solprop.backends.numba_cpu.tables#
Evaluate the reference z-table representation into persistent host stage arrays.
Classes
|
Hold a Chebyshev table's Clenshaw work and stage output on the host. |
- class solprop.backends.numba_cpu.tables.HostTable(table)[source]#
Bases:
objectHold a Chebyshev table’s Clenshaw work and stage output on the host.
The operation order follows NumPy’s
chebval. Only this adapter handles the general z table; the existing constant and expansion stage routes are unchanged.- Parameters:
- tablePiecewiseChebyshev
Array-valued complex table sampled in translation.
- Attributes:
- tablePiecewiseChebyshev
The immutable translated coefficients.
- worknumpy.ndarray
Three Clenshaw arrays, shape
(3, *value_shape), complex128.- valuenumpy.ndarray
Evaluation result, shape value_shape, complex128.
- leftnumpy.ndarray
Saved left endpoint for a difference, shape value_shape, complex128.
- at(z)[source]#
Evaluate the table at one position into its reusable result array.
- Parameters:
- zfloat
Position in m, within the sampled span.
- Returns:
- numpy.ndarray
The result, overwritten by the next call.
- delta(z0, z1)[source]#
Subtract two endpoint values into the reusable result array.
- Parameters:
- z0, z1float
Endpoint positions in m, within the sampled span.
- Returns:
- numpy.ndarray
Table(z1) - table(z0), overwritten by the next call.
- property nbytes: int[source]#
Return the bytes of owned work arrays.
- Returns:
- int
Resident scratch and results in bytes, excluding immutable coefficients.
- table: PiecewiseChebyshev#