Re: [Matplotlib-users] symlog linear to log transition plot position?

[Benjamin R. <ben...@ou...>]

On Wed, May 2, 2012 at 3:59 PM, David Erickson <der...@st...>wrote:

>  On 5/1/2012 5:41 AM, Benjamin Root wrote:
>
>
>
> On Tue, May 1, 2012 at 6:42 AM, David Erickson <der...@st...>wrote:
>
>> Hi I need to use the symlog yscale in my graph, I have a lot of data
>> that needs to be displayed linearly, with a small fraction at the upper
>> end of the range that needs to be displayed in log scale due to its
>> distance from the main data.  The symlog scale works great, however I'd
>> like to adjust the actual vertical graph position where it crosses from
>> linear to log (not the threshold), because right now only ~25% of the y
>> space is being given to linear, and I'd like it to be more like 80%.  Is
>> this possible?
>>
>> Thanks!
>> David
>>
>>
> Isn't linthreshy what you are looking for?  It denotes the range where the
> scale is linear.  So, if it is 25, then from -25 to 25 the scale will be
> linear.  After 25, it will be log.
>
> Maybe I am missing something in your description?
>
> Cheers!
> Ben Root
>
>
> Hi Ben,
> No unfortunately linthreshy only controls the crossover point from linear
> to log scale, it does not give you control over where this occurs on the
> figure's y coordinates.  I've attached a picture to explain, currently the
> linear part of the graph is only being given around 25% of the vertical
> space, I'd like to reverse that and give linear ~75% and log at the top
> only the remaining 25%.  Is this possible?  I've been digging around in the
> SymmetricalLogScale and SymmetricalLogLocator classes and can't even tell
> how this range is allocated.
>
> Thanks in advance!!
> -David
>

David,

Thanks, that is much clearer what you are looking for.  You are right, I
can't seem to find any sort of obvious way to get what you want.  Reading
the docs for the SymLog scale indicates to me that the author intended for
the logrithmic portion to be most interesting and the linear portion was
only supposed to be a work-around the whole log(0) issue.  Could you file
an issue on the github page so that we can mark it as a feature request?

Thanks,
Ben Root