artursp. Posts - Answer Overflow

artursp.

•Created by artursp. on 2/3/2025 in #questions-issues

go.Scatter selection randomly not triggering

I'm using go.Scatter and its lasso selection to select points and further work with the returned indices. However, this is unreliable. Most of the times the selection just doesn't do anything and I have to repeat it 1-5 times. I think it gets more unreliable the bigger the total number of points is (but doesn't depend on the number of selected points). I found this issue that might be connected, but couldn't translate it into a solution myself: https://github.com/zauberzeug/nicegui/issues/3762 Here is a somewhat minimal example.

1 replies

SSolara

•Created by artursp. on 3/26/2024 in #questions-issues

Using existing plotly widget to update figures

With go.FigureWidget it is possible to update the data/properties of a figure without redrawing it. E.g. I can create a scatter plot with a lot of points:

N = 1000000
fig = go.FigureWidget(
    data=go.Scattergl(
        x = np.random.randn(N),
        y = np.random.randn(N),     
        mode='markers',
        selected=go.scattergl.Selected(marker = {"color":"red", "size":25}),
        selectedpoints=[5,10,25]
    )
)

N = 1000000
fig = go.FigureWidget(
    data=go.Scattergl(
        x = np.random.randn(N),
        y = np.random.randn(N),     
        mode='markers',
        selected=go.scattergl.Selected(marker = {"color":"red", "size":25}),
        selectedpoints=[5,10,25]
    )
)

and then update the highlighted points without redrawing it the whole thing:

scatter = fig.data[0]
scatter.selectedpoints = np.random.randint(0,N,size=50)

scatter = fig.data[0]
scatter.selectedpoints = np.random.randint(0,N,size=50)

The straight forward solara equivalent could look like this:

N = 1000000
selected = solara.reactive([5,10,25])

x = np.random.randn(N)
y = np.random.randn(N)

def rand_sel():
    selected.set(np.random.randint(0,N,size=50))

@solara.component
def Page():
    fig = go.FigureWidget(
        data=go.Scattergl(
            x = x,
            y = y,     
            mode='markers',
            selected=go.scattergl.Selected(marker = {"color":"red", "size":25}),
            selectedpoints=selected.value
        )
    )
    
    solara.FigurePlotly(fig)
    solara.Button('Random select', on_click=rand_sel)
    
Page()

N = 1000000
selected = solara.reactive([5,10,25])

x = np.random.randn(N)
y = np.random.randn(N)

def rand_sel():
    selected.set(np.random.randint(0,N,size=50))

@solara.component
def Page():
    fig = go.FigureWidget(
        data=go.Scattergl(
            x = x,
            y = y,     
            mode='markers',
            selected=go.scattergl.Selected(marker = {"color":"red", "size":25}),
            selectedpoints=selected.value
        )
    )
    
    solara.FigurePlotly(fig)
    solara.Button('Random select', on_click=rand_sel)
    
Page()

but this will generally run slower as the figure is redrawn IIUC. Is there a way to update the figure using the plotly widget within solara?

3 replies

SSolara

•Created by artursp. on 1/3/2024 in #questions-issues

Keep zoom state of plotly figure when re-render is triggered.

I built an app where I have a coloured scatter plot with many thousand points each associated with an image from a set of classes. I can select points, look at the image and reclassify. When I want to re render my scatter plot so that it has the updated classes I change a reactive dataframe. This all works fine, the only issue is that I have to zoom into a region of the scatter plot to resolve the points. Re-rendering resets the axes so I have to find that region again (but I have to re-render regularly to keep track of which points I processed). So the question is if there is a way to update the scatter plot (i.e. the point colors) without resetting the axes.

4 replies

SSolara

•Created by artursp. on 12/21/2023 in #questions-issues

Remove controls from plotly figure

I'm using plotly to create clickable images that are tightly arranged. For this I need to remove the plotly controls. This can be done by passing config={'displayModeBar': False} to plotly.show, but as this is not a figure property I'm not sure how to accomplish this when doing solara.FigurePlotly(fig).

9 replies

Gaming

Programming