Analyzing Quasiconcave to Quasiconvex Transformation

Let's examine the relationship between a quasiconcave function and its negative. A function $f(x)$ is defined as quasiconcave if for any two points $x_1$ and $x_2$ in its domain, and for any scalar $\lambda$ such that $0 \le \lambda \le 1$, the following inequality holds:

$f(\lambda x_1 + (1-\lambda) x_2) \ge \min\{f(x_1), f(x_2)\}$

Now consider the function $g(x) = -f(x)$. For $g(x)$ to be quasiconvex, it must satisfy:

$g(\lambda x_1 + (1-\lambda) x_2) \le \max\{g(x_1), g(x_2)\}$

Substituting $g(x) = -f(x)$, we get:

$-f(\lambda x_1 + (1-\lambda) x_2) \le \max\{-f(x_1), -f(x_2)\}$

The term $\max\{-f(x_1), -f(x_2)\}$ is equivalent to $-\min\{f(x_1), f(x_2)\}$. So the inequality becomes:

$-f(\lambda x_1 + (1-\lambda) x_2) \le -\min\{f(x_1), f(x_2)\}$

Multiplying by -1 reverses the inequality sign:

$f(\lambda x_1 + (1-\lambda) x_2) \ge \min\{f(x_1), f(x_2)\}$

This is exactly the definition of $f(x)$ being quasiconcave. Therefore, if $f(x)$ is quasiconcave, then $-f(x)$ is quasiconvex. Statement 1 is correct.

Linear Function Properties Explained

A linear function can be represented as $f(x) = ax + b$. For such a function:

• Concavity: $f(\lambda x_1 + (1-\lambda) x_2) = a(\lambda x_1 + (1-\lambda) x_2) + b = \lambda(ax_1 + b) + (1-\lambda)(ax_2 + b) - \lambda b - (1-\lambda)b + b = \lambda f(x_1) + (1-\lambda) f(x_2)$. Since $f(\lambda x_1 + (1-\lambda) x_2) = \lambda f(x_1) + (1-\lambda) f(x_2)$, it satisfies the condition for both concavity (where the inequality is $\ge$) and convexity (where the inequality is $\le$) because equality holds.
• Quasiconcavity/Quasiconvexity: Since linear functions are both concave and convex, they are also both quasiconcave and quasiconvex. The condition $f(\lambda x_1 + (1-\lambda) x_2) \ge \min\{f(x_1), f(x_2)\}$ and $f(\lambda x_1 + (1-\lambda) x_2) \le \max\{f(x_1), f(x_2)\}$ are both met because equality holds in the definition.

Thus, Statement 2 is correct.

Concave vs. Quasiconcave Relationship

The relationship between concave and quasiconcave functions is as follows:

• Concave implies Quasiconcave: If a function $f(x)$ is concave, it means $f(\lambda x_1 + (1-\lambda) x_2) \ge \lambda f(x_1) + (1-\lambda) f(x_2)$. Since $\lambda f(x_1) + (1-\lambda) f(x_2) \ge \min\{f(x_1), f(x_2)\}$ is always true (because $f(x_1)$ and $f(x_2)$ are the boundary values, and the weighted average lies between them or equals one of them), it follows that $f(\lambda x_1 + (1-\lambda) x_2) \ge \min\{f(x_1), f(x_2)\}$. This confirms that any concave function is indeed quasiconcave.
• Converse is not true: There are functions that are quasiconcave but not concave. For instance, consider the function $f(x) = -|x|$. This function is V-shaped, peaking at $x=0$. It is quasiconcave, but it is not concave because the line segment connecting two points on the graph lies below the graph for most points (except the peak). The second derivative test confirms this: $f''(x)=0$ for $x \ne 0$, but the function is not linear. A simpler example often used is $f(x) = \sqrt{x}$ for $x \ge 0$, which is concave. Consider $f(x)=1/x$ for $x>0$, which is neither concave nor convex, but is quasiconcave.

Therefore, Statement 3 is correct.

Convex Function Converse Analysis

Let's analyze the statement: "Any convex function is quasiconvex and its converse also holds."

• Convex implies Quasiconvex: If a function $f(x)$ is convex, it satisfies $f(\lambda x_1 + (1-\lambda) x_2) \le \lambda f(x_1) + (1-\lambda) f(x_2)$. Since $\lambda f(x_1) + (1-\lambda) f(x_2) \le \max\{f(x_1), f(x_2)\}$ is always true, it follows that $f(\lambda x_1 + (1-\lambda) x_2) \le \max\{f(x_1), f(x_2)\}$. This means any convex function is indeed quasiconvex.
• Converse: Quasiconvex does not imply Convex: This part of the statement claims that if a function is quasiconvex, it must be convex. This is not generally true. A function is quasiconvex if its upper-level sets $\{x | f(x) \le \alpha\}$ are convex for all $\alpha$. A function is convex if its epigraph $\{(x, y) | y \ge f(x)\}$ is a convex set. While convex functions have convex epigraphs (and thus convex upper-level sets), the converse is not always true. For example, the function $f(x) = x^4 - 4x^3 + 4x^2 + x$ is quasiconvex but not convex over certain intervals. A simpler counterexample is $f(x) = -1/x$ for $x>0$, which is quasiconvex but not convex. The defining inequality for quasiconvexity, $f(\lambda x_1 + (1-\lambda) x_2) \le \max\{f(x_1), f(x_2)\}$, is weaker than the condition for convexity.

Since the converse ("its converse also holds") is false, Statement 4 is not correct.

Conclusion on Incorrect Statement

The question asks to identify the statement that is not correct. Based on our analysis, Statement 4, which claims that the converse relationship holds for convex and quasiconvex functions (i.e., any quasiconvex function is convex), is false. While all convex functions are quasiconvex, not all quasiconvex functions are convex.