How can I accomplish `set_xlim` or `set_ylim` in Bokeh?

Question

I create a figure in a function, e.g.

import numpy
from bokeh.plotting import figure, show, output_notebook
output_notebook()

def make_fig():
    rows = cols = 16
    img = numpy.ones((rows, cols), dtype=numpy.uint32)
    view = img.view(dtype=numpy.uint8).reshape((rows, cols, 4))
    view[:, :, 0] = numpy.arange(256)
    view[:, :, 1] = 265 - numpy.arange(256)
    fig = figure(x_range=[0, c], y_range=[0, rows])
    fig.image_rgba(image=[img], x=[0], y=[0], dw=[cols], dh=[rows])
    return fig

Later I want to zoom in on the figure:

fig = make_fig()
# <- zoom in on plot, like `set_xlim` from matplotlib
show(fig)

How can I do programmatic zoom in bokeh?

Solution 1

One way is to can things with a simple tuple when creating a figure:

figure(..., x_range=(left, right), y_range=(bottom, top))

But you can also set the x_range and y_range properties of a created figure directly. (I had been looking for something like set_xlim or set_ylim from matplotlib.)

from bokeh.models import Range1d

fig = make_fig()
left, right, bottom, top = 3, 9, 4, 10
fig.x_range=Range1d(left, right)
fig.y_range=Range1d(bottom, top)
show(fig)

Solution 2

As of Bokeh 2.X, it seems it is not possible to replace figure.{x,y}_range with a new instance of Range1d from DataRange1d or vice versa.

Instead one has to set figure.x_range.start and figure.x_range.end for a dynamic update.

See https://github.com/bokeh/bokeh/issues/8421 for further details on this issue.

Solution 3

Maybe a naive solution, but why not passing the lim axis as argument of your function?

import numpy
from bokeh.plotting import figure, show, output_notebook
output_notebook()

def make_fig(rows=16, cols=16,x_range=[0, 16], y_range=[0, 16], plot_width=500, plot_height=500):
    img = numpy.ones((rows, cols), dtype=numpy.uint32)
    view = img.view(dtype=numpy.uint8).reshape((rows, cols, 4))
    view[:, :, 0] = numpy.arange(256)
    view[:, :, 1] = 265 - numpy.arange(256)
    fig = figure(x_range=x_range, y_range=y_range, plot_width=plot_width, plot_height=plot_height)
    fig.image_rgba(image=[img], x=[0], y=[0], dw=[cols], dh=[rows])
    return fig

Comments

[asofyan]

How to implement this on y_axis_type = 'log'? It doesn't work with that bottom and top range.

[tommy.carstensen]

Doesn't work for me. I had to do p.x_range = Range1d(x_min, x_max)

[tommy.carstensen]

Doesn't work for me. I had to do fig.x_range = Range1d(x_min, x_max)

[bigreddot]

set was really an implementation detail, and it has since been removed. I have updated the answer.

[Marc Compere]

bingo - thank you - works like a champ in 1.4.0 and Bokeh 2