Skip to main content
10 events
when toggle format what by license comment
Nov 10, 2020 at 14:33 history edited Emil Jeřábek CC BY-SA 4.0
added 3626 characters in body
Nov 7, 2020 at 8:38 comment added user44143 You are correct. I'll leave up the comment so others can see the problem with the idea.
Nov 7, 2020 at 8:35 comment added Emil Jeřábek @MattF. How would that be enough? I can in fact give you all continuous bijections $[0,1]\to[0,1]$, the resulting structure is still superstable. But even in this language, I don’t see how to define a single continuous function $[0,1]^2\to[0,1]$ which takes four different values at the corners.
Nov 7, 2020 at 8:28 comment added user44143 How about adding one unary function for each pair of rationals, $f_{q,r}(x) = \max(0, \min(q + r x, 1))$? That would be enough to define continuous functions that interpolate between the boxes.
Nov 4, 2020 at 13:24 comment added tomasz @EmilJeřábek: Me neither, but it does not seem implausible.
Nov 2, 2020 at 16:23 comment added Emil Jeřábek @tomasz They might be easier to control, but I don’t see how to do continuous approximation of continuous binary functions using just unary affine functions.
Nov 2, 2020 at 16:06 comment added tomasz @EmilJeřábek: That is true, but $+$ is binary. Unary functions are easier to control.
Nov 2, 2020 at 11:36 comment added Emil Jeřábek If you expand the structure with as simple an affine function as $+$ (restricted to $[0,1]$), it is no longer stable, as it can define the usual order on $[0,1]$.
Nov 2, 2020 at 11:32 comment added Johannes Hahn Why can't one require $g$ to be continuous? Shouldn't piecewise affine functions work just as well as piecewise constant functions?
Nov 2, 2020 at 10:40 history answered Emil Jeřábek CC BY-SA 4.0