jax.scipy.linalg.expm#
- jax.scipy.linalg.expm(A, *, upper_triangular=False, max_squarings=16)[source]#
Compute the matrix exponential
JAX implementation of
scipy.linalg.expm()
.- Parameters:
- Returns:
An array of shape
(..., N, N)
containing the matrix exponent ofA
.- Return type:
Notes
This uses the scaling-and-squaring approximation method, with computational complexity controlled by the optional
max_squarings
argument. Theoretically, the number of required squarings ismax(0, ceil(log2(norm(A))) - c)
wherenorm(A)
is the L1 norm andc=2.42
for float64/complex128, orc=1.97
for float32/complex64.See also