The p-norm is defined as:
$$ \ \|x\|_p=\left(|x_1|^p+|x_2|^p+\dotsb+|x_n|^p\right)^{\frac{1}{p}} $$
When $p<1$, this is no longer a "norm" because it violates the triangle inequality (- it is super additive and not subadditive). However, it is still valid to ask, what is its limit when p goes to 0?
My guess is that:
- If all coordinates are 0, then $l_p=0$, and it remains like this when p=0.
- If exactly one coordinate, say $x_i$, is non-zero, then $l_p=x_i$, and it remains like this when p=0.
- If more than one coordinate (say, $x_i$ and $x_j$) are non-zero, then $l_p>x_i$, and because the exponent goes to $\infty$, the value of $l_p$ also goes to $\infty$.
Is this correct?
(Note that this is not equal to the $l_0$ "norm" = the number of nonzero elements. This is also not equal to the scaled norm, in which there is an additional $1/n$ factor).