Examine linear dependence of f_1(x)=x, f_2(x)=x^3…
Examine linear dependence of $f_1(x)=x,\qquad f_2(x)=x^3,\qquad f_3(x)=|x^3|$.
Explanation of Method
Show that $x,\ x^3,\ |x^3|$ are dependent on each half-line but independent on all of $\mathbb{R}$, even though $W\equiv0$ everywhere $-$ a classic counterexample to the converse Wronskian test.
Step 1: Preliminary observation
Since $|x^3|=|x|^3$, \[f_3(x)=|x^3|=\begin{cases}x^3,&x\ge0,\\-x^3,&x<0.\end{cases}\]
$f_3$ is continuous everywhere and, in fact, twice differentiable everywhere ($f_3''(x)=6|x|$ is continuous, even at $x=0$, where $f_3''(0)=0$); but its third derivative fails to exist at $x=0$, so $f_3$ is not real‑analytic there. This is exactly the feature that makes the case interesting. Dependence is examined separately on $(0,+\infty)$, on $(-\infty,0)$, and on all of $\mathbb{R}$.
Step 2: On each half-line
$x>0$.
Here $|x^3|=x^3=f_2$, so
\[ 0\cdot x+1\cdot x^3+(-1)\cdot|x^3|=x^3-x^3=0, \]
a non-trivial relation ($c_2=1,\ c_3=-1$). Dependent on $(0,+\infty)$.
$x<0$.
Here $|x^3|=-x^3=-f_2$, so
\[ 0\cdot x+1\cdot x^3+1\cdot|x^3|=x^3+(-x^3)=0, \]
a non-trivial relation ($c_2=1,\ c_3=1$). Dependent on $(-\infty,0)$.
Note the two relations are different: $c_3=-c_2$ is needed on the right, $c_3=+c_2$ on the left. A single triple $(c_1,c_2,c_3)$ cannot satisfy both unless $c_2=c_3=0$ – which is exactly what forces independence globally.
Step 3: On all of $\mathbb{R}$
Suppose $c_1x+c_2x^3+c_3|x^3|\equiv0$ for every $x\in\mathbb{R}$. Splitting into the two half-lines (where $x,x^3$ are themselves independent, by Problem 61, $3^\circ$): \[x>0:\quad c_1x+(c_2+c_3)x^3\equiv0\ \Longrightarrow\ c_1=0,\ \ c_2+c_3=0.\]
\[ x<0:\quad c_1x+(c_2-c_3)x^3\equiv0\ \Longrightarrow\ c_1=0,\ \ c_2-c_3=0. \]
Adding and subtracting $c_2+c_3=0$ and $c_2-c_3=0$ gives $c_2=0$ and $c_3=0$. So $c_1=c_2=c_3=0$ is the only solution.
Equivalent parity argument: $x$ and $x^3$ are odd functions while $|x^3|$ is even. If $c_1x+c_2x^3+c_3|x^3|\equiv0$, the odd part $c_1x+c_2x^3$ and the even part $c_3|x^3|$ must each vanish identically on their own (the zero function is the unique function that is both), giving $c_1=c_2=0$ (odd polynomials independent) and $c_3=0$ (since $|x^3|\not\equiv0$).
The three functions are linearly independent on $\mathbb{R}$.
Step 4: The Wronskian on $\mathbb{R}$
For $x>0$, $f_3=f_2=x^3$, so columns 2 and 3 of $W$ are identical: \[W(x)=\begin{vmatrix}x&x^3&x^3\\1&3x^2&3x^2\\0&6x&6x\end{vmatrix}=0\qquad(x>0).\]
For $x<0$, $f_3=-f_2$, so column 3 is $(-1)$ times column 2, and again $\det=0$. At $x=0$, using the actual values $f_1(0)=0,f_1'(0)=1,f_1''(0)=0$ and $f_2(0)=f_2'(0)=f_2''(0)=0=f_3(0)=f_3'(0)=f_3''(0)$, the first row of $W(0)$ is all zeros, so $W(0)=0$ too.
Thus $W(x)\equiv0$ on all of $\mathbb{R}$.
Step 5: Summary
$(0,+\infty)$.
$W\equiv0$ – dependent ($|x^3|=x^3$).
$(-\infty,0)$.
$W\equiv0$ – dependent ($|x^3|=-x^3$).
$\mathbb{R}$.
$W\equiv0$ – nonetheless independent (direct/parity proof).
Result
Final answer:
\[\boxed{\,\text{Linearly independent on } \mathbb{R}\,}\]
Final answer:
\[\boxed{\,W(x)\equiv0 \text{ on all of } \mathbb{R}\,}\]
Remark
The subtlety.
Here $W\equiv0$ on $\mathbb{R}$, yet the system is linearly independent on $\mathbb{R}$. This does not contradict the earlier principle ($W\not\equiv0\Rightarrow$ independent) – that direction is always valid. What fails is the converse ($W\equiv0\Rightarrow$ dependent), which is only guaranteed when $f_1,\ldots,f_n$ are solutions of a common $n$-th order linear homogeneous ODE with continuous coefficients on the interval. No such ODE can have $|x^3|$ as a solution near $x=0$, since a third-order linear ODE would force $f_3$ to be three-times differentiable there, which it is not. $|x^3|$ is a "fake" analytic function: it agrees with $x^3$ on one side and $-x^3$ on the other, which is enough to make the Wronskian vanish locally without making the system globally dependent.