Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

I'm reading Fulton's book and I'm trying to solve the following exercise :

Let $T : \Bbb A^2 \rightarrow \Bbb A^2$ be a polynomial map, $T (Q) = P$.

(a) Show that $m_Q(F^T ) \geq m_P (F)$.

(b) Let $T = (T1,T2)$, and define $J_QT = (\partial T_i /\partial X_j (Q))$ to be the Jacobian matrix of T at Q.

Show that $m_Q(F^T ) = m_P (F)$ if $J_QT$ is invertible.

a) For the first question I can write $ \partial^n F^T/ \partial X_j^n (Q) $ in terms of $ \partial^n F / \partial X_j^n(P) $ (using for example Faà di Bruno's formula). For the first derivative for instance one can easily show that $ F_X^T(Q) = F_X(P)T_X(Q) $ and thus : If $ \partial^n F^T/ \partial X_j^n (Q) \neq 0 $ Then there exists an $ m \leq n $ such that $ \partial^m F / \partial X_j^m(P) \neq 0 $.

b) Do you have some suggestions for the second question ?

share|improve this question
add comment

1 Answer

up vote 4 down vote accepted

a) Let $T : \Bbb{A}^2 \rightarrow \Bbb{A}^2$ be the polynomial map where $T(Q)=P$. We want to show that $m_Q(F^T) \geq m_P(F)$. We show this first for $Q = (0,0)$ and $P = (0,0)$. Let $T = (T_1, T_2)$. We have that $T(0,0) = (0,0)$. So $T_1$ and $T_2$ have no constant term. Now we have that $F^T = F(T_1(X,Y),T_2(X,Y)$ and $T_1$ and $T_2$ have no constant term. So $T$ does not decrease the lowest degree term of $F$. Therefore $m_Q(F^T) = m_P(F)$.

Now assume either $Q \neq (0,0)$ or $P \neq (0,0)$. Let $T'$ be the translation that takes $(0,0)$ to $Q$ and $T''$ be the translation that take $(0,0)$ to $P$. Then we have that $T \circ T'(0,0) = T''(0,0)$. Now we have that $m_Q(F^T) = m_{(0,0)}(F^{T\circ T'})$ and $m_P(F) = m_{(0,0)}(F^{T''})$. Also we know that $$ (T'')^{-1}\circ T \circ T'(0,0) = (0,0)$. $$ Let $G = F^{T''}$. Then we have that $m_{(0,0)}(F^{T\circ T'}) = m_{(0,0)}(G^{(T'')^{-1}\circ T \circ T'})$. Now let $\bar{T} = (T'')^{-1} \circ T \circ T'$. So $\bar{T}(0,0) = (0,0)$. So by the ealier case, we have that $m_{(0,0)}(G^{\bar{T}}) \geq m_{(0,0)}(G)$. Therefore $m_Q(F^T) \geq m_P(F)$.

b) Let $T = (T_1,T_2)$ and $$ J_Q T = \begin{pmatrix} \dfrac{\partial T_1}{\partial X}(Q) & \dfrac{\partial T_1}{\partial Y}(Q) \\ \dfrac{\partial T_2}{\partial X}(Q) & \dfrac{\partial T_2}{\partial Y}(Q) \end{pmatrix}. $$ As before we show the result for the case $P = (0,0)$ and $Q = (0,0)$.

Assume $J_Q T$ is invertible. Since $J_Q T$ is invertible, we can't have both $\dfrac{\partial T_1}{\partial X}(Q) = 0$ and $\dfrac{\partial T_1}{\partial Y}(Q) = 0$ or both $\dfrac{\partial T_2}{\partial X}(Q) = 0$ and $\dfrac{\partial T_2}{\partial Y}(Q) = 0$. Since $Q = (0,0)$, this implies that the decomposition of $T_1$ and $T_2$ into homogeneous polynomials are $$ T_1 = G_1 + G_2 + G_3 + \ldots $$ and $$ T_2 = H_1 + H_2 + H_3 + \ldots $$ where the $G_i$'s and the $H_i$'s are forms. By part (a), $T_1$ and $T_2$ have no constant terms. Thus, $$ F^T = F(T_1,T_2) = F_m(T_1,T_2) + F_{m+1}(T_1,T_2) + \ldots $$ Since the lowest degree forms of $T_1$ and $T_2$ are of degree $1$, we have that $T$ does not decrease of the form $F_m(T_1,T_2)$. Similarly, $T$ does not decrease the degree of $F_{m+1}(T_1,T_2), F_{m+2}(T_1,T_2), \ldots$. Therefore we have that $m_{(0,0)}(F^T) = m_{(0,0)}(F)$.

Now assume that either $Q = (a_1,b_1) \neq (0,0)$ or $P = (a_2,b_2) \neq (0,0)$. Assume that $J_Q T$ is invertible. Let $T'$ be the translation that takes $(0,0)$ to $Q$ and $T''$ be the translation that takes $(0,0)$ to $P$. Then we have that $$ \begin{pmatrix} \dfrac{\partial T_1}{\partial X}(X+a_1) & \dfrac{\partial T_1}{\partial Y}(X+a_1) \\ \dfrac{\partial T_2}{\partial X}(Y+b_1) & \dfrac{\partial T_2}{\partial Y}(Y+b_1) \end{pmatrix} = \begin{pmatrix} 1 & 0 \\ 0 & 1 \end{pmatrix} $$ and $$ \begin{pmatrix} \dfrac{\partial T_1}{\partial X}(X+a_2) & \dfrac{\partial T_1}{\partial Y}(X+a_2) \\ \dfrac{\partial T_2}{\partial X}(Y+b_2) & \dfrac{\partial T_2}{\partial Y}(Y+b_2) \end{pmatrix} = \begin{pmatrix} 1 & 0 \\ 0 & 1 \end{pmatrix}. $$ So by the chain rule, we have that $J_Q T = J_{(0,0)}((T'')^{-1} \circ T \circ T')$. Letting $\bar{T} = (T'')^{-1} \circ T \circ T'$, we have that $J_0 \,\bar{T}$ is invertible and the result follows from our ealier case.

share|improve this answer
add comment

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.