[proofplan]
We separate the constant polynomial case from the positive-degree case. When $d=0$, the quotient is identically $1$. When $d \ge 1$, we divide the polynomial by its leading term and obtain $1$ plus finitely many lower-order terms of the form $c_i x^{i-d}$, where $i-d<0$. We then prove directly from the definition of the limit as $|x| \to \infty$ that the finite lower-order sum tends to $0$.
[/proofplan]
custom_env
admin
[step:Handle the constant polynomial case]
Suppose first that $d=0$. Then $p(x)=a_0$ for every $x \in \mathbb{R}$, and $a_d=a_0 \ne 0$. Since $x^0=1$ for every $x \in \mathbb{R}$, we have
\begin{align*}
\frac{p(x)}{a_d x^d}=\frac{a_0}{a_0}=1
\end{align*}
for every $x \in \mathbb{R}$. Hence
\begin{align*}
\lim_{|x| \to \infty} \frac{p(x)}{a_d x^d}=1.
\end{align*}
[/step]
custom_env
admin
[step:Rewrite the normalized polynomial as one plus lower-order terms]Assume now that $d \ge 1$. For every $x \in \mathbb{R}$ with $|x| \ge 1$, we have $x \ne 0$, so $a_d x^d \ne 0$ because $a_d \ne 0$. Therefore the quotient is defined for all $x$ with $|x| \ge 1$.
For each integer $i \in \{0,\dots,d-1\}$, define the real constant
\begin{align*}
c_i=\frac{a_i}{a_d}.
\end{align*}
Then, for every $x \in \mathbb{R}$ with $|x| \ge 1$,
\begin{align*}
\frac{p(x)}{a_d x^d}=1+\sum_{i=0}^{d-1} c_i x^{i-d}.
\end{align*}[/step]
custom_env
admin
[guided]The purpose of normalizing by $a_d x^d$ is to isolate the leading term. Since $d \ge 1$, the denominator $a_d x^d$ is nonzero whenever $x \ne 0$; in particular it is nonzero for all $x$ satisfying $|x| \ge 1$.
For each integer $i \in \{0,\dots,d-1\}$, define
\begin{align*}
c_i=\frac{a_i}{a_d}.
\end{align*}
This is well-defined because the leading coefficient satisfies $a_d \ne 0$. Dividing the expression
\begin{align*}
p(x)=a_d x^d+\sum_{i=0}^{d-1} a_i x^i
\end{align*}
by $a_d x^d$ gives
\begin{align*}
\frac{p(x)}{a_d x^d}=1+\sum_{i=0}^{d-1} \frac{a_i}{a_d} x^{i-d}.
\end{align*}
Using the constants $c_i$ just defined, this becomes
\begin{align*}
\frac{p(x)}{a_d x^d}=1+\sum_{i=0}^{d-1} c_i x^{i-d}.
\end{align*}
Every exponent $i-d$ in the remaining sum is negative, because $i \le d-1$. Thus each term is a lower-order correction to the leading term.[/guided]
custom_env
admin
[step:Show the lower-order finite sum tends to zero]
Define the function $r: \mathbb{R} \setminus \{0\} \to \mathbb{R}$ by
\begin{align*}
r(x)=\sum_{i=0}^{d-1} c_i x^{i-d}.
\end{align*}
We prove that $\lim_{|x| \to \infty} r(x)=0$.
Let $\varepsilon>0$. For each $i \in \{0,\dots,d-1\}$, define
\begin{align*}
b_i=|c_i|.
\end{align*}
Define the nonnegative constant
\begin{align*}
B=\sum_{i=0}^{d-1} b_i.
\end{align*}
If $B=0$, then $r(x)=0$ for every $x \ne 0$, so the desired limit is immediate.
Assume $B>0$. Define
\begin{align*}
R=\max\{1,B/\varepsilon\}.
\end{align*}
If $|x|>R$, then $|x|>1$. Since $i-d \le -1$ for every $i \in \{0,\dots,d-1\}$, we have
\begin{align*}
|x^{i-d}|=|x|^{i-d}\le |x|^{-1}.
\end{align*}
By the triangle inequality,
\begin{align*}
|r(x)|\le \sum_{i=0}^{d-1} |c_i| |x^{i-d}|.
\end{align*}
Using the preceding bound term by term gives
\begin{align*}
|r(x)|\le \sum_{i=0}^{d-1} b_i |x|^{-1}=B|x|^{-1}<\varepsilon.
\end{align*}
Thus, for every $\varepsilon>0$, there exists $R>0$ such that $|x|>R$ implies $|r(x)|<\varepsilon$. Therefore
\begin{align*}
\lim_{|x| \to \infty} r(x)=0.
\end{align*}
[/step]
custom_env
admin
[step:Conclude that the normalized polynomial tends to one]
For every $x \in \mathbb{R}$ with $|x| \ge 1$,
\begin{align*}
\frac{p(x)}{a_d x^d}=1+r(x).
\end{align*}
Since $\lim_{|x| \to \infty} r(x)=0$, it follows directly from the definition of the limit that
\begin{align*}
\lim_{|x| \to \infty} \frac{p(x)}{a_d x^d}=1.
\end{align*}
Combining this with the already handled case $d=0$ proves the theorem.
[/step]