Re: Non-linear axis

[Irene Dumkow <irene.dumkow(at)uni-essen.de>]

Martin Schultz wrote:
> 
> Craig Markwardt wrote:
> >
> >
> > Pay no attention to Martin.  :-)   ...
> 
> Oh Craig. Thanks for bringing me back on Earth ;-) Guess, I just got
> carried away by the word non-linear axis ... You only need resampling
> if you want to display an image which was sampled in linear "bins" on
> a non-linear axis.
> 
> Never mind,
> Martin
> 
I have the impression, that the elegant solution does need the
resampling
part. I am not sure if I brought my point across the first time. What I
have
is an image which needs two different y-axes, one is a standard linear
scale, the other is related to the first one as each value is
essentially
a*y1*y1. I have seen the same style of axis-labeling for phase diagramms
with
one x-axis label linearly with values of 1000 K/ T and the second x-axis
labeled with standard Celsius temperatures as most thermometers aren't
labeled in 1000K / T. 
The solution I found now with the hints given here (but which I consider
rather clumsy and almost like cheating):
While doing the contourplot getting the tickvalues with
Y_TICK_GET=y_value
Calculating the corresponding values for my second y-axis (y2_value).
Drawing the second y-axis using the AXIS procedure with positing the
ticks with YTICKV=y_value, labeling the ticks via YTICKNAME=y2_value,
and
using YTICKS to make sure I have the correct number of ticks.
This does not give me very nice numbers and if I want to get that I
supposed
I have to do quite some calculating back and forth. I think what I am
dreaming about is something kind of like the YLOG keyword.

Thanks for the help I have gotten so far

Irene