I have a enhancement proposal: Could you please add the possibility to pass absolute values to axhline and axvline?
Background
I found it a bit inconvenient to draw horizontal or vertical lines to mark maxima or minima using the relative values if you are plotting actual data.
Code example
importnumpyasnpimportmatplotlib.pyplotaspltfig, ax=plt.subplots()
x=np.linspace(-1, 1)
defcurve(x):
return-np.random.rand()*(2)*x**2-np.random.rand() # the random numbers have just been introduced to avoid an easy guessing of the numbers in the followingfixed_curve=curve(x)
ax.plot(x, fixed_curve)
ylim=ax.get_ylim()
ytmp= (np.amax(fixed_curve) -ylim[0])/(ylim[1]-ylim[0])
ax.axvline(0, 0, ytmp) # better(?): ax.axvline(0, ylim[0], np.amax(fixed_curve), absolute = True)
Code explanation
In the example given above I want to mark to maximum with the coordinates (x, y) by a vertical line going from the maximum of the curve to the x axis to mark the value on the x axis. However, to exactly hit the maximum I first have to calculate the position of the maximum in respect the limits of the plot. Otherwise the line does either not "reach" the maximum or it "crosses" the maximum what would be counterintuitive.
This could be shortened by a parameter absolute or similar that allows you to draw a axvline or axhline from one absolute coordinate to another absolute coordinate.
Axvline is primarily intended to draw lines spanning the whole vertical range (or a fraction of it). Making it end on a specific data coordinate is not the intention.
@jakobjakobson13 Thanks for the interest in enhancing Matplotlib, but I agree with @timhoffm that this is already trivially easy. axhline is needed as its own method because it puts the line at the x value regardless of the ylimits. I'm going to close this as an active issue, but if you think we have misunderstood, or have more comments please feel free to make them.
@jakobjakobson13 Thanks for the interest in enhancing Matplotlib, but I agree with @timhoffm that this is already trivially easy. axhline is needed as its own method because it puts the line at the x value regardless of the ylimits. I'm going to close this as an active issue, but if you think we have misunderstood, or have more comments please feel free to make them.
Just for reference: If one want a line from a data position down to the x-axis, that's not really possible. One cannot mix coordinate systems here. As a workaround, one can draw the line to large negative values and exclude it from autoscaling by passing in_layout=False by the yet to be implemented method described in #15595.
Activity
timhoffm commented on Nov 1, 2019
You can simply
Axvline is primarily intended to draw lines spanning the whole vertical range (or a fraction of it). Making it end on a specific data coordinate is not the intention.
jklymak commented on Nov 1, 2019
@jakobjakobson13 Thanks for the interest in enhancing Matplotlib, but I agree with @timhoffm that this is already trivially easy. axhline is needed as its own method because it puts the line at the x value regardless of the ylimits. I'm going to close this as an active issue, but if you think we have misunderstood, or have more comments please feel free to make them.
jakobjakobson13 commented on Nov 1, 2019
No thanks, it's fine that way.
timhoffm commented on Nov 1, 2019
Just for reference: If one want a line from a data position down to the x-axis, that's not really possible. One cannot mix coordinate systems here. As a workaround, one can draw the line to large negative values and exclude it from autoscaling by passing in_layout=False by the yet to be implemented method described in #15595.
ImportanceOfBeingErnest commented on Nov 2, 2019
For a line with both ends in different coordinate systems also see ConnectionPatch, examplified in connect_simple01.