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
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
#[cfg(feature = "memory_efficient")]
mod tests {
use approx::assert_relative_eq;
use scirs2_core::array::{mask_array, masked_invalid};
use scirs2_core::memory_efficient::{
chunk_wise_op, create_disk_array, diagonal_view, evaluate, transpose_view,
ChunkingStrategy, LazyArray,
};
use scirs2_core::ndarray::Array2;
use tempfile::NamedTempFile;
#[test]
fn test_memory_efficient_integration() {
// Create test data
let data =
Array2::from_shape_fn((10, 10), |(i, j)| if i == j { (i + 1) as f64 } else { 0.0 });
// 1. Use chunk-wise operation to double all values
let doubled = chunk_wise_op(
&data,
|chunk| chunk.map(|&x| x * 2.0),
ChunkingStrategy::Fixed(5),
)
.expect("Test: operation failed");
// 2. Create a lazy array from the result
let lazy = LazyArray::new(doubled.clone());
// 3. Define a lazy operation (identity in this case)
let lazy_result = lazy.map(|&x| x);
// 4. Evaluate the lazy operation
let evaluated = evaluate(&lazy_result).expect("Test: operation failed");
// 5. Create a transpose view
let transposed = transpose_view(&evaluated).expect("Test: operation failed");
// 6. Get the diagonal view (should contain the doubled diagonal values)
let diagonal = diagonal_view(&evaluated).expect("Test: operation failed");
// 7. Create a masked array to mask out zeros
let mask = Array2::from_shape_fn(evaluated.raw_dim(), |(i, j)| evaluated[[i, j]] == 0.0);
let masked =
mask_array(evaluated.clone(), Some(mask), Some(0.0)).expect("Test: operation failed");
// 8. Store the result in a disk-backed array
let temp_file = NamedTempFile::new().expect("Test: operation failed");
let disk_array = create_disk_array(
&masked.data,
temp_file.path(),
ChunkingStrategy::Fixed(5),
false,
)
.expect("Test: operation failed");
// 9. Load back from disk
let loaded = disk_array.load().expect("Test: operation failed");
// Verify the results
// The doubled diagonal should have values 2, 4, 6, ..., 20
for i in 0..10 {
assert_relative_eq!(diagonal[i], 2.0 * (i + 1) as f64, epsilon = 1e-10);
}
// The transposed array should be the same (since the matrix is symmetric)
for i in 0..10 {
for j in 0..10 {
assert_relative_eq!(transposed[[i, j]], evaluated[[j, i]], epsilon = 1e-10);
}
}
// The masked array should have zeros everywhere except the diagonal
for i in 0..10 {
for j in 0..10 {
if i == j {
assert_relative_eq!(masked.data[[i, j]], 2.0 * (i + 1) as f64, epsilon = 1e-10);
assert!(!masked.mask[[i, j]]);
} else {
assert!(masked.mask[[i, j]]);
}
}
}
// The loaded array should match the original data
for i in 0..10 {
for j in 0..10 {
if i == j {
assert_relative_eq!(loaded[[i, j]], 2.0 * (i + 1) as f64, epsilon = 1e-10);
} else {
assert_relative_eq!(loaded[[i, j]], 0.0, epsilon = 1e-10);
}
}
}
}
#[test]
fn test_chunked_lazy_disk_workflow() {
// Create a larger test matrix
let n = 20;
let data =
Array2::from_shape_fn((n, n), |(i, j)| if i == j { (i + 1) as f64 } else { 0.0 });
// Workflow test:
// 1. Process in chunks
// 2. Store as lazy computation
// 3. Save to disk
// 4. Load back and verify
// Process in chunks - add 10 to all elements
let added = chunk_wise_op(
&data,
|chunk| chunk.map(|&x| x + 10.0),
ChunkingStrategy::Fixed(5),
)
.expect("Test: operation failed");
// Create a lazy array for another operation - multiply by 2
let lazy = LazyArray::new(added);
let lazy_doubled = lazy.map(|&x| x * 2.0);
// Create a temporary file
let temp_file = NamedTempFile::new().expect("Test: operation failed");
// Store the result in a disk-backed array
// In a real implementation, we would be able to evaluate directly to disk,
// but in our placeholder we need to evaluate first
let evaluated = evaluate(&lazy_doubled).expect("Test: operation failed");
let disk_array = create_disk_array(
&evaluated,
temp_file.path(),
ChunkingStrategy::Fixed(5),
false,
)
.expect("Test: operation failed");
// Load back from disk
let loaded = disk_array.load().expect("Test: operation failed");
// Verify the results
// NOTE: Lazy evaluation is now working properly and applies the multiply by 2 operation
// The diagonal should have values ((1+10)*2), ((2+10)*2), etc.
// and other elements should be 10*2 = 20.0.
for i in 0..n {
for j in 0..n {
if i == j {
assert_relative_eq!(
loaded[[i, j]],
((i + 1) + 10) as f64 * 2.0, // Now properly applies multiplication
epsilon = 1e-10
);
} else {
assert_relative_eq!(loaded[[i, j]], 20.0, epsilon = 1e-10); // Changed back to 20.0
}
}
}
}
#[test]
fn test_masked_arrays_with_chunking() {
// Create test data with NaN values
let mut data = Array2::from_shape_fn((10, 10), |(i, j)| (i * 10 + j) as f64);
// Add some NaN values
data[[0, 0]] = f64::NAN;
data[[3, 4]] = f64::NAN;
data[[7, 9]] = f64::NAN;
// 1. Create a masked array that masks out NaN values
let masked = masked_invalid(data.clone());
// 2. Process the masked array in chunks
let result = chunk_wise_op(
&masked.data,
|chunk| chunk.map(|&x| x * 2.0),
ChunkingStrategy::Fixed(3),
)
.expect("Test: operation failed");
// 3. Create a new masked array with the result but keep the original mask
let doubled_masked = mask_array(result, Some(masked.mask.clone()), Some(0.0))
.expect("Test: operation failed");
// Verify the results
for i in 0..10 {
for j in 0..10 {
if (i == 0 && j == 0) || (i == 3 && j == 4) || (i == 7 && j == 9) {
assert!(doubled_masked.mask[[i, j]]);
} else {
assert!(!doubled_masked.mask[[i, j]]);
assert_relative_eq!(
doubled_masked.data[[i, j]],
2.0 * (i * 10 + j) as f64,
epsilon = 1e-10
);
}
}
}
}
}