linear -> spline interp for waveform data

ML4GW · Nov 5, 2024 · b9f6863 · b9f6863
1 parent 032639d
commit b9f6863
Show file tree

Hide file tree

Showing 2 changed files with 12 additions and 9 deletions.
diff --git a/ml4gw/waveforms/phenom_d.py b/ml4gw/waveforms/phenom_d.py
@@ -2,6 +2,7 @@
 from jaxtyping import Float
 
 from ml4gw.constants import MTSUN_SI, PI
+from ml4gw.transforms import SplineInterpolate
 from ml4gw.types import BatchTensor, FrequencySeries1d
 
 from .phenom_d_data import QNMData_a, QNMData_fdamp, QNMData_fring
@@ -583,7 +584,9 @@ def fring_fdamp(self, eta, eta2, chi1, chi2):
         finspin = self.FinalSpin0815(eta, eta2, chi1, chi2)
         Erad = self.PhenomInternal_EradRational0815(eta, eta2, chi1, chi2)
 
-        fRD, fDM = self._linear_interp_finspin(finspin)
+        spline_interpolant = SplineInterpolate(self.qnmdata_a, kx=3, sx=1e-8)
+        fRD = spline_interpolant(self.qnmdata_fring, x_out=finspin).squeeze()
+        fDM = spline_interpolant(self.qnmdata_fdamp, x_out=finspin).squeeze()
         fRD /= 1.0 - Erad
         fDM /= 1.0 - Erad
 

diff --git a/tests/waveforms/test_cbc_waveforms.py b/tests/waveforms/test_cbc_waveforms.py
@@ -225,16 +225,16 @@ def test_phenom_d(
     hc_torch = hc_torch[torch_mask]
 
     assert np.allclose(
-        1e21 * hp_lal_data.real, 1e21 * hp_torch.real.numpy(), atol=2e-4
+        1e21 * hp_lal_data.real, 1e21 * hp_torch.real.numpy(), atol=3e-4
     )
     assert np.allclose(
-        1e21 * hp_lal_data.imag, 1e21 * hp_torch.imag.numpy(), atol=2e-4
+        1e21 * hp_lal_data.imag, 1e21 * hp_torch.imag.numpy(), atol=3e-4
     )
     assert np.allclose(
-        1e21 * hc_lal_data.real, 1e21 * hc_torch.real.numpy(), atol=2e-4
+        1e21 * hc_lal_data.real, 1e21 * hc_torch.real.numpy(), atol=3e-4
     )
     assert np.allclose(
-        1e21 * hc_lal_data.imag, 1e21 * hc_torch.imag.numpy(), atol=2e-4
+        1e21 * hc_lal_data.imag, 1e21 * hc_torch.imag.numpy(), atol=3e-4
     )
 
 
@@ -342,14 +342,14 @@ def test_phenom_p(chirp_mass, mass_ratio, chi1z, chi2z, distance, sample_rate):
     hc_torch = hc_torch[torch_mask]
 
     assert np.allclose(
-        1e21 * hp_lal_data.real, 1e21 * hp_torch.real.numpy(), atol=2e-3
+        1e21 * hp_lal_data.real, 1e21 * hp_torch.real.numpy(), atol=3e-3
     )
     assert np.allclose(
-        1e21 * hp_lal_data.imag, 1e21 * hp_torch.imag.numpy(), atol=2e-3
+        1e21 * hp_lal_data.imag, 1e21 * hp_torch.imag.numpy(), atol=3e-3
     )
     assert np.allclose(
-        1e21 * hc_lal_data.real, 1e21 * hc_torch.real.numpy(), atol=2e-3
+        1e21 * hc_lal_data.real, 1e21 * hc_torch.real.numpy(), atol=3e-3
     )
     assert np.allclose(
-        1e21 * hc_lal_data.imag, 1e21 * hc_torch.imag.numpy(), atol=2e-3
+        1e21 * hc_lal_data.imag, 1e21 * hc_torch.imag.numpy(), atol=3e-3
     )