pybird.utils module

pybird.utils.diff_all(f, x, max_order=4, epsilon=0.002)

Computes all derivatives of a function f: ℝⁿ → ℝᵐ at point x up to max_order using central finite differences with memoized evaluation reuse.

Parameters:

• f – callable, ℝⁿ → ℝᵐ

• x – 1D array of shape (n,)

• max_order – int, in [0, 4]; highest derivative order to compute

• epsilon – scalar or array of shape (n,); finite difference step size

Returns:

[0] → f(x) of shape (m,)

[1] → Jacobian of shape (m, n) [2] → Hessian of shape (m, n, n) [3] → 3rd-order tensor of shape (m, n, n, n) [4] → 4th-order tensor of shape (m, n, n, n, n)

For example, if max_order = 2, the return value is:

[f(x), Jacobian, Hessian]

Return type:

List of derivatives up to max_order, where each element is

pybird.utils.get_data_path()

Get the path to the emulator data directory, works for both development and PyPI installation.

Returns:

Path to the emulator data directory

Return type:

Path

pybird.utils.reshape_bird(array, L)

Reshape 1D concatenated array back to the original arrays (P11l, Pctl, Ploopl).

Parameters:

• array (-) – The input array (1D for original, higher dimensions for derivatives)

• L (-) – The object that contains correlator_sky objects and original shapes

Returns:

• A list of reshaped tuples (P11l, Pctl, Ploopl) for each correlator_sky