Skip to main content
>>
>>
Version: v1.3.0

Sparse Matrix

Sparse matrices are frequently involved in solving linear systems in science and engineering. Taichi provides useful APIs for sparse matrices on the CPU and CUDA backends.

To use sparse matrices in Taichi programs, follow these three steps:

  1. Create a builder using ti.linalg.SparseMatrixBuilder().
  2. Call ti.kernel to fill the builder with your matrices' data.
  3. Build sparse matrices from the builder.
WARNING

The sparse matrix feature is still under development. There are some limitations:

  • The sparse matrix data type on the CPU backend only supports f32 and f64.
  • The sparse matrix data type on the CUDA backend only supports f32.

Here's an example:

import taichi as ti
arch = ti.cpu # or ti.cuda
ti.init(arch=arch)

n = 4
# step 1: create sparse matrix builder
K = ti.linalg.SparseMatrixBuilder(n, n, max_num_triplets=100)

@ti.kernel
def fill(A: ti.types.sparse_matrix_builder()):
    for i in range(n):
        A[i, i] += 1  # Only +=  and -= operators are supported for now.

# step 2: fill the builder with data.
fill(K)

print(">>>> K.print_triplets()")
K.print_triplets()
# outputs:
# >>>> K.print_triplets()
# n=4, m=4, num_triplets=4 (max=100)(0, 0) val=1.0(1, 1) val=1.0(2, 2) val=1.0(3, 3) val=1.0

# step 3: create a sparse matrix from the builder.
A = K.build()
print(">>>> A = K.build()")
print(A)
# outputs:
# >>>> A = K.build()
# [1, 0, 0, 0]
# [0, 1, 0, 0]
# [0, 0, 1, 0]
# [0, 0, 0, 1]

The basic operations like +, -, *, @ and transpose of sparse matrices are supported now.

print(">>>> Summation: C = A + A")
C = A + A
print(C)
# outputs:
# >>>> Summation: C = A + A
# [2, 0, 0, 0]
# [0, 2, 0, 0]
# [0, 0, 2, 0]
# [0, 0, 0, 2]

print(">>>> Subtraction: D = A - A")
D = A - A
print(D)
# outputs:
# >>>> Subtraction: D = A - A
# [0, 0, 0, 0]
# [0, 0, 0, 0]
# [0, 0, 0, 0]
# [0, 0, 0, 0]

print(">>>> Multiplication with a scalar on the right: E = A * 3.0")
E = A * 3.0
print(E)
# outputs:
# >>>> Multiplication with a scalar on the right: E = A * 3.0
# [3, 0, 0, 0]
# [0, 3, 0, 0]
# [0, 0, 3, 0]
# [0, 0, 0, 3]

print(">>>> Multiplication with a scalar on the left: E = 3.0 * A")
E = 3.0 * A
print(E)
# outputs:
# >>>> Multiplication with a scalar on the left: E = 3.0 * A
# [3, 0, 0, 0]
# [0, 3, 0, 0]
# [0, 0, 3, 0]
# [0, 0, 0, 3]

print(">>>> Transpose: F = A.transpose()")
F = A.transpose()
print(F)
# outputs:
# >>>> Transpose: F = A.transpose()
# [1, 0, 0, 0]
# [0, 1, 0, 0]
# [0, 0, 1, 0]
# [0, 0, 0, 1]

print(">>>> Matrix multiplication: G = E @ A")
G = E @ A
print(G)
# outputs:
# >>>> Matrix multiplication: G = E @ A
# [3, 0, 0, 0]
# [0, 3, 0, 0]
# [0, 0, 3, 0]
# [0, 0, 0, 3]

print(">>>> Element-wise multiplication: H = E * A")
H = E * A
print(H)
# outputs:
# >>>> Element-wise multiplication: H = E * A
# [3, 0, 0, 0]
# [0, 3, 0, 0]
# [0, 0, 3, 0]
# [0, 0, 0, 3]

print(f">>>> Element Access: A[0,0] = {A[0,0]}")
# outputs:
# >>>> Element Access: A[0,0] = 1.0

Sparse linear solver

You may want to solve some linear equations using sparse matrices. Then, the following steps could help:

  1. Create a solver using ti.linalg.SparseSolver(solver_type, ordering). Currently, the factorization types supported on CPU backends are LLT, LDLT, and LU, and supported orderings include AMD and COLAMD. The sparse solver on CUDA supports the LLT factorization type only.
  2. Analyze and factorize the sparse matrix you want to solve using solver.analyze_pattern(sparse_matrix) and solver.factorize(sparse_matrix)
  3. Call x = solver.solve(b), where x is the solution and b is the right-hand side of the linear system. On CPU backends, x and b can be NumPy arrays, Taichi Ndarrays, or Taichi fields. On the CUDA backend, x and b must be Taichi Ndarrays.
  4. Call solver.info() to check if the solving process succeeds.

Here's a full example.

import taichi as ti

arch = ti.cpu # or ti.cuda
ti.init(arch=arch)

n = 4

K = ti.linalg.SparseMatrixBuilder(n, n, max_num_triplets=100)
b = ti.ndarray(ti.f32, shape=n)

@ti.kernel
def fill(A: ti.types.sparse_matrix_builder(), b: ti.template(), interval: ti.i32):
    for i in range(n):
        A[i, i] += 2.0

        if i % interval == 0:
            b[i] += 1.0

fill(K, b, 3)

A = K.build()
print(">>>> Matrix A:")
print(A)
print(">>>> Vector b:")
print(b)
# outputs:
# >>>> Matrix A:
# [2, 0, 0, 0]
# [0, 2, 0, 0]
# [0, 0, 2, 0]
# [0, 0, 0, 2]
# >>>> Vector b:
# [1. 0. 0. 1.]
solver = ti.linalg.SparseSolver(solver_type="LLT")
solver.analyze_pattern(A)
solver.factorize(A)
x = solver.solve(b)
isSuccess = solver.info()
print(">>>> Solve sparse linear systems Ax = b with the solution x:")
print(x)
print(f">>>> Computation was successful?: {isSuccess}")
# outputs:
# >>>> Solve sparse linear systems Ax = b with the solution x:
# [0.5 0.  0.  0.5]
# >>>> Computation was successful?: True

Examples

Please have a look at our two demos for more information:

  • Stable fluid: A 2D fluid simulation using a sparse Laplacian matrix to solve Poisson's pressure equation.
  • Implicit mass spring: A 2D cloth simulation demo using sparse matrices to solve the linear systems.
Was this helpful?