问题

我正在尝试使用plotly来绘制一组点的三维凸包。我使用了Mesh3d对象，但表面没有正确创建（请参见下面的图片）。我该如何修复这个问题？

import plotly.graph_objects as go
import itertools, math, numpy as np

from scipy.spatial import ConvexHull

# This simply creates a set of points:
n = 3
m = 3
E = np.array(list(itertools.product(np.arange(-1, 1.1, .5), repeat=m)))
V = [
 [-2.20676418,  1.53670924, -1.5541674 ],
 [ 0.63437404,  0.07306301,  3.82253086],
 [ 3.19989112,  0.71987311,  2.79373418]
]
x = np.array([np.dot(V, e) for e in E])

# Then I compute the convex hull using scipy:
xc = x[ConvexHull(x).vertices]

fig = go.Figure()
fig.add_trace(go.Mesh3d(x=xc[:, 0], y=xc[:, 1], z=xc[:, 2], color="blue", opacity=.5))
fig

当前输出是:

import plotly.graph_objects as go
import itertools, math, numpy as np

from scipy.spatial import ConvexHull

# This simply creates a set of points:
n = 3
m = 3
E = np.array(list(itertools.product(np.arange(-1, 1.1, .5), repeat=m)))
V = [
 [-2.20676418,  1.53670924, -1.5541674 ],
 [ 0.63437404,  0.07306301,  3.82253086],
 [ 3.19989112,  0.71987311,  2.79373418]
]
x = np.array([np.dot(V, e) for e in E])

# Then I compute the convex hull using scipy:
xc = x[ConvexHull(x).vertices]

fig = go.Figure()
fig.add_trace(go.Mesh3d(x=xc[:, 0], y=xc[:, 1], z=xc[:, 2], color="blue", opacity=.5))
fig

答案1

得分: 2

I think what you miss is the alphahull option.

If you want a convex hull you need to add alphahull=0:

import plotly.graph_objects as go
import itertools, math, numpy as np

from scipy.spatial import ConvexHull

# This simply creates a set of points:
n = 3
m = 3
E = np.array(list(itertools.product(np.arange(-1, 1.1, .5), repeat=m)))
V = [
 [-2.20676418,  1.53670924, -1.5541674 ],
 [ 0.63437404,  0.07306301,  3.82253086],
 [ 3.19989112,  0.71987311,  2.79373418]
]
x = np.array([np.dot(V, e) for e in E])

# Then I compute the convex hull using scipy:
xc = x[ConvexHull(x).vertices]

fig = go.Figure()
fig.add_trace(go.Mesh3d(x=xc[:, 0], 
                        y=xc[:, 1], 
                        z=xc[:, 2], 
                        color="blue", 
                        opacity=.5,
                        alphahull=0))
fig

将给你想要的立方体：

<details>
<summary>英文:</summary>

I think what you miss is the [alphahull](https://plotly.com/python/reference/#mesh3d-alphahull) option.

If you want a convex hull you need to add `alphahull=0`:

import plotly.graph_objects as go
import itertools, math, numpy as np

from scipy.spatial import ConvexHull

This simply creates a set of points:

n = 3
m = 3
E = np.array(list(itertools.product(np.arange(-1, 1.1, .5), repeat=m)))
V = [
[-2.20676418, 1.53670924, -1.5541674 ],
[ 0.63437404, 0.07306301, 3.82253086],
[ 3.19989112, 0.71987311, 2.79373418]
]
x = np.array([np.dot(V, e) for e in E])

Then I compute the convex hull using scipy:

xc = x[ConvexHull(x).vertices]

fig = go.Figure()
fig.add_trace(go.Mesh3d(x=xc[:, 0],
y=xc[:, 1],
z=xc[:, 2],
color="blue",
opacity=.5,
alphahull=0))
fig

will give you the vube you want:

![resutl][1]

[1]: https://i.stack.imgur.com/WuizR.png

</details>