PtyLabX/tests/test_utils.py at main · PtyLab/PtyLabX · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
"""Tests for PtyLabX.utils.utils — FFT, shifts, orthogonalization, geometry."""

import jax.numpy as jnp
import numpy as np
from numpy.testing import assert_allclose

from PtyLabX.utils.utils import circ, fft2c, fraccircshift, ifft2c, orthogonalizeModes, posit, rect


class TestFFT2c:
    """Tests for centered FFT/IFFT."""

    def test_fft2c_ifft2c_roundtrip(self):
        """FFT(IFFT(x)) == x and IFFT(FFT(x)) == x (unitarity)."""
        rng = np.random.default_rng(0)
        E_in = rng.standard_normal((5, 100, 100)) + 1j * rng.standard_normal((5, 100, 100))
        E_in = jnp.array(E_in, dtype=jnp.complex64)
        assert_allclose(np.asarray(ifft2c(fft2c(E_in))), np.asarray(E_in), atol=1e-5)
        assert_allclose(np.asarray(fft2c(ifft2c(E_in))), np.asarray(E_in), atol=1e-5)

    def test_fft2c_with_fftshift_switch(self):
        """Verify fftshiftSwitch=True path also round-trips."""
        rng = np.random.default_rng(1)
        E_in = jnp.array(rng.standard_normal((32, 32)) + 1j * rng.standard_normal((32, 32)), dtype=jnp.complex64)
        result = ifft2c(fft2c(E_in, fftshiftSwitch=True), fftshiftSwitch=True)
        assert_allclose(np.asarray(result), np.asarray(E_in), atol=1e-5)

    def test_fft2c_energy_preservation(self):
        """Parseval's theorem: sum(|x|^2) == sum(|FFT(x)|^2) for unitary FFT."""
        rng = np.random.default_rng(2)
        E_in = jnp.array(rng.standard_normal((64, 64)) + 1j * rng.standard_normal((64, 64)), dtype=jnp.complex64)
        energy_in = float(jnp.sum(jnp.abs(E_in) ** 2))
        energy_out = float(jnp.sum(jnp.abs(fft2c(E_in)) ** 2))
        assert_allclose(energy_in, energy_out, rtol=1e-4)

    def test_fft2c_jit_compiled(self):
        """Verify fft2c runs under JIT (it's decorated with @jax.jit)."""
        E_in = jnp.ones((16, 16), dtype=jnp.complex64)
        result = fft2c(E_in)
        assert result.shape == (16, 16)


class TestFraccircshift:
    """Tests for fractional circular shift."""

    def test_integer_shift_matches_roll(self):
        """Integer shifts should match jnp.roll exactly."""
        rng = np.random.default_rng(3)
        A = jnp.array(rng.standard_normal((32, 32)), dtype=jnp.float32)
        shift = jnp.array([3.0, -2.0])
        result = fraccircshift(A, shift)
        expected = jnp.roll(jnp.roll(A, 3, axis=0), -2, axis=1)
        assert_allclose(np.asarray(result), np.asarray(expected), atol=1e-5)

    def test_zero_shift_identity(self):
        """Zero shift should return the same array."""
        rng = np.random.default_rng(4)
        A = jnp.array(rng.standard_normal((16, 16)), dtype=jnp.float32)
        result = fraccircshift(A, jnp.array([0.0, 0.0]))
        assert_allclose(np.asarray(result), np.asarray(A), atol=1e-6)

    def test_subpixel_shift_interpolation(self):
        """Half-pixel shift should average adjacent pixels."""
        A = jnp.array([[1.0, 0.0, 0.0, 0.0], [0.0, 0.0, 0.0, 0.0], [0.0, 0.0, 0.0, 0.0], [0.0, 0.0, 0.0, 0.0]])
        result = fraccircshift(A, jnp.array([0.5, 0.0]))
        # Half shift along axis 0: 0.5 * A[0,:] + 0.5 * A[1,:]
        assert float(result[0, 0]) > 0
        assert float(result[1, 0]) > 0


class TestOrthogonalizeModes:
    """Tests for SVD-based mode orthogonalization."""

    def test_svd_roundtrip(self):
        """Orthogonalized modes should be orthogonal."""
        rng = np.random.default_rng(5)
        p = jnp.array(rng.standard_normal((3, 32, 32)) + 1j * rng.standard_normal((3, 32, 32)), dtype=jnp.complex64)
        p_orth, eigenvalues, _ = orthogonalizeModes(p)
        assert p_orth.shape == p.shape
        assert eigenvalues.shape == (3,)
        # Eigenvalues should sum to 1 (normalized)
        assert_allclose(float(jnp.sum(eigenvalues)), 1.0, atol=1e-5)

    def test_snapshots_method(self):
        """Test the snapShots method path."""
        rng = np.random.default_rng(6)
        p = jnp.array(rng.standard_normal((3, 32, 32)) + 1j * rng.standard_normal((3, 32, 32)), dtype=jnp.complex64)
        p_orth, eigenvalues, V = orthogonalizeModes(p, method="snapShots")
        assert p_orth.shape == p.shape
        assert_allclose(float(jnp.sum(eigenvalues)), 1.0, atol=1e-5)


class TestCirc:
    """Tests for circular mask generation."""

    def test_basic_shape(self):
        """Circ should produce a boolean mask of the correct shape."""
        x, y = jnp.meshgrid(jnp.linspace(-1, 1, 64), jnp.linspace(-1, 1, 64))
        mask = circ(x, y, 1.0)
        assert mask.shape == (64, 64)
        assert mask.dtype == jnp.bool_

    def test_circ_center_is_true(self):
        """Center pixel of circ should be True for nonzero diameter."""
        x, y = jnp.meshgrid(jnp.linspace(-1, 1, 65), jnp.linspace(-1, 1, 65))
        mask = circ(x, y, 1.0)
        assert bool(mask[32, 32])


class TestRect:
    """Tests for rect function."""

    def test_basic(self):
        x = jnp.array([-1.0, -0.3, 0.0, 0.3, 1.0])
        result = rect(x)
        expected = jnp.array([False, True, True, True, False])
        assert jnp.array_equal(result, expected)


class TestPosit:
    """Tests for posit (ReLU-like) function."""

    def test_positive_values_unchanged(self):
        x = jnp.array([1.0, 2.0, 3.0])
        assert_allclose(np.asarray(posit(x)), np.asarray(x), atol=1e-6)

    def test_negative_values_zero(self):
        x = jnp.array([-1.0, -2.0, -3.0])
        assert_allclose(np.asarray(posit(x)), np.zeros(3), atol=1e-6)