Visual Tools
Calculators
Tables
Mathematical Keyboard
Converters
Other Tools


Differentials






Separating dy and dx


The derivative dydx\frac{dy}{dx} is defined as a limit—a single object, not a fraction. But Leibniz notation invites the question: can dydy and dxdx stand on their own? Differentials answer yes, with precise meaning. The symbol dxdx represents an independent small change in xx, and dy=f(x)dxdy = f'(x)\,dx represents the corresponding change in yy predicted by the tangent line.

This formalism does more than justify notation. It provides a direct tool for approximating function values near a known point, estimating how errors in measurement propagate through calculations, and explaining why Leibniz notation behaves like fraction arithmetic in the chain rule and in integration.



The Differential dx


The differential dxdx is an independent variable. It represents a change in xx—an increment away from a given value. Unlike the hh or Δx\Delta x in the limit definition, dxdx is not required to approach zero. It is a finite quantity that can be positive, negative, or zero.

There is no formula for dxdx—it is chosen freely. Choosing dx=0.1dx = 0.1 means considering what happens when xx shifts by 0.10.1. Choosing dx=2dx = -2 means shifting xx by 2-2. The differential dxdx sets the scale for the approximation that follows.

The notation is consistent with Leibniz notation for the derivative. When dy/dxdy/dx is treated as a ratio of differentials rather than a limit symbol, the algebraic manipulations that make Leibniz notation powerful—cancellation in the chain rule, separation in differential equations—become formally valid rather than merely suggestive.

The Differential dy


Given y=f(x)y = f(x) where ff is differentiable, the differential of yy is defined as

dy=f(x)dxdy = f'(x) \cdot dx


The differential dydy depends on two things: the point xx (which determines the slope f(x)f'(x)) and the increment dxdx (which determines the scale). For fixed xx, dydy is a linear function of dxdx—doubling dxdx doubles dydy.

Geometrically, dydy is the vertical change along the tangent line at xx when the horizontal position shifts by dxdx. The tangent line rises (or falls) at rate f(x)f'(x), so a horizontal shift of dxdx produces a vertical shift of f(x)dxf'(x) \cdot dx.

For a linear function f(x)=mx+bf(x) = mx + b, the differential dy=mdxdy = m \cdot dx equals the actual change in ff exactly, because the tangent line to a linear function is the function itself. For nonlinear functions, dydy is an approximation—exact in the limit as dx0dx \to 0, and increasingly approximate as dxdx grows.

dy Versus Δy


Two quantities measure the change in yy when xx changes by dxdx:

Δy=f(x+dx)f(x)(actual change)\Delta y = f(x + dx) - f(x) \qquad \text{(actual change)}


dy=f(x)dx(tangent line estimate)dy = f'(x) \cdot dx \qquad \text{(tangent line estimate)}


The actual change Δy\Delta y follows the curve. The differential dydy follows the tangent line. The difference Δydy\Delta y - dy is the error introduced by the linear approximation.

For small dxdx, the error Δydy\Delta y - dy is much smaller than dxdx itself. Precisely, limdx0Δydydx=0\lim_{dx \to 0} \frac{\Delta y - dy}{dx} = 0—the error shrinks faster than the increment. This is what makes the tangent line a good approximation near the point of tangency.

For larger dxdx, the gap between Δy\Delta y and dydy widens. The tangent line is a local tool: reliable near xx, increasingly unreliable far from it. The curvature of ff—governed by the second derivative—determines how quickly the approximation deteriorates.

Linear Approximation


The differential formula rearranges into an approximation for function values:

f(x+dx)f(x)+f(x)dxf(x + dx) \approx f(x) + f'(x) \cdot dx


This is the linearization of ff at xx. Given a known value f(x)f(x) and the slope f(x)f'(x), the tangent line estimates ff at nearby points.

To approximate 4.03\sqrt{4.03}: take f(x)=xf(x) = \sqrt{x}, x=4x = 4, dx=0.03dx = 0.03. Then f(4)=2f(4) = 2 and f(x)=12xf'(x) = \frac{1}{2\sqrt{x}}, so f(4)=14f'(4) = \frac{1}{4}. The estimate is 2+14(0.03)=2.00752 + \frac{1}{4}(0.03) = 2.0075. The actual value is 4.032.00749...\sqrt{4.03} \approx 2.00749...—the approximation is accurate to five decimal places.

