Counter-streaming beams go turbulent

The lossy problem the solved real-space step was measured on: two components collide and the flow breaks up.

/home/runner/work/NLSE/NLSE/NLSE/solvers/nlse.py:182: UserWarning: The MLX backend is not installed here, so CPU will be used instead — the fastest one available. Install mlx to use MLX, or pass backend='CPU' to silence this.
  self._backend: Backend = get_backend(backend, grid_size=(NX, NY))
/home/runner/work/NLSE/NLSE/NLSE/solvers/nlse.py:380: UserWarning: delta_z=1.50e-04 exceeds split-step accuracy limit (4.98e-05). Reducing to 4.48e-05.
  delta_z = self._capped_delta_z(delta_z, A, method)





Propagation:   0%|          | 0.00/100 [00:00<?, ?%/s]




Propagation: 100%|██████████| 100/100 [01:27<00:00, 1.14%/s]





Time spent to solve : 87.63338106900005 s (CPU)


array([[ 0.27637655-2.9725555e-02j,  1.8697624 -8.7581235e-01j,
         3.4343312 -1.4051603e+00j, ..., -0.87621176+2.7425733e+00j,
        -0.86121935+1.1756768e+00j, -0.64081347+6.1063677e-01j],
       [ 3.0515254 +1.6852412e+00j,  4.6509304 +3.6947879e-01j,
         5.1936407 -1.3372231e+00j, ..., -2.1129737 +5.6220288e+00j,
        -1.0708597 +4.2313528e+00j,  0.76944596+2.9165740e+00j],
       [ 4.231117  +2.8455992e+00j,  4.079919  +3.2829599e+00j,
         3.297777  +2.7439342e+00j, ...,  1.351612  +1.1168299e+00j,
         2.6642046 +2.4315631e+00j,  3.804654  +2.5183561e+00j],
       ...,
       [ 4.1957855 +2.8297093e+00j,  3.8348446 +2.5621474e+00j,
         2.691773  +2.4575994e+00j, ...,  1.5773629 +1.4359707e+00j,
         3.289304  +2.7020240e+00j,  4.041312  +3.2242475e+00j],
       [ 3.1344655 +1.6810782e+00j,  0.8473452 +2.9397113e+00j,
        -1.0274792 +4.2616773e+00j, ...,  5.212818  -3.2864480e+00j,
         5.139399  -1.2709531e+00j,  4.7043586 +3.9898223e-01j],
       [ 0.32870656-3.7216526e-03j, -0.66872203+5.8379656e-01j,
        -0.8776942 +1.1258004e+00j, ...,  4.470187  -1.9946876e+00j,
         3.456886  -1.3825043e+00j,  1.8917439 -8.3645284e-01j]],
      shape=(1024, 1024), dtype=complex64)

import numpy as np
from NLSE import NLSE

N = 1024
n2 = -1.6e-9
window = 8e-3
power = 1.05
Isat = 10e4
L = 20e-2
alpha = 20
waist = 2e-3
waist_d = 1e-3
backend = "MLX"
simu = NLSE(alpha, power, window, n2, None, L, NX=N, NY=N, Isat=Isat, backend=backend)
simu.V = 1e-4 * np.exp(-(np.hypot(simu.XX, simu.YY) ** 2) / waist_d**2)
kp = 2 * np.pi * 5e3
E0 = np.exp(-(np.hypot(simu.XX, simu.YY) ** 2) / waist**2).astype(np.complex64)
E0[0 : N // 2, :] *= np.exp(1j * kp * simu.XX[0 : N // 2, :])
E0[N // 2 :, :] *= np.exp(-1j * kp * simu.XX[N // 2 :, :])
simu.out_field(E0, L, plot=True)