The codes are implemented in tensorflow--1.14 or 1.15 under the interpreter python3.6 or python3.7. Additionally, if the codes are runned on a Server, one should use the miniconda3 for python 3.7 or 3.6. However, if you dowmload the latest version of miniconda3 from https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh, you will get a miniconda3 based on python 3.8. Hence, you should redirect to the https://docs.conda.io/en/latest/miniconda.html, then download the miniconda3 based on python3.7.
The original codes are found in https://github.com/Blue-Giant/MscaleDNN_tf1
We have also submitted the new codes for solving multi-scale PDEs on Tensorflow 2.X, one can refer to https://github.com/Blue-Giant/MscaleDNN-tf2
NumPy 矩阵库(Matrix) https://www.runoob.com/numpy/numpy-matrix.html
Numpy中矩阵对象（matrix） https://blog.csdn.net/qq403977698/article/details/47254539
Numpy库中矩阵相关操作(矩阵创建，截取，求和，方差，最大值) https://blog.csdn.net/m0_54361811/article/details/124156665
numpy教程：矩阵matrix及其运算 https://blog.csdn.net/pipisorry/article/details/48791403
Numpy（二）np.matrix概述 https://blog.csdn.net/weixin_39258979/article/details/109358408
Tensorflow之矩阵基础 https://blog.csdn.net/z_xindong/article/details/90678904
tensorflow中去上下三角矩阵：matrix_band_part 和tf.linalg.band_part https://blog.csdn.net/ACM_hades/article/details/88790013