Timeline for Interpolation[] gives negative values when all the initial data is positive
Current License: CC BY-SA 4.0
4 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Oct 30, 2019 at 19:25 | comment | added | John Doty |
The problem is intrinsically unsymmetrical: zero is a line that must not be crossed, but there is no upper limit. Looking at the function, the peaks are rounded, but the valleys are nearly flat. The Log method reflects this, while you seem to insist that a proper method should not.
|
|
| Oct 30, 2019 at 12:10 | comment | added | anderstood | @yarchik Of course, but with the same drawback more or less amplified (based on the slope of $g$). Bottom line: what is the missing information that is sought between the known points? If the purpose of interpolating is to take the derivative, that could be a very bad strategy. | |
| Oct 30, 2019 at 10:41 | comment | added | yarchik |
However, the Log-Exp pair can be replaced with any other functional pair such that $f(g(x))=x$ and $f(x)>0$ for domain of interest. For instance g[x]=Sqrt[x] and f[x]=x^2.
|
|
| Oct 30, 2019 at 9:08 | history | answered | anderstood | CC BY-SA 4.0 |