The linearization can also be written as L(x)=f(a)+f(a)(xa)L(x) = f(a) + f'(a)(x - a), the tangent line at x=ax = a used as a function. For xx near aa, L(x)f(x)L(x) \approx f(x). The quality of the approximation depends on two factors: the size of xa|x - a| and the magnitude of ff'' near aa, which controls how quickly the curve diverges from its tangent.

Error Estimation


If a quantity xx is measured with error dxdx, the computed value f(x)f(x) inherits an error approximately equal to the differential:

absolute error in fdy=f(x)dx\text{absolute error in } f \approx |dy| = |f'(x)| \cdot |dx|


The derivative acts as an error amplification factor. Where f(x)|f'(x)| is large, small measurement errors in xx produce large errors in f(x)f(x). Where f(x)|f'(x)| is small, errors are suppressed.

The relative error normalizes by the function value:

dyf(x)=f(x)f(x)dx\frac{|dy|}{|f(x)|} = \frac{|f'(x)|}{|f(x)|} \cdot |dx|


The ratio f(x)f(x)=ddx[lnf(x)]\frac{f'(x)}{f(x)} = \frac{d}{dx}[\ln|f(x)|] is the logarithmic derivative. Relative error in the output equals the logarithmic derivative times the absolute error in the input.

For f(x)=xnf(x) = x^n: f(x)f(x)=nx\frac{f'(x)}{f(x)} = \frac{n}{x}, so the relative error in xnx^n is nn times the relative error in xx. Squaring a measurement doubles its relative error. Cubing triples it. This scaling rule is a standard tool in experimental science for propagating uncertainties through power-law relationships.

Differentials of Multiple Variables


When a formula involves several measured quantities, each contributes to the total error. If z=f(x,y)z = f(x, y), the total differential is

dz=fxdx+fydydz = \frac{\partial f}{\partial x}\,dx + \frac{\partial f}{\partial y}\,dy


Each partial derivative weights the contribution of the corresponding variable's error. The total differential estimates how zz responds to simultaneous small changes in all inputs.

For the area of a rectangle A=lwA = lw: dA=wdl+ldwdA = w\,dl + l\,dw. If l=10±0.1l = 10 \pm 0.1 and w=5±0.05w = 5 \pm 0.05, then dA=5(0.1)+10(0.05)=1.0dA = 5(0.1) + 10(0.05) = 1.0. The area is approximately 50±1.050 \pm 1.0.

For the volume of a cylinder V=πr2hV = \pi r^2 h: dV=2πrhdr+πr2dhdV = 2\pi r h\,dr + \pi r^2\,dh. The radius error is amplified by 2πrh2\pi r h and the height error by πr2\pi r^2. Since rr enters as a square, its error contributes more heavily—consistent with the power-law scaling of relative error.

This extension uses partial derivatives, which generalize the single-variable derivative to functions of several variables. The total differential remains a linear approximation, now in multiple dimensions.

Differentials and Leibniz Notation


Differentials retroactively justify the algebraic behavior of Leibniz notation.

The chain rule states dydx=dydududx\frac{dy}{dx} = \frac{dy}{du} \cdot \frac{du}{dx}. In differential form: dy=dydududy = \frac{dy}{du}\,du and du=dudxdxdu = \frac{du}{dx}\,dx. Substituting the second into the first gives dy=dydududxdxdy = \frac{dy}{du} \cdot \frac{du}{dx}\,dx, which is dy=dydxdxdy = \frac{dy}{dx}\,dx. The intermediate variable dudu cancels as though these were fractions—and with differentials, they are.

Integration notation f(x)dx\int f(x)\,dx also uses the differential dxdx meaningfully. The substitution rule u=g(x)u = g(x), du=g(x)dxdu = g'(x)\,dx replaces dxdx with an expression involving dudu—a literal change of variable in the differential. The notation is not merely symbolic; it reflects the algebraic structure of differentials.

Separation of variables in differential equations—writing dydx=g(x)h(y)\frac{dy}{dx} = g(x)h(y) as dyh(y)=g(x)dx\frac{dy}{h(y)} = g(x)\,dx and integrating both sides—depends on treating dydy and dxdx as independent objects that can be rearranged. Differentials make this manipulation rigorous rather than heuristic.