so I am working with relatively large arrays (size (13, 8192)) to plot some Figures on a website. It has already been implemented like this, so changes are difficult to make.
As I am running out of memory using the local storage of the browser, I have to use directly a given complex NumPy array and then split it into the real and imaginary parts in another Callback. The problem is that I can't JSON serialize complex-like arrays. Does someone know what can I do to "save" this sort of array using the dcc.Store component of Dash? Thanks in advance.
Here is an example of the code (it is a really short version of it).
import dash
import dash_core_components as dcc
import dash_html_components as html
from dash.dependencies import Input, Output
import plotly.graph_objects as go
import numpy as np
app = dash.Dash(__name__)
T0 = 1E-12 # duration of input
N = 8192 # number of points
dt = 750*T0/N
T = np.arange(-N/2, N/2)*dt
m = 1
C = 0
def envelopef(T,T0,C,m):
U = (np.exp(-((1 1j*C)/2)*((T/T0)**(2*m)))).astype(complex)
UI = np.absolute(U)**2
return U, UI
z = np.arange(-10,10)
U, UI = envelopef(T,T0,C,m)
scatter1 = go.Scatter(x=T/T0,y=UI)
figure1 = go.Figure(data=[scatter1]).update_layout( )
env_graph = dcc.Graph(id='envelopesss',
animate=True,
figure=figure1.update_layout(width=600, height=600,
xaxis = dict(range = [-8, 8])))
M_slider = dcc.Slider(
id='m_slider',
min=1,
max=10,
step=1,
value=m,
marks={
1: {'label': '1'},
10: {'label': '10'}},
)
app.layout = html.Div([
M_slider,
dcc.Store(id='session', storage_type='local'),
dcc.Loading(id="loading1",children=[html.Div([env_graph]) ],type="circle",),
])
@app.callback(
Output("loading1", "children"),
Output("session", "data"),
[Input("m_slider", "value")])
def update_bar_chart(mn):
U, UI = envelopef(T,T0,C,mn)
phase = np.angle(U)
scatter1 = go.Scatter(x=T/T0,y=UI)
figure1 = go.Figure(data=[scatter1]).update_layout(width=600, height=600,
xaxis = dict(range = [-8, 8]))
data = {'u': U , 'ui':UI, 'up': phase}
env_graph = dcc.Graph(figure=figure1)
return env_graph, data
app.run_server(debug=True)
CodePudding user response:
You might want to take a look at the ServersideOutput
component from dash-extensions
. It keeps the data server side (which should improve the performance of your app), and since the default serializer is pickle, complex-valued arrays work out of the box. You can install it via pip,
pip install dash-extensions==0.0.66
To enable use of serverside outputs, replace the app initialization code by
from dash_extensions.enrich import DashProxy, html, dcc, Input, Output, ServersideOutput, ServersideOutputTransform
app = DashProxy(__name__, transforms=[ServersideOutputTransform()])
Next, replace the Output
by a ServersideOutput
,
@app.callback(
Output("loading1", "children"),
ServersideOutput("session", "data"),
[Input("m_slider", "value")])
That's it. Your app should now work. For completeness, here is the full app code,
import plotly.graph_objects as go
import numpy as np
from dash_extensions.enrich import DashProxy, html, dcc, Input, Output, ServersideOutput, ServersideOutputTransform
app = DashProxy(__name__, transforms=[ServersideOutputTransform()])
T0 = 1E-12 # duration of input
N = 8192 # number of points
dt = 750 * T0 / N
T = np.arange(-N / 2, N / 2) * dt
m = 1
C = 0
def envelopef(T, T0, C, m):
U = (np.exp(-((1 1j * C) / 2) * ((T / T0) ** (2 * m)))).astype(complex)
UI = np.absolute(U) ** 2
return U, UI
z = np.arange(-10, 10)
U, UI = envelopef(T, T0, C, m)
scatter1 = go.Scatter(x=T / T0, y=UI)
figure1 = go.Figure(data=[scatter1]).update_layout()
env_graph = dcc.Graph(id='envelopesss',
animate=True,
figure=figure1.update_layout(width=600, height=600,
xaxis=dict(range=[-8, 8])))
M_slider = dcc.Slider(
id='m_slider',
min=1,
max=10,
step=1,
value=m,
marks={
1: {'label': '1'},
10: {'label': '10'}},
)
app.layout = html.Div([
M_slider,
dcc.Store(id='session', storage_type='local'),
dcc.Loading(id="loading1", children=[html.Div([env_graph])], type="circle", ),
])
@app.callback(
Output("loading1", "children"),
ServersideOutput("session", "data"),
[Input("m_slider", "value")])
def update_bar_chart(mn):
U, UI = envelopef(T, T0, C, mn)
phase = np.angle(U)
scatter1 = go.Scatter(x=T / T0, y=UI)
figure1 = go.Figure(data=[scatter1]).update_layout(width=600, height=600,
xaxis=dict(range=[-8, 8]))
data = {'u': U, 'ui': UI, 'up': phase}
env_graph = dcc.Graph(figure=figure1)
return env_graph, data
app.run_server(port=7777)