`axiswise=True` does not work

HoloViews
1

Hello everyone,

I have two figures side by side. They are two independent figures because they show two different things. I understand that Holoview checks if the axes are the same or not, and accordingly axes will be shared. But in this situation, I dont want them to be shared.

You can see here that y-axis is shared, and second figure is clipping my data.

Screenshot 2020-07-23 at 21.57.51
Screenshot 2020-07-23 at 21.57.512408×796 99.4 KB

I added opts.(axiswise=True) command for both figures but no chance. Firstly, I realized that X axis of both figures have the same name. I modified one of them, so x-axis is not shared anymore. However, I cannot break y-axis sharing, although they do not have the same y-axis name…

Maybe you can help me out :innocent:

This is my code;

import pandas as pd
import numpy as np
import panel as pn
import matplotlib.pyplot as plt
import hvplot.pandas
from matplotlib.figure import Figure
from matplotlib.backends.backend_agg import FigureCanvas
from decimal import Decimal

pn.extension()

%matplotlib inline
%matplotlib notebook

# Global variables
fig_width = 600
fig_height = 300

d = np.genfromtxt('visdata.dat', delimiter='\t')
row_data = pd.DataFrame(data=d[0:,0:])
d1 = row_data.iloc[1:,0:] # we use this data frame for plotting time traces.

def plot_multitimtrace(pixels = ["200","300","400"]):
    fig = d1.hvplot.line(x='0', 
                         y=pixels,
                         xlim=(-3, 20),
                         xlabel='Time (ps)', 
                         ylabel='∆Absorbance (OD)',
                         width = fig_width, 
                         height = fig_height).opts(axiswise=True)
    return pn.Column(fig)

def plot_multitimspec(time = ["2", "30", "50", "70"]):
    x1= row_data.iloc[0,1:] #slicing first row of data as x-axis (which is pixels).       
    
    y1= row_data.iloc[int(time[0]),1:] #slicing time'th' row of data as y-axis (which is defined by time)
    y2= row_data.iloc[int(time[1]),1:]
    y3= row_data.iloc[int(time[2]),1:]
    y4= row_data.iloc[int(time[3]),1:]
    data_to_plot = pd.concat([x1, y1, y2, y3, y4], axis=1)
    mapping = {data_to_plot.columns[0]:'wl'}
    data_to_plot = data_to_plot.rename(columns=mapping)
    fig = data_to_plot.hvplot.line(x='wl',
                                  y=time,
                                  xlabel='Wavelength (nm)', 
                                  ylabel='∆Absorbance (OD)',
                                  width = fig_width, 
                                  height = fig_height).opts(axiswise=True)
    return pn.Column(fig)

graph3 = pn.Column("<br>\n#Plot Multi Time-Traces\n.", pn.interact(plot_multitimtrace))
graph4 = pn.Column("<br>\n#Plot Time-Spectra\n.",pn.interact(plot_multitimspec))

p = pn.Row(graph3, graph4)
p
1 Like
2

Instead of returning pn.Column from your functions I’d do this:

return pn.pane.HoloViews(fig, linked_axes=False)

Returning a Column is (slightly) inefficient since it adds an additional layer of nesting and internally just renders with a pn.pane.HoloViews anyway. The linked_axes=False options will ensure the axes aren’t shared.

2 Likes
3

Hi Philip,

Thanks! That worked! :slightly_smiling_face: I want to understand those concepts (layers, panels) better. I am reading the documentation of panel and holoview but I feel sometimes lost. Some stuff are still a little bit abstract to me… :sweat:

1 Like
4

Totally understandable. It’s also the case that axiswise should have worked here. Which version of Panel are you using?

5

Panel version is 0.8.3.

6

Oh, right, latest version is 0.9.7, I’d urge you to upgrade.