在使用numpy linalg svd时遇到问题

import numpy as np

# 示例：分块处理大型矩阵
def block_svd(matrix, block_size):
    num_blocks = matrix.shape[0] // block_size
    U_list = []
    S_list = []
    Vt_list = []
    
    for i in range(num_blocks):
        block = matrix[i*block_size:(i+1)*block_size, :]
        U, S, Vt = np.linalg.svd(block)
        U_list.append(U)
        S_list.append(S)
        Vt_list.append(Vt)
    
    # 合并结果
    U = np.concatenate(U_list, axis=0)
    S = np.concatenate(S_list, axis=0)
    Vt = np.concatenate(Vt_list, axis=0)
    
    return U, S, Vt

import numpy as np

# 示例：使用更高精度浮点数
matrix = np.array(matrix, dtype=np.float64)
U, S, Vt = np.linalg.svd(matrix)

import numpy as np

# 示例：确保矩阵形状正确
if matrix.shape[0] != matrix.shape[1]:
    matrix = np.pad(matrix, ((0, max(matrix.shape) - matrix.shape[0]), (0, max(matrix.shape) - matrix.shape[1])), 'constant')
U, S, Vt = np.linalg.svd(matrix)

import dask.array as da

# 示例：使用Dask进行并行计算
matrix = da.from_array(matrix, chunks=(1000, 1000))
U, S, Vt = da.linalg.svd(matrix)