指数基本不等式直觉探究：m<n时mⁿ与nᵐ的大小及普适性证明

Great question! This is a classic exponential inequality problem that feels counterintuitive at first, but framing it with calculus makes the pattern clear. Let’s break this down from intuition to a formal proof.

Intuition Behind the Problem #

Instead of directly comparing (m^n) and (n^m), let’s take the natural logarithm of both (since log is a monotonically increasing function, the inequality direction stays the same). We need to compare:
(n\ln m) vs (m\ln n)

Divide both sides by (mn) (positive, so inequality direction doesn’t change) — now we’re comparing:
(\frac{\ln m}{m}) vs (\frac{\ln n}{n})

This transforms our problem into analyzing the function (f(x) = \frac{\ln x}{x}). If we can figure out how (f(x)) behaves as (x) increases, we can directly compare (f(m)) and (f(n)) for (m < n).

When Does (m^n > n^m) Hold? (The "Universal" Rules) #

First, let’s get the exceptions out of the way — the conclusion isn’t fully universal. Here’s the breakdown for positive integers (m < n):

• Case 1: (m = 1)
  (1^n = 1) and (n^1 = n). Since (n > 1), (1^n < n^m) every time.
• Case 2: (m = 2)
  • (n = 3): (2^3 = 8 < 3^2 = 9) (the only exception for (m=2))
  • (n \geq 4): (2^n > n^2) (e.g., (2⁴⁼¹⁶⁼⁴2), (2^5=32>25=52), and this gap only widens as (n) grows)
• Case 3: (m \geq 3)
  For any (n > m), (m^n > n^m) holds universally. Your examples (3^7 > 7^3) and (4^5 > 5^4) fit this rule perfectly.

Rigorous Proof Using Calculus #

Let’s formalize the intuition with the function (f(x) = \frac{\ln x}{x}) for (x > 0):

1. Compute the derivative to find monotonicity:
   [
   f'(x) = \frac{\frac{1}{x} \cdot x - \ln x \cdot 1}{x^2} = \frac{1 - \ln x}{x^2}
   ]
2. Analyze the derivative:
   • When (0 < x < e) (where (e \approx 2.718), the base of natural logs), (1 - \ln x > 0), so (f'(x) > 0). This means (f(x)) is strictly increasing on ((0, e)).
   • When (x > e), (1 - \ln x < 0), so (f'(x) < 0). This means (f(x)) is strictly decreasing on ((e, \infty)).
   • At (x = e), (f(x)) reaches its maximum value (\frac{1}{e}).

Now let’s apply this to our cases:

• For (m=1, n>1):
  (f(1) = \frac{\ln 1}{1} = 0), and (f(n) = \frac{\ln n}{n} > 0) (since (n>1)). So (f(1) < f(n)) → (\frac{\ln 1}{1} < \frac{\ln n}{n}) → (n\ln1 < m\ln n) → (1^n < n^1).

• For (m=2, n=3):
  (f(2) = \frac{\ln2}{2} \approx 0.3466), (f(3) = \frac{\ln3}{3} \approx 0.3662). Since (2 < 3 < e), (f(x)) is increasing here, so (f(2) < f(3)) → (3\ln2 < 2\ln3) → (2^3 < 3^2).

• For (m=2, n\geq4):
  (n \geq4 > e), so (f(x)) is decreasing on ((e, \infty)). Thus (f(n) \leq f(4) = \frac{\ln4}{4} = \frac{\ln2}{2} = f(2)). This gives (\frac{\ln n}{n} \leq \frac{\ln2}{2}) → (2\ln n \leq n\ln2) → (n^2 \leq 2^n) (equality only when (n=4)).

• For (m\geq3, n>m):
  Both (m) and (n) are greater than (e), and (f(x)) is decreasing on ((e, \infty)). Since (m < n), (f(m) > f(n)) → (\frac{\ln m}{m} > \frac{\ln n}{n}) → (m\ln n < n\ln m) → (n^m < m^n).

Summary #

To recap the rule of thumb for positive integers (m < n):

• (1^n < n^1) always
• (2^3 < 3^2) (the only small-case exception)
• (2^n \geq n^2) for (n \geq4)
• (m^n > n^m) for all (m \geq3) and (n > m)

内容的提问来源于stack exchange，提问作者user_1_1_1