Why Compare a Derivative With Nearby Slopes?
A symbolic derivative gives a formula for instantaneous change. A nearby difference quotient gives the average change over a short interval. If the derivative formula is correct, these quantities should be close when the interval is sufficiently small and lies within the function's domain.
This comparison is a practical check, not a replacement for differentiation or proof. It can reveal a sign error, a missing coefficient, or an implausible value at a selected input. It cannot establish that a derivative formula is correct everywhere: numerical checks examine only particular inputs and step sizes.
The Difference Quotient as a Numerical Check
A positive \(h\) compares the value at \(x\) with a point to its right; a negative \(h\) compares it with a point to its left. Using both directions can be informative. If the function is defined only on one side of \(x\), use steps from that side only. In every case, \(h\) must be nonzero and \(x+h\) must be in the domain.
The practical comparison has three parts: calculate the derivative symbolically, evaluate it at the chosen input, and calculate one or more difference quotients near that input. The estimates need not equal the derivative exactly. They should be interpreted in light of the step sizes: a larger step averages change over a longer interval, while a smaller step usually probes closer to the instantaneous slope.
Why Nearby Quotients Approach the Derivative
The reason this check works is contained in the definition of the derivative. For a differentiable function, the difference between \(Q_x(h)\) and \(f'(x)\) becomes arbitrarily small as \(h\) approaches zero. A polynomial example makes the size and direction of that difference explicit.
Let \(f(x)=x^3-2x+1\). Expanding the numerator of the difference quotient gives:
Direct differentiation gives \(f'(x)=3x^2-2\). Therefore, for this polynomial, the forward quotient differs from the derivative by \(3xh+h^2\). For a fixed input \(x\), both extra terms approach zero as \(h\) approaches zero. This verifies directly that the nearby quotient approaches \(f'(x)\) in this example. It also shows why a finite-step estimate may not match exactly.
The expansion is specific to this polynomial, but the general principle comes from the definition: differentiability means precisely that the difference quotient has limit \(f'(x)\). That limit justifies the comparison; it does not guarantee that any one quotient with a nonzero step will be close enough for every purpose.
Worked Comparisons
Worked Example: Checking a Cubic Derivative in Both Directions
Take \(f(x)=x^3-2x+1\) and compare the symbolic derivative at \(x=2\) with nearby forward and backward quotients using steps of \(0.1\).
First differentiate and evaluate:
Now calculate the function value at the input:
For the forward step \(h=0.1\), the new input is \(2.1\):
For the backward step \(h=-0.1\), the new input is \(1.9\):
The two nearby slopes, \(10.61\) and \(9.41\), lie on opposite sides of the symbolic value \(10\). That is plausible: the function's slopes vary across the interval, and neither secant slope is the tangent slope itself.
As an additional check, use the central quotient, which compares the two endpoints on either side of \(2\):
The central estimate is also close to \(10\). The symbolic value checks out at this input. Substituting \(x=2\) into the derivative formula confirms the comparison: \(3(2)^2-2=10\).
Worked Example: Checking a Rational Derivative
Let \(f(x)=\frac{1}{x+1}\) and check the derivative at \(x=1\) with steps \(0.1\) and \(-0.1\). Both nearby inputs are in the domain, which excludes \(x=-1\).
Write the function as a negative power and differentiate:
At \(x=1\), the symbolic value is:
The function value is \(f(1)=\frac12\). For \(h=0.1\), \(x+h=1.1\), so \(f(1.1)=\frac{1}{2.1}=\frac{10}{21}\). Then:
For \(h=-0.1\), \(x+h=0.9\), and \(f(0.9)=\frac{1}{1.9}=\frac{10}{19}\). Thus:
Both estimates are near \(-0.25\), with the forward estimate slightly above it and the backward estimate slightly below it. The exact symbolic check is \(f'(1)=-1/4\), while the quotients are \(-5/21\) and \(-5/19\). Their signs and sizes are consistent with that result.
Worked Example: Spotting a Sign Error for a Negative Power
Consider \(f(x)=x^{-2}\) at \(x=2\). This example shows how a nearby slope can flag an incorrect positive sign in a derivative formula.
Apply the power rule and evaluate:
The function value is \(f(2)=2^{-2}=\frac14=0.25\). With \(h=0.1\), the new input is \(2.1=\frac{21}{10}\), and:
With \(h=-0.1\), the new input is \(1.9=\frac{19}{10}\), and:
Both estimates are negative and close to \(-0.25\), supporting the symbolic calculation. If a sign error produced \(+2x^{-3}\), its value at \(2\) would be \(+\frac14\), opposite in sign to both nearby slopes. The numerical check would flag that candidate. Substitution into the correct derivative gives \(-2(2^{-3})=-\frac14\), as required.
Common Pitfall: Reversing the Numerator
The quotient's numerator and denominator must use the same direction of change. If the step is \(h\), the numerator is \(f(x+h)-f(x)\), and the denominator is \(h\). Reversing the numerator but keeping the denominator unchanged reverses the sign of the estimate.
For example, take \(f(x)=x^2\), \(x=3\), and \(h=0.1\). The correct computation is:
The derivative is \(f'(x)=2x\), so \(f'(3)=6\), making \(6.1\) a plausible nearby estimate. The incorrect reversed numerator would give:
This negative value conflicts with the positive symbolic derivative. The correction is to keep the order consistent: when \(h=0.1\), subtract the value at \(3\) from the value at \(3.1\), then divide by \(0.1\). A negative estimate can be legitimate for a negative step, but the denominator must then be negative as well.
How to Interpret the Comparison
A useful check uses more than one nearby step when practical. If forward and backward quotients both have the expected sign and are reasonably close to the symbolic value, that is reassuring. If they differ sharply from the proposed derivative, first check the arithmetic, the sign of \(h\), the function values, and whether the chosen inputs are in the domain.
There are limits to the method. A step that is too large may give a secant slope that differs substantially from the derivative. A very small step can make decimal calculations sensitive to rounding: subtracting two nearly equal function values can lose accuracy. Keep exact fractions where convenient, and avoid treating rounded decimal agreement as an exact result.
Most importantly, agreement at selected inputs does not prove a formula for every input. A formula can pass a few numerical checks and still fail elsewhere. Differentiation rules and algebraic justification establish the result; nearby quotients are a diagnostic tool for catching possible mistakes.
Check Your Understanding
Use the difference quotient and the symbolic derivative to assess each comparison.
- For \(f(x)=x^2+1\) at \(x=2\), calculate the forward quotient with \(h=0.1\) and compare it with \(f'(2)\).
- For \(f(x)=x^{-1}\) at \(x=2\), should a step of \(h=-3\) be used? Explain using the domain of the function.
- Why can a forward quotient with a nonzero step differ from the derivative even when the derivative formula is correct?
- For \(f(x)=x^2\) at \(x=3\), identify the sign mistake in \(\frac{f(3)-f(3.1)}{0.1}\).
- Does agreement between numerical estimates and a proposed derivative at two inputs prove the formula is correct for all inputs? Explain.
