transformation formulas give incorrect values for exponents

[matthewParksViome]

I noticed that the values calculated by the trans formula for sec_axis are incorrect.

In particular, I want to give the log values and have a sec_axis that plots both the log and non-log values.

It seems to be a problem when the trans formula has an exponential form other than exp.

Minimal reproducible example:

df=data.frame(x=seq(from=0,to=-0.95,by=-0.05) , y=rnorm(20))
# just some numbers between [-1,0] on x axis. 
# They are intended to represent logarithmic values, and the the goal
# is to plot the non-logarithmic values as a second y-axis.
# Note the simple change of base to log10

ggplot(data=df , aes( x = x/log(10) , y = y ) ) +
 geom_point() + 
scale_x_continuous(sec.axis = sec_axis(~10^(.)) )

Since all x values are in [0,1], then by definition, the sec_axis should not give any values above 1. But it does. Note how the rightmost data point on the x-axis has log value = 0, but on the secondary axis it has value > 1.0. This is, of course, incorrect since a^0 = 1 for all numbers a.

[batpigandme]

I think this is the same bug as #2978 Ticks misaligned for sec_axis with some scale transformations and data

[matthewParksViome]

@batpigandme , so it is still unresolved? i.e. the bug remains in the latest version of ggplot 3.1.0 ?

[batpigandme]

Correct. When it's resolved, the issue will be closed.

[dpseidel]

If you install PR #3040's branch, this has been patched there. We hope to merge it soon.

[batpigandme]

Great. Since This is a dupe for #2978, I'm going to close this as a duplicate issue.

[lock]

This old issue has been automatically locked. If you believe you have found a related problem, please file a new issue (with reprex) and link to this issue. https://reprex.tidyverse.org/