Multivariable Calculus · Lesson 2 of 3 · ~6 min

Partial derivatives

Predict

At the saddle's center $(0, 0)$ the map shows higher ground east and west, lower ground north and south. What are the two slice slopes $f_x$ and $f_y$ there?

Hold that thought. Press saddle x²−y² and park the probe on the exact center — watch both arrows.
Try
Try itOn the bowl, drag the probe to $(1.00,\ 0.50)$. The stats read $f_x = 2.000$ and $f_y = 1.000$, and the two arrows at the probe draw those slopes. Slide the probe left across the $y$-axis and watch the $x$-arrow flip.
Notice

At $(1, 0.5)$ the east arrow is twice the north arrow: stepping east, the bowl climbs at rate $2$; stepping north, at rate $1$. Cross to $x < 0$ and the $x$-arrow points west — that slice now descends. At the saddle's center both arrows vanish and the stats read $0.000$ and $0.000$, yet the map around the probe is anything but featureless.

Explain

A partial derivative is an ordinary derivative applied to a slice. Freeze $y$ at $0.5$ and the bowl collapses to a one-variable function $g(x) = x^2 + 0.25$, whose derivative $g'(x) = 2x$ is the partial $f_x$. Differentiate treating $x$ as the constant instead and $f_y = 2y$. At $(1, 0.5)$ that gives $f_x = 2$ and $f_y = 1$ — the two numbers on the stats. The dashed crosshair through the probe shows the two slices being measured.

PREREQUISITEEach arrow is a tangent slope from single-variable calculus — one slice at a time.

The saddle settles the prediction: $f_x = 2x$ and $f_y = -2y$ both vanish at the center, exactly as at a bowl's bottom — but here the $x$-slice curves up while the $y$-slice curves down. Two flat slices do not make a peak or a pit. Points where both partials vanish are called critical points; deciding which kind takes more than the two slopes.

Apply
Checkpoint — prove it to yourself

On the bowl at $(1, 0.5)$ the stats read $f_x = 2.000$. What does that number mean?

For the bowl, $f_x = 2x$ and $f_y = 2y$. What do the stats read at $(-1,\ 1.5)$?

Transfer: $g(x, y) = 3x + x^2 y$. What is $g_y$, the partial with respect to $y$?

Key idea

A partial derivative freezes every variable but one and differentiates the slice that remains. $f_x$ and $f_y$ are the surface's slopes along the two axis directions — and both can vanish at a point that is neither a peak nor a pit.

← All lessons