from __future__ import annotations
import numpy as np
from matplotlib import pyplot as plt
from multinterp.unstructured._gpytorch import GaussianProcessRegression
%matplotlib inline
%load_ext autoreload
%autoreload 2def function_1(x, y):
return x * (1 - x) * np.cos(4 * np.pi * x) * np.sin(4 * np.pi * y**2) ** 2rng = np.random.default_rng(0)
rand_x, rand_y = rng.random((2, 1000))
values = function_1(rand_x, rand_y)grid_x, grid_y = np.meshgrid(
np.linspace(0, 1, 100), np.linspace(0, 1, 100), indexing="ij"
)interp = GaussianProcessRegression(
values.astype("float32"), (rand_x.astype("float32"), rand_y.astype("float32"))
)
new_grid = interp(grid_x.astype("float32"), grid_y.astype("float32"))C:\Users\alujan\GitHub\alanlujan91\multinterp\src\multinterp\grids.py:133: UserWarning: Creating a tensor from a list of numpy.ndarrays is extremely slow. Please consider converting the list to a single numpy.ndarray with numpy.array() before converting to a tensor. (Triggered internally at C:\cb\pytorch_1000000000000\work\torch\csrc\utils\tensor_new.cpp:248.)
self.grids = BACKEND_MODULES[backend].asarray(grids)
c:\Users\alujan\mambaforge\envs\multinterp-dev\lib\site-packages\torch\nn\parallel\data_parallel.py:32: UserWarning:
There is an imbalance between your GPUs. You may want to exclude GPU 1 which
has less than 75% of the memory or cores of GPU 0. You can do so by setting
the device_ids argument to DataParallel, or by setting the CUDA_VISIBLE_DEVICES
environment variable.
warnings.warn(imbalance_warn.format(device_ids[min_pos], device_ids[max_pos]))
c:\Users\alujan\mambaforge\envs\multinterp-dev\lib\site-packages\torch\cuda\nccl.py:15: UserWarning: PyTorch is not compiled with NCCL support
warnings.warn('PyTorch is not compiled with NCCL support')
Iter 1/50 - Loss: -1.057 lengthscale: 0.704 noise: 0.012
Convergence reached!
C:\Users\alujan\GitHub\alanlujan91\multinterp\src\multinterp\unstructured\_gpytorch.py:177: UserWarning: The use of `x.T` on tensors of dimension other than 2 to reverse their shape is deprecated and it will throw an error in a future release. Consider `x.mT` to transpose batches of matrices or `x.permute(*torch.arange(x.ndim - 1, -1, -1))` to reverse the dimensions of a tensor. (Triggered internally at C:\cb\pytorch_1000000000000\work\aten\src\ATen\native\TensorShape.cpp:3575.)
).T
plt.imshow(function_1(grid_x, grid_y).T, extent=(0, 1, 0, 1), origin="lower")
plt.plot(rand_x, rand_y, "ok", ms=2, label="input points")
plt.title("Original")
plt.legend(loc="lower right")
plt.imshow(new_grid.mean.cpu(), extent=(0, 1, 0, 1), origin="lower")
plt.plot(rand_x, rand_y, "ok", ms=2, label="input points")
plt.title("Original")
plt.legend(loc="lower right")