Visual walkthrough — Cauchy-Euler (Equidimensional) equation

---

Step 1 — What do the symbols even mean?

WHAT. Before we solve anything, let us read the equation like a sentence. The three characters are:

• $y$ — an unknown function of $x$. Think of it as a curve whose height at position $x$ we don't yet know.
• $y^{\prime }$ — the slope of that curve (how fast the height changes as we step right). The prime mark ${}^{\prime }$ means "rate of change with respect to $x$".
• $y^{\prime \prime }$ — the slope of the slope (how much the curve bends).

The full equation is $\underset{\text{number}}{\underbrace{a}}\underset{\text{coeff.}}{\underbrace{x^2}}\underset{\text{bend}}{\underbrace{y^{\prime \prime }}}+\underset{\text{number}}{\underbrace{b}}\underset{\text{coeff.}}{\underbrace{x}}\underset{\text{slope}}{\underbrace{y^{\prime }}}+\underset{\text{number}}{\underbrace{c}}\underset{\text{height}}{\underbrace{y}}=0.$

WHY. Notice the pattern in the coefficients: the bend term wears $x^2$, the slope term wears $x^1$, the height term wears $x^0=1$. The power of $x$ always equals the number of primes. That is the "equidimensional" balance — it is the single fact this whole page exploits.

PICTURE. Below, a curve $y(x)$ with its height, slope arrow (tangent), and bend marked at one point.

---

Step 2 — Why a power $x^m$ is the natural guess

WHAT. We propose the trial shape $y=x^m$, where $m$ is a number we do not yet know (it could be $2$, or $-1$, or even imaginary — we will find out). Here $x^m$ just means "$x$ raised to the power $m$".

WHY this shape and not a wave? In an ordinary constant-coefficient ODE (see Constant-Coefficient Linear ODEs) the magic shape is $e^{rx}$, because differentiating $e^{rx}$ just multiplies it by $r$ — the shape survives. Here the coefficients are powers of $x$, not constants, so the surviving shape must interact nicely with multiplying by $x$. A power does exactly that: multiplying $x^m$ by $x$ gives $x^{m+1}$ (bump the power up), differentiating $x^m$ gives $m{x}^{m-1}$ (bump the power down). Those two moves are inverses. So $x^2{y}^{\prime \prime }$ pushes the power down twice then up twice — it returns to $x^m$.

PICTURE. The two operations shown as arrows on a "power ladder": differentiation steps down one rung and scales, multiply-by-$x$ steps back up.

---

Step 3 — Differentiate the guess, watch the powers slide

WHAT. Apply the differentiation rule $\frac{d}{dx}{x}^p=p{x}^{p-1}$ twice:

$y=x^m,y^{\prime }=\underset{\text{scale}}{\underbrace{m}}{x}^{\stackrel{\text{power dropped 1}}{\overbrace{m-1}}},y^{\prime \prime }=\underset{\text{scale}}{\underbrace{m(m-1)}}{x}^{\stackrel{\text{power dropped 2}}{\overbrace{m-2}}}.$

WHY. Each derivative peels one off the exponent and leaves the old exponent behind as a multiplying number. After two derivatives the power has dropped by $2$ and we have collected the product $m(m-1)$ out front. That product will be the star of the algebra to come.

PICTURE. The exponent value plotted as it slides $m\to m-1\to m-2$, with the accumulated scale factor written beside each.

---

Step 4 — The powers cancel and the equation collapses

WHAT. Substitute $y,y^{\prime },y^{\prime \prime }$ into $a{x}^2{y}^{\prime \prime }+bx{y}^{\prime }+cy=0$:

$a{x}^2\cdot m(m-1)x^{m-2}+bx\cdot m{x}^{m-1}+c{x}^m=0.$

Now combine the powers of $x$ term by term:

$\underset{\text{first term}}{\underbrace{x^2\cdot x^{m-2}=x^m}},\underset{\text{second term}}{\underbrace{x\cdot x^{m-1}=x^m}},\underset{\text{third term}}{\underbrace{x^0\cdot x^m=x^m}}.$

Every single term became a number times $x^m$:

$[am(m-1)+bm+c]x^m=0.$

WHY. This is the whole point of the equidimensional balance from Step 1. The $x^2$ in front cancelled the two powers the second derivative removed; the $x^1$ cancelled the one the first derivative removed. So $x^m$ factors out cleanly and we are left with a pure number puzzle inside the brackets.

PICTURE. Three coloured tiles, each collapsing to the common height $x^m$, then merging into one bracket.

---

Step 5 — The indicial equation: a quadratic decides everything

WHAT. For any $x\ne 0$ the factor $x^m$ is not zero, so the bracket must vanish:

$am(m-1)+bm+c=0.$

Expand $m(m-1)=m^2-m$ and gather:

$a{m}^2+\underset{\text{watch this shift!}}{\underbrace{(b-a)}}m+c=0.$

This is the indicial (auxiliary) equation — the same idea as the Characteristic / Auxiliary Equation but for powers instead of exponentials.

WHY the $(b-a)$ and not just $b$? The bend term $x^2{y}^{\prime \prime }$ produced $m(m-1)=m^2-m$, and that hidden $-m$ steals an $a$ from the linear coefficient. This is the single most common slip — the linear coefficient is $b-a$, not $b$.

PICTURE. The quadratic $f(m)=a{m}^2+(b-a)m+c$ drawn as a parabola; its crossings of the horizontal axis are the roots $m$ we seek.

