A function is convex if and only if

  • Thread starter Thread starter Arian.D
  • Start date Start date
  • Tags Tags
    Convex Function
Arian.D
Messages
101
Reaction score
0

Homework Statement


Show that a differentiable function f is convex if and only if the following inequality holds for each fixed point x0 in Rn:
f(x) ≥ f(x0) + ∇tf(x0)(x-x0) for all x in Rn, where ∇tf(x0) is the gradient vector of f at x0.


Homework Equations





The Attempt at a Solution



by definition of a convex function we have:

f(λx + (1-λ)x0) ≤ λf(x) + (1-λ)f(x0)

f(λ(x-x0)+x0) ≤ λ(f(x) - f(x0)) + f(x0)

f(λ(x-x0)+x0) - f(x0) ≤ λ(f(x) - f(x0))

Setting Δx = λx-x0 we'll have:

f(Δx + x0) - f(x0) ≤ λ(f(x) - f(x0))

I'm very hesitant to use this step, because vector division is not defined, but if we were dealing with real numbers (i.e, vectors of dimension 1) I could've gone further to obtain:

\frac{f(\Delta{x} + x_0) - f(x_0)}{\lambda} \leq f(x) - f(x_0)
\frac{f(\Delta{x} + x_0) - f(x_0)}{\lambda(x-x_0)}(x-x_0) \leq f(x) - f(x_0)
\frac{f(\Delta{x} + x_0) - f(x_0)}{\Delta{x}}(x-x_0) \leq f(x) - f(x_0)

taking limits from both sides as Δx goes to 0 gives us the desired result.

I don't know how to show that the converse is true, and also I don't know how to generalize what I've written to functions of several variables since vector division is not defined. any helps would be appreciated.
 
Physics news on Phys.org
Arian.D said:

Homework Statement


Show that a differentiable function f is convex if and only if the following inequality holds for each fixed point x0 in Rn:
f(x) ≥ f(x0) + ∇tf(x0)(x-x0) for all x in Rn, where ∇tf(x0) is the gradient vector of f at x0.


Homework Equations





The Attempt at a Solution



by definition of a convex function we have:

f(λx + (1-λ)x0) ≤ λf(x) + (1-λ)f(x0)

f(λ(x-x0)+x0) ≤ λ(f(x) - f(x0)) + f(x0)

f(λ(x-x0)+x0) - f(x0) ≤ λ(f(x) - f(x0))

Setting Δx = λx-x0 we'll have:

f(Δx + x0) - f(x0) ≤ λ(f(x) - f(x0))

I'm very hesitant to use this step, because vector division is not defined, but if we were dealing with real numbers (i.e, vectors of dimension 1) I could've gone further to obtain:

\frac{f(\Delta{x} + x_0) - f(x_0)}{\lambda} \leq f(x) - f(x_0)
\frac{f(\Delta{x} + x_0) - f(x_0)}{\lambda(x-x_0)}(x-x_0) \leq f(x) - f(x_0)
\frac{f(\Delta{x} + x_0) - f(x_0)}{\Delta{x}}(x-x_0) \leq f(x) - f(x_0)

taking limits from both sides as Δx goes to 0 gives us the desired result.

I don't know how to show that the converse is true, and also I don't know how to generalize what I've written to functions of several variables since vector division is not defined. any helps would be appreciated.

Most of theses types of results generalize almost immediately from the 1-dimensional case to the multivariate case, because if ##x_0, \, x \in \mathbb{R}^n## and ##p = x - x_0,## then
f \text{ convex in }\mathbb{R}^n \Rightarrow \phi(t) = f(x_0 + t p)\text{ is convex in } t \in \mathbb{R}, and conversly. So, you need to show that ##\phi(t)## is convex if and only if
\phi(t) \leq \phi(t_0) + \phi'(t_0) \, (t - t_0) \, \forall \, t, t_0.

RGV
 
There are two things I don't understand about this problem. First, when finding the nth root of a number, there should in theory be n solutions. However, the formula produces n+1 roots. Here is how. The first root is simply ##\left(r\right)^{\left(\frac{1}{n}\right)}##. Then you multiply this first root by n additional expressions given by the formula, as you go through k=0,1,...n-1. So you end up with n+1 roots, which cannot be correct. Let me illustrate what I mean. For this...

Similar threads

Back
Top