\text{deg}(f) = 1$. Thus we have $\text{deg}(b) = 0$, such that $ba^{-1}x =y$ for some non-zero $a,b \in k$, which is a contradiction. If $\text{deg}(f) = 0$, then $(f) = k[x,y]$ as $f$ is a unit, which contradicts $(f) = (x,y)$. Therefore you need at least two generators for $\mathcal{m}$.

I also don't understand your argument for the rest.


For $n = 1$ we have the generators $x,y$.

For $n = 2$ we have the generators $x^2,xy,y^2$.

For $n = 3$ we have the generators $x^3,x^2y,xy^2,y^3$.


I guess we are working over a field in order for $\mathcal{m}$ to be a maximal ideal.

  • $\begingroup$ If possible, can you please provide me, full solution for $(a)$? $\endgroup$ – Kumar Jun 27 at 17:29

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.