AcousticLSRTM

class AcousticLSRTM(
    spatial_order=4,
    device="cpu",
    backend="torch",
)

Implementation:

• src/sweep/equations/acoustic_lsrtm.py

Linearized acoustic equation for LSRTM-style modeling with background and scattered wavefields.

Parameters

• spatial_order (int, optional): Finite-difference order. It should be an even number.
• device (device or str, optional): Target device used to place operator kernels and backend-specific tensors.
• backend (str, optional): Numerical backend. If you plan to run with PropCUDA, this should still normally be "torch" rather than "cuda".

Models

• models (list[str]): ["vp", "mp"]

Required models:

• vp: background velocity model
• mp: perturbation or reflectivity-like model term used by the scattered wavefield update

Wavefields

• wavefields (list[str]): ["h1", "h2", "psix", "psiz", "zetax", "zetaz", "sh1", "sh2", "spsix", "spsiz", "szetax", "szetaz"]

This class carries both background and scattered states.

Backend Behavior

• implemented as a second-order equation
• currently does not expose a compiled torch binding hook

Torch Binding

• supports_torch_binding() : False