- #1
maze
- 662
- 4
What is your favorite real function?
My top 5 are
1. f:R->R
[tex]f(x) = e^{-\frac{1}{x^2}}[/tex]
for x > 0; 0 for [itex]x \le 0[/itex]. This function is infinitely differentiable everywhere, but the taylor series tells you nothing about half of the function.
2. f:[0,1]\Q->R
[tex]f(x) = \sum_{n=1}^\infty 2^{-n} ln |x - q_n|[/tex]
for all rational qn in [0,1]. This function is ultra-spikey. It has a singularity at every rational number between 0 and 1, but yet the area under the curve is finite.
3. f:Rd->R
[tex]f(x) = (\epsilon+|x|^2)^\frac{1}{2}[/tex]
This smooth function approximates |x|, but without the sharp point at 0.
4. The Devil's staircase. This function is continuous, has derivative zero almost everywhere, but yet it is nonzero.
5. f:Rd->R
[tex]f(x) = \frac{x_1}{\sqrt{x_1^2 + x_2^2 + ... + x_k^2}}[/tex]
for [itex]1 < k \le d[/itex]. This function is discontinuous, but the dimension of the discontinuity can be varied by adjusting k. For example if d=3 and k=3, then it is a 3D function discontinuous only at a point. If you take a derivative of it, you get a singularity (instead of a delta distribution).
My top 5 are
1. f:R->R
[tex]f(x) = e^{-\frac{1}{x^2}}[/tex]
for x > 0; 0 for [itex]x \le 0[/itex]. This function is infinitely differentiable everywhere, but the taylor series tells you nothing about half of the function.
2. f:[0,1]\Q->R
[tex]f(x) = \sum_{n=1}^\infty 2^{-n} ln |x - q_n|[/tex]
for all rational qn in [0,1]. This function is ultra-spikey. It has a singularity at every rational number between 0 and 1, but yet the area under the curve is finite.
3. f:Rd->R
[tex]f(x) = (\epsilon+|x|^2)^\frac{1}{2}[/tex]
This smooth function approximates |x|, but without the sharp point at 0.
4. The Devil's staircase. This function is continuous, has derivative zero almost everywhere, but yet it is nonzero.
5. f:Rd->R
[tex]f(x) = \frac{x_1}{\sqrt{x_1^2 + x_2^2 + ... + x_k^2}}[/tex]
for [itex]1 < k \le d[/itex]. This function is discontinuous, but the dimension of the discontinuity can be varied by adjusting k. For example if d=3 and k=3, then it is a 3D function discontinuous only at a point. If you take a derivative of it, you get a singularity (instead of a delta distribution).