rmwoods

import pandas as pd
import numpy as np

df1 = pd.DataFrame([[1,np.nan]], index=[1], columns=["a","b"])
df2 = pd.DataFrame([[np.nan, 2]], ndex=[1], columns=["a","b"]) 
df3 = df1.append(df2)
#     a    b
# 1  1.0  NaN
# 1  NaN  2.0

rmwoods

a = np.arange(20).reshape(4,5)
#array([[ 0,  1,  2,  3,  4],
#       [ 5,  6,  7,  8,  9],
#       [10, 11, 12, 13, 14],
#       [15, 16, 17, 18, 19]])
b = np.repeat(a,2).reshape(*a.shape,2)
#array([[[ 0,  0],
#        [ 1,  1],
#        [ 2,  2],
#        [ 3,  3],