---

Step 6 — Case A: two distinct real roots

WHAT. If the parabola crosses the axis at two different real points $m_1\ne m_2$, we get two power solutions and combine them freely:

$y=C_1{x}^{m_1}+C_2{x}^{m_2}.$

Here $C_1,C_2$ are arbitrary constants (a 2nd-order ODE has a 2-parameter family of answers).

WHY. $x^{m_1}$ and $x^{m_2}$ are genuinely different shapes (independent), so no combination of one can fake the other — together they span every solution.

PICTURE. Two power curves of different exponents (e.g. $x^4$ growing, $x^{-1}$ decaying) plotted for $x>0$.

---

Step 7 — Case B: a repeated root, and where $\ln x$ hides

WHAT. If the parabola just touches the axis (its lowest point sits on it), the two roots merge into one value $m$. Now we only have ONE power $x^m$. The second solution is $y=(C_1+C_2\ln x)x^m,$ and the mystery is: why $\ln x$?

WHY. Undisguise the equation with $x=e^t$, i.e. $t=\ln x$ (this is the Reduction of Order / substitution trick the parent proves). In the $t$-world the equation becomes constant-coefficient, and a repeated root there gives the pair $e^{mt}$ and $t{e}^{mt}$. Translate back: $e^{mt}=x^m$ and $t{e}^{mt}=(\ln x)x^m$. The extra partner is a factor of $t=\ln x$, NOT a factor of $x$. Writing $x\cdot x^m$ would be copying the wrong world.

PICTURE. Left: the parabola tangent to the axis (double root). Right: $x^2$ and its partner $(\ln x)x^2$ pulling apart so you can see they are independent.

---

Step 8 — Case C: complex roots, a curve that spins as it grows

WHAT. If the parabola never touches the axis, the roots are complex: $m=\alpha \pm i\beta$, where $\alpha$ is the real part and $\beta$ the imaginary part, and $i=\sqrt{-1}$. The solution is $y=x^{\alpha }[C_1\cos (\beta \ln x)+C_2\sin (\beta \ln x)].$

WHY. Split $x^{\alpha +i\beta }=x^{\alpha }\cdot x^{i\beta }$. The strange piece $x^{i\beta }=e^{i\beta \ln x}$, and by Euler's Formula $e^{i\theta }=\cos \theta +i\sin \theta$ with $\theta =\beta \ln x$. So the imaginary exponent becomes a real oscillation in the variable $\ln x$. The $x^{\alpha }$ sets the overall growth/decay envelope; the cosine/sine make the curve wobble as $x$ increases — a spiral in disguise.

PICTURE. An oscillation whose horizontal axis is $\ln x$ (so wiggles bunch up near $x=0$ and stretch out for large $x$), riding an $x^{\alpha }$ envelope.

---

Step 9 — The degenerate edge: what happens at $x=0$ and for $x<0$

WHAT. At $x=0$ the coefficient $a{x}^2$ of $y^{\prime \prime }$ vanishes, so the equation loses its leading term — $x=0$ is a singular point (specifically a regular singular point; see Frobenius Method & Regular Singular Points). For $x<0$ the raw $\ln x$ is undefined.

WHY / the fix. Solutions like $x^{-1}$ or $\ln x$ can blow up or become undefined near $x=0$, so we always restrict to one side. For $x<0$ replace $x^m\to |x{|}^m$ and $\ln x\to \ln |x|$ — the algebra of Steps 3–5 is untouched because it never cared about the sign, only the power arithmetic.

PICTURE. The $x$-axis split: a green working region $x>0$, a red forbidden point at $x=0$, and the mirrored region $x<0$ handled with $|x|$.

---

The one-picture summary

Everything above, compressed: guess a power → derivatives slide the exponent → the $x$-powers cancel → a quadratic pops out → its roots (real/repeated/complex) pick one of three answer-shapes.

Recall Feynman: the whole walkthrough in plain words

We have an equation that doesn't care how big we draw $x$ — zoom in or out, it looks the same. Shapes that survive zooming are power curves, $x$-to-some-power, so we guess one. When we differentiate a power, the exponent slides down and drops a number in front; the $x$'s stuck to each derivative slide it right back up. Because of that perfect cancellation, the whole differential equation shrinks into one little quadratic asking "which powers fit?" Solve the quadratic. If it has two clean answers, we get two power curves and add them. If it has a double answer we're one solution short, so we peek into the disguised $x=e^t$ world where the missing partner is an "extra $t$" — translate it back and that $t$ becomes $\ln x$. If the quadratic's answers are imaginary, Euler's formula turns them into a curve that gently spins as it grows, giving cosines and sines of $\ln x$. And we always remember: at $x=0$ the leading term dies, so we stay on one side, using $|x|$ if we wander left.

Recall

The trial solution for a Cauchy-Euler equation ::: $y=x^m$ After substitution, every term becomes a number times ::: $x^m$ The indicial equation for $a{x}^2{y}^{\prime \prime }+bx{y}^{\prime }+cy=0$ ::: $am(m-1)+bm+c=0$, i.e. $a{m}^2+(b-a)m+c=0$ Repeated root $m$ gives the extra factor ::: $\ln x$ (not $x$) Complex roots $\alpha \pm i\beta$ give ::: $x^{\alpha }[C_1\cos (\beta \ln x)+C_2\sin (\beta \ln x)]$ Why $x=0$ is special ::: the coefficient of $y^{\prime \prime }$ vanishes there — a singular point