Moving to new jaxdf (#220)

* removed missing function from test_geometry
* added setting class for time domain simulations (TimeWavePropagationSettings )
* upgraded jaxdf
* use jaxdf.Module

CHANGELOG.md

@@ -7,6 +7,16 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 ### Fixed
 - Fixed arguments error in helmholtz notebook
 
+### Changed
+- `Medium` objects are now `jaxdf.Module`s, which is based on `equinox` modules. It is also a [parametric module for dispatching operators](https://beartype.github.io/plum/parametric.html), meaning that there's a type difference betwee `Medium[FourierSeries]` and `Medium[FiniteDifferences]`, for example.
+- The settings of time domain acoustic simulations are now set using a `TimeWavePropagationSettings`. This also includes an attribute to explicity set the reference sound speed.
+
+### Added
+- Added a logger in `jwave.logger`
+
+### Removed
+- Removed `pressure_from_density` from `jwave.acoustics.conversion`, as it was a duplicate
+
 ## [0.1.5] - 2023-09-27
 ### Added
 - Added `numbers_with_smallest_primes` utility to find grids with small primes for efficient FFT when using FourierSeries

jwave/acoustics/conversion.py

@@ -16,29 +16,6 @@
 import numpy as np
 from jax import numpy as jnp
 
-from jwave.geometry import Sensors
-
-
-def pressure_from_density(sensors_data: jnp.ndarray, sound_speed: jnp.ndarray,
-                          sensors: Sensors) -> jnp.ndarray:
-    r"""
-    Calculate pressure from acoustic density given by the raw output of the
-    timestepping scheme.
-
-    Args:
-        sensors_data: Raw output of the timestepping scheme.
-        sound_speed: Sound speed of the medium.
-        sensors: Sensors object.
-
-    Returns:
-        jnp.ndarray: Pressure time traces at sensor locations
-    """
-    if sensors is None:
-        return jnp.sum(sensors_data[1], -1) * (sound_speed**2)
-    else:
-        return jnp.sum(sensors_data[1], -1) * (sound_speed[sensors.positions]**
-                                               2)
-
 
 def db2neper(
     alpha: jnp.ndarray,

jwave/acoustics/time_harmonic.py

@@ -185,9 +185,9 @@ def _cbs_norm_units(medium, omega, k0, src):
     # Update fields
     src = FourierSeries(src.on_grid, domain)
     if issubclass(type(medium.sound_speed), FourierSeries):
-        medium.sound_speed = FourierSeries(c, domain)
-    else:
-        medium.sound_speed = c
+        c = FourierSeries(c, domain)
+
+    medium = medium.replace("sound_speed", c)
 
     # Update k0
     k0 = k0 * _conversion["dx"]