# ErrorListLogPlot gives wrong error bars on log scale

From this and this post I learned how to use ErrorListLogPlot and its family to add error bars on top of a list of mean values. However, consider this simple example:

Needs["ErrorBarLogPlots"]

ErrorListLogPlot[{{1, 0.6}, {2, 0.5}, {3, 1.5}}, Joined -> True,
PlotRange -> {{0, 4}, {0.1, 5}},
GridLines -> {{}, {0.4, 1.5, 1.6, 2.5, 4.5}}]


where {1, 2, 3} are mean values and {0.6, 0.5, 1.5} are the associated standard errors. I added GridLines to indicate the positions where the end points of error bars are supposed to be located. It is clear from the plot above that the gridlines and end points do not intersect (on Mathematica 9), so the error bars must be wrong. I wonder whether something can be fixed in the ErrorBarLogPlots package. Note that the following examples, including built-in ErrorListPlot, give correct error bars (either on linear or log scale):

ErrorListPlot[{{1, 0.6}, {2, 0.5}, {3, 1.5}}, Joined -> True,
PlotRange -> {{0, 4}, {0.1, 5}},
GridLines -> {{}, {0.4, 1.5, 1.6, 2.5, 4.5}}]

ListPlot[{{1, 2, 3}, {0.4, 1.5, 1.5}, {1.6, 2.5, 4.5}},
Joined -> {False, True, True}, PlotRange -> {{0, 4}, {0.1, 5}},
GridLines -> {{}, {0.4, 1.5, 1.6, 2.5, 4.5}},
Filling -> {2 -> {3}}]

ListLogPlot[{{1, 2, 3}, {0.4, 1.5, 1.5}, {1.6, 2.5, 4.5}},
Joined -> {False, True, True}, PlotRange -> {{0, 4}, {0.1, 5}},
GridLines -> {{}, {0.4, 1.5, 1.6, 2.5, 4.5}},
Filling -> {2 -> {3}}]
`
• The explanation to this problem is summarized here. The thing is basically that error-bars do not transform (when you pass to log-scale) taking logarithm at the end of the bars, relative error is needed instead of the absolute one. (sorry for the bad english) Mar 27, 2018 at 14:40
• Comments are not for extended discussion; this conversation has been moved to chat.
– Kuba
Mar 27, 2018 at 20:09