关于微分记号$df=g(x)dx$的精确定义咨询

Great question—this is one of those notation quirks that feels intuitive (and super handy for tricks like substitution) but makes you stop and go, "Wait, does this actually have a formal definition?" The short answer is yes, and it depends a bit on the context you're working in. Let's break down the main rigorous interpretations:

1. Elementary Calculus: Differentials as Linear Approximations #

In intro calc, we can define the differential of $f$ at a point $x$ as the linear part of the function's change near $x$. Since you already understand $rac{d}{dx}f(x) = g(x)$, here's how it translates:

• Let $\Delta x$ be a small increment in $x$. The actual change in $f$ is $\Delta f = f(x+\Delta x) - f(x) = g(x)\Delta x + o(\Delta x)$, where $o(\Delta x)$ is an error term that vanishes faster than $\Delta x$ as $\Delta x o 0$.
• The differential $df$ is exactly that linear approximation: we set $dx = \Delta x$, so $df = g(x)dx$. This is the definition that justifies using $df = g(x)dx$ for things like estimating small changes or doing u-substitution in integrals.

2. Advanced Analysis/Differential Geometry: Differentials as 1-Forms #

Once you move to multivariable calculus or more abstract math, this notation gets a precise meaning using differential 1-forms:

• A 1-form is a linear map that acts on tangent vectors at each point in space. For single-variable functions, the tangent space at any $x$ is just $\mathbb{R}$ (think of it as the line of possible "directions" you can move from $x$).
• The differential $df$ is the 1-form that takes a tangent vector $v$ (which we can identify with $dx$, the increment in $x$) and returns $df(v) = f'(x)v = g(x)v$.
• We define the standard 1-form $dx$ as the identity map on tangent vectors: $dx(v) = v$. So when we write $df = g(x)dx$, we're saying that for any tangent vector $v$, $df(v) = g(x) \cdot dx(v)$—which is exactly a formal way to encode the derivative's meaning.

This is the strict definition used in fields like differential geometry, where we need to work with curved spaces or higher-dimensional functions.

3. Riemann Integration: Notational Shorthand for Antiderivatives #

In the context of integration, $df = g(x)dx$ is often used as a shorthand to say that $f$ is an antiderivative of $g$. In other words, $\int g(x)dx = f(x) + C$. While this is more of a convenience than a standalone definition, it ties back to the linear approximation idea—integrating $g(x)dx$ is just summing up all those tiny linear approximations $df$ over an interval.

A quick note on "algebraic manipulation" #

You might have seen people rearrange $df = g(x)dx$ to $dx = df/g(x)$ (when $g(x) \neq 0$) for substitution. Even though this feels like basic algebra, it's actually justified by the formal definitions above. For example, if $u = f(x)$, then $du = f'(x)dx = g(x)dx$, so substituting into an integral just replaces one set of linear approximations ($g(x)dx$) with another ($du$)—the notation works because it mirrors the underlying linearity.

内容的提问来源于stack exchange，提问作者MH Yip