# Start a new discussion

## Not signed in

Want to take part in these discussions? Sign in if you have an account, or apply for one below

## Site Tag Cloud

Vanilla 1.1.10 is a product of Lussumo. More Information: Documentation, Community Support.

• CommentRowNumber1.
• CommentAuthorTodd_Trimble
• CommentTimeSep 6th 2015
• (edited Sep 6th 2015)

I started an article bijective proof.

In a section on polynomial identities, I give a proof that polynomials in several variables are uniquely determined by their values at natural number arguments, an intuitively obvious statement if there ever was one. Without bothering to look up whether there are standard nice proofs, I cooked up a proof myself. Please let me know if you know of nicer proofs.

Edit: Having written this, it’s painfully obvious how to prove more general statements even more simply. Ah well. I still invite comments.

• CommentRowNumber2.
• CommentAuthorZhen Lin
• CommentTimeSep 6th 2015

It suffices to prove the following:

Let $A$ be an infinite integral domain and let $B$ be an infinite subset of $A$. Given $p (x_1, \ldots, x_n) \in A [x_1, \ldots, x_n]$, if $p (b_1, \ldots, b_n) = 0$ for all $(b_1, \ldots, b_n) \in B^n$, then $p = 0$.

We proceed by induction on $n$. The case $n = 1$ is well known (at least if $A$ is an algebraically closed field – but we can embed an integral domain in the algebraic closure of its fraction field). If the claim holds for $n = k$, then given $p (x_1, \ldots, x_{k+1}) \in A [x_1, \ldots, x_{k+1}]$ such that $p (b_1, \ldots, b_{k+1}) = 0$ for all $(b_1, \ldots, b_{k+1}) \in B^{k+1}$, we know that $p (x_1, \ldots, x_k, b)$ is the zero polynomial for all $b \in B$; but then $p (x_1, \ldots, x_k, x_{k+1})$ is a one-variable polynomial over $A [x_1, \ldots, x_k]$ that vanishes at infinitely many points, so $p = 0$.

• CommentRowNumber3.
• CommentAuthorTodd_Trimble
• CommentTimeSep 6th 2015

Yep, thanks. I would have added the (obvious) remark that we first embed $A[x_1, \ldots, x_k]$ into its field of fractions $K$, and then use well-known facts about $K[x]$.

• CommentRowNumber4.
• CommentAuthorTodd_Trimble
• CommentTimeSep 6th 2015

So I went ahead and supplanted the earlier clunky proof with a version of Zhen Lin’s (thanks again).

• CommentRowNumber5.
• CommentAuthorDavidRoberts
• CommentTimeSep 6th 2015

Doron Zeilberger is fond of proving identities by checking them at the smallest few nontrivial inputs, eg 1,2 and 3, where these are trivial to calculate.

• CommentRowNumber6.
• CommentAuthorTodd_Trimble
• CommentTimeApr 20th 2016

I added some details to some examples that had been there. Will probably write (and rewrite) more later.