Let $a$ and $b$ be real numbers with $a<b$. Let $n>1$, and suppose that the mapping $g:[a, b] \longrightarrow \mathbb{R}^n$ is continuous and that $g$ is differentiable on the open interval $(a, b)$. It is tempting to generalize the mean value theorem (Theorem 1.2.3) to the assertion
$$
g(b)-g(a)=g^{\prime}(c)(b-a) \quad \text { for some } c \in(a, b) .
$$
The assertion is grammatically meaningful, since it posits an equality between two $n$-vectors. The assertion would lead to a slight streamlining of the proof of Lemma 5.1.3, since there would be no need to reduce to scalar output. However, the assertion is false.
(a) Let $g:[0,2 \pi] \longrightarrow \mathbb{R}^2$ be $g(t)=(\cos t, \sin t)$. Show that (5.1) fails for this $g$. Describe the situation geometrically.
(b) Let $g:[0,2 \pi] \longrightarrow \mathbb{R}^3$ be $g(t)=(\cos t, \sin t, t)$. Show that (5.1) fails for this $g$. Describe the situation geometrically.
(c) Here is an attempt to prove (5.1): Let $g=\left(g_1, \ldots, g_n\right)$. Since each $g_i$ is scalar-valued, we have for $i=1, \ldots, n$ by the mean value theorem,
$$
g_i(b)-g_i(a)=g_i^{\prime}(c)(b-a) \text { for some } c \in(a, b) .
$$
Assembling the scalar results gives the desired vector result.
What is the error here?