太极开发者小伙伴，你们好，我目前遇到一个问题，我在求解一个线性方程Ku=F的时候，发现同样的K,F，使用taichi求解器计算出的u一直为0，与使用numpy或者matlab计算出的结果不一致。下面是我的代码和结果：
import taichi as ti
import numpy as np
Knumpy = np.array([
[ 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[ 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[ 0, 0, 519.2, 115.4, -158.7, -108.2, 0, 0, 57.69, 57.69, -129.8, -79.33],
[ 0, 0, 115.4, 519.2, -7.212, -259.6, 0, 0, 57.69, 259.6, -79.33, -129.8],
[ 0, 0, -158.7, -7.212, 259.6, 7.212, 0, 0, -129.8, 21.63, 28.85, -21.63],
[ 0, 0, -108.2, -259.6, 7.212, 259.6, 0, 0, 21.63, -129.8, 79.33, 129.8],
[ 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0],
[ 0, 0, 57.69, 57.69, -129.8, 21.63, 0, 0, 519.2, 115.4, -158.7, -7.212],
[ 0, 0, 57.69, 259.6, 21.63, -129.8, 0, 0, 115.4, 519.2, -108.2, -259.6],
[ 0, 0, -129.8, -79.33, 28.85, 79.33, 0, 0, -158.7, -108.2, 259.6, 108.2],
[ 0, 0, -79.33, -129.8, -21.63, 129.8, 0, 0, -7.212, -259.6, 108.2, 259.6],
])
fnumpy = np.array([
0,
0,
0,
0,
9.375,
0,
0,
0,
0,
0,
9.375,
0
])
np.set_printoptions(precision=3, suppress=True)
np.set_printoptions(formatter={'float': '{: 0.3f}'.format})
print(np.linalg.inv(Knumpy) @ fnumpy)
ti.init(arch=ti.cpu)
f = ti.ndarray(ti.f64, 12)
f[4]=9.375
f[10]=9.375
KBuilder = ti.linalg.SparseMatrixBuilder(12, 12, max_num_triplets=10000)
@ti.kernel
def buildSparseMatrix(K1: ti.types.sparse_matrix_builder(),K2: ti.types.ndarray()):
for i, j in ti.ndrange(K2.shape[0],K2.shape[1]):
K1[i,j] += K2[i,j]
@ti.kernel
def print_arr(arr : ti.types.ndarray(dtype=ti.f64, ndim=1)):
for i in ti.grouped(arr):
print(arr[i])
buildSparseMatrix(KBuilder,Knumpy)
kb = KBuilder.build()
solver = ti.linalg.SparseSolver(solver_type="LDLT")
solver.analyze_pattern(kb)
solver.factorize(kb)
dv = solver.solve(f)
print_arr(dv)
这边的问题是需要显式指定一下