Error when multiplying a non-square SparseMatrix with an ndarray

[wang-mengdi]

Describe the bug
Cannot multiply a non-square sparse matrix with an ndarray.

To Reproduce

import taichi as ti

ti.init(arch=ti.cpu)

n_rows, n_cols = 4, 3
#n_rows, n_cols = 4, 4 # this works
sparse_matrix_builder = ti.linalg.SparseMatrixBuilder(n_rows, n_cols, max_num_triplets=12)

@ti.kernel
def fill_sparse_matrix(sparse_matrix_builder: ti.types.sparse_matrix_builder()):
    for i in range(n_rows):
        for j in range(n_cols):
            sparse_matrix_builder[i, j] += i
fill_sparse_matrix(sparse_matrix_builder)
sparse_matrix = sparse_matrix_builder.build()

#dense_vector = ti.field(dtype=ti.f32, shape=n_cols) # this works
dense_vector = ti.ndarray(dtype=ti.f32, shape=n_cols)


@ti.kernel
def initialize_vector():
    for i in range(n_cols):
        dense_vector[i] = i * 2.0

dense_vector = sparse_matrix@dense_vector

print("\nResult:")
for i in range(n_rows):
    print("Row {}: {}".format(i, dense_vector[i]))

Log

[Taichi] version 1.7.0, llvm 15.0.1, commit 2fd24490, win, python 3.11.5
[Taichi] Starting on arch=x64
Assertion failed: rows == this->rows() && cols == this->cols() && "DenseBase::resize() does not actually allow to resize.", file C:\Users\buildbot\actions-runner\_work\taichi\taichi\external\eigen\Eigen\src/Core/DenseBase.h, line 257

Additional comments
Currently the followings are OK:

1. Multiply a square sparse matrix with a ti.ndarray (see comment 1)
2. Multiply a non-square sparse matrix with a ti.field (see comment 2)

However, multiplying a non-square matrix with a ti.ndarray will produce errors.