Take the 2-minute tour ×
MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

Edit: question has been changed from 'lexicographic' (cf. "D K"'s answer below) to 'degree' minimality.

Let $x_1,x_2,x_3$ be indeterminates. Fix an integer $k> 3$. Consider the set $M$ of all monomials of the form $x_1^{i_1}.x_2^{i_2}.x_3^{i_3}$ where each $i_j\in \mathbb{N}$ with $i_j\geq 1$ and $i_1+i_2+i_3 = k$.

Fix an arbitrary $F\in M$ and write $F = x_1^{a}.x_2^{b}.x_3^{c}$.

Question When does there exist a triple $(w_1, w_2, w_3)$, with each $w_i\in \mathbb{N}$ and $w_i\geq 1$, such that $$a.w_1 + b.w_2 + c.w_3 < i_1.w_1 + i_2.w_2 + i_3.w_3, \forall G = x_1^{i_1}.x_2^{i_2}.x_3^{i_3}\in M \;\text{with} \; G\neq F ?$$ In the exponential case, same question replaced by $a^{w_1}+ b^{w_2}+ c^{w_3} < i_1^{w_1}+i_2^{w_2}+ i_3^{w_3}$.

Positive example in additive case: $k = 4$ on three indeterminates. Then if $F$ corresponds to the element of $B_4$ with exponents $(2,1,1)$, then for any $w_2, w_3>1$, $w = (1,w_2,w_3)$ has the desired property.

Sometimes this question has a negative answer: e.g. k=4 on only two indeterminates. if we take F to correspond to the element in $B_4$ with exponents $(2,2)$ then we cannot achieve both $2w_1 + 2w_2 < w_1 + 3 w_2$ and $2w_1+2w_2 < 3w_1+w_2$.


Same question for $n$-variables $x_1,\ldots, x_n$ and $k > n.$

Apologies if this question has been answered; (general and specific) references also appreciated.

share|improve this question

2 Answers 2

up vote 1 down vote accepted

I haven't completely checked, but examples suggest that $F$ must be, up to reordering the $x_i$'s, the minimal lexicographic element in $B_k$ (in general, you can also replace $B_k$ with a subset of $B_k$, and then take the minimal element of this new set)

Example: take $k=5$. Then $(a,b,c)$ must have one of $a,b,$ or $c$ equal to $3$, i.e. no exponent of 2 is allowed in $F$ for a positive answer.

Explicitly, for the affirmative: we can assume without loss of generality that $F = xyz^3$. Then we get that $w_3 < w_1$, $w_3 < w_2$ and $2w_3 < w_1+w_2$, which is infinitely possible. Take, e.g. $w = (2,2,1)$. On the other hand if $F = xy^2z^2$ there is no solution since we must satisfy $w_3 < w_2$ and $w_2 < w_3$.

share|improve this answer
2  
ok, I think that's correct in the general case: for $n\geq 2, k>n$, then if $F$ is, up to reordering, the minimal element of $B_k$ in the lexicographic ordering, then there are infinitely many $w's$ for which you have a positive answer to your question. It's probably true that this condition is also sufficient, i.e. if $F$ is not minimal, then there are no solutions. –  user34821 Jun 11 '13 at 9:28

It is clearly not the case, as the optimal choice for $a,b,c$ once the weights are fixed is always to put $k$ on the minimal weight and $0$ on the others.

You can see that if you start from a different configuration, the total weight will decrease when you move to this one, as you transfer "mass" from "heavy" to "light".

So if $a,b,c$ are not $k,0,0$ (in any order), then it cannot be a minimal choice, no matter what the weights are.

This is still true in the exponentiation setting.

share|improve this answer
    
thank you, though i have now changed the question. and yes, you are completely correct that in the original (and wrong) version of the question, we won't get a desired lexicographic ordering unless the chosen $f$ was already the minimal wrt lex. ordering. –  Freddy Jun 9 '13 at 15:42
    
as a very easy example of positivity for the new question: take $k=4$ for three indeterminates. Then the exponents of the three elements of $B_k$ have coordinates $(2,1,1), (1,2,1),$ and $(1,1,2)$. Say we take $F\in B_k$ to be the element with $(2,1,1)$. Then in fact there are infinitely-many good $w's$, namely $w = (1, w_2, w_3)$ where $w_2,w_3>1$ –  Freddy Jun 9 '13 at 16:05

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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