spherical.grid

Submodule providing the Grid object

Class Grid

Grid(input_array, *args, **kwargs)

Source: spherical/grid/__init__.py

Object to store SWHS values on a grid This class subclasses numpy's ndarray object, so that it should act like a numpy array in many respects, even with functions like np.zeros_like. NOTE: The functions np.copy(grid) and np.array(grid, copy=True) return ndarray objects; they lose information about the SWSH attributes, and have different ufuncs. If you wish to keep this information, use grid.copy(). Also note that pickling works as expected, as do copy.copy and copy.deepcopy. The number of dimensions is arbitrary as long as it is 2 or more, but the grid must be stored in the last 2 axes. Specifically, the grid must have shape (n_theta, n_phi). (See the function spherical.theta_phi for an example of the actual grid locations expected.) For example, a SWSH function of time may be stored as a 3-d array where the first axis represents different times, and the second and third axes represent the function values at each instant of time. This class also does two important things that are unlike numpy arrays: 1) It tracks the spin weight of the function represented by this data. 2) It overrides most of numpy's "universal functions" (ufuncs) to work appropriately for spin-weighted functions. Specifically, these ufuncs are interpreted as acting on the spin-weighted function itself, rather than just the grid values. The returned values are -- where possible -- Grid objects. Most importantly, we have these overriden methods: a) Multiplying two Grid objects will result in a new Grid object that represents the pointwise product of the functions themselves (and will correctly have spin weight given by the sum of the spin weights of the first two functions). Division is only permitted when the divisor has spin weight zero. b) Addition (and subtraction) is permitted for functions of the same spin weight, but it does not make sense to add (or subtract) functions with different spin weights, so any attempt to do so raises a ValueError. Note that a constant is implicitly a function of spin weight zero, and is treated as such. Numerous other ufuncs -- such as log, exp, trigonometric ufuncs, bit-twiddling ufuncs, and so on -- are disabled because they don't result in objects of specific spin weights. It is possible to treat the underlying data of a Grid object grid as an ordinary numpy array by taking grid.view(np.ndarray). However, it is hoped that this class already performs all reasonable operations. If you find a missing feature that requires you to resort to this, please feel free to open an issue in this project's github page to discuss it. Also, be aware that ndarrays also have various built-in methods that cannot be overridden very easily, such as max, copysign, etc. If you try to use -- even indirectly -- those functions that don't have any clear interpretation for spin-weighted functions, things will likely break.

Methods

bar

bar

Source: spherical/grid/algebra.py

Return Grid object corresponding to conjugated function The result has spin weight equal to the negative of this object's spin weight.

imag

imag

Source: spherical/grid/algebra.py

Return imaginary-valued part of function values on the grid This only makes sense when the function has spin weight 0; anything else will raise an error.