rstsr-core 0.7.3

An n-Dimension Rust Tensor Toolkit
Documentation 
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

use crate::prelude_dev::*;

/* #region assign */

pub trait TensorAssignAPI<TRB> {
    fn assign_f(a: &mut Self, b: TRB) -> Result<()>;
    fn assign(a: &mut Self, b: TRB) {
        Self::assign_f(a, b).rstsr_unwrap()
    }
}

pub fn assign_f<TRA, TRB>(a: &mut TRA, b: TRB) -> Result<()>
where
    TRA: TensorAssignAPI<TRB>,
{
    TRA::assign_f(a, b)
}

pub fn assign<TRA, TRB>(a: &mut TRA, b: TRB)
where
    TRA: TensorAssignAPI<TRB>,
{
    TRA::assign(a, b)
}

impl<RA, DA, RB, DB, TA, TB, B> TensorAssignAPI<TensorAny<RB, TB, B, DB>> for TensorAny<RA, TA, B, DA>
where
    RA: DataMutAPI<Data = <B as DeviceRawAPI<TA>>::Raw>,
    RB: DataAPI<Data = <B as DeviceRawAPI<TB>>::Raw>,
    DA: DimAPI,
    DB: DimAPI,
    B: DeviceAPI<TA> + DeviceAPI<TB> + OpAssignAPI<TA, DA, TB>,
{
    fn assign_f(a: &mut Self, b: TensorAny<RB, TB, B, DB>) -> Result<()> {
        // get tensor views
        let mut a = a.view_mut();
        let b = b.view();
        // check device
        rstsr_assert!(a.device().same_device(b.device()), DeviceMismatch)?;
        let device = a.device().clone();
        // check layout
        rstsr_assert!(!a.layout().is_broadcasted(), InvalidLayout, "cannot assign to broadcasted tensor")?;
        let la = a.layout().to_dim::<IxD>()?;
        let lb = b.layout().to_dim::<IxD>()?;
        let default_order = a.device().default_order();
        let (la_b, lb_b) = broadcast_layout_to_first(&la, &lb, default_order)?;
        let la_b = la_b.into_dim::<DA>()?;
        let lb_b = lb_b.into_dim::<DA>()?;
        // assign
        device.assign(a.raw_mut(), &la_b, b.raw(), &lb_b)
    }
}

impl<RA, DA, RB, DB, TA, TB, B> TensorAssignAPI<&TensorAny<RB, TB, B, DB>> for TensorAny<RA, TA, B, DA>
where
    RA: DataMutAPI<Data = <B as DeviceRawAPI<TA>>::Raw>,
    RB: DataAPI<Data = <B as DeviceRawAPI<TB>>::Raw>,
    DA: DimAPI,
    DB: DimAPI,
    B: DeviceAPI<TA> + DeviceAPI<TB> + OpAssignAPI<TA, DA, TB>,
{
    fn assign_f(a: &mut Self, b: &TensorAny<RB, TB, B, DB>) -> Result<()> {
        TensorAssignAPI::assign_f(a, b.view())
    }
}

impl<S, D> TensorBase<S, D>
where
    D: DimAPI,
{
    pub fn assign_f<TRB>(&mut self, b: TRB) -> Result<()>
    where
        Self: TensorAssignAPI<TRB>,
    {
        assign_f(self, b)
    }

    pub fn assign<TRB>(&mut self, b: TRB)
    where
        Self: TensorAssignAPI<TRB>,
    {
        assign(self, b)
    }
}

/* #endregion */

/* #region fill */

pub trait TensorFillAPI<T> {
    fn fill_f(a: &mut Self, b: T) -> Result<()>;
    fn fill(a: &mut Self, b: T) {
        Self::fill_f(a, b).rstsr_unwrap()
    }
}

pub fn fill_f<TRA, T>(a: &mut TRA, b: T) -> Result<()>
where
    TRA: TensorFillAPI<T>,
{
    TRA::fill_f(a, b)
}

pub fn fill<TRA, T>(a: &mut TRA, b: T)
where
    TRA: TensorFillAPI<T>,
{
    TRA::fill(a, b)
}

impl<RA, DA, TA, TB, B> TensorFillAPI<TB> for TensorAny<RA, TA, B, DA>
where
    RA: DataMutAPI<Data = <B as DeviceRawAPI<TA>>::Raw>,
    DA: DimAPI,
    B: DeviceAPI<TA> + OpAssignAPI<TA, DA, TB>,
{
    fn fill_f(a: &mut Self, b: TB) -> Result<()> {
        // check layout
        rstsr_assert!(!a.layout().is_broadcasted(), InvalidLayout, "cannot fill broadcasted tensor")?;
        let la = a.layout().clone();
        let device = a.device().clone();
        device.fill(a.raw_mut(), &la, b)
    }
}

impl<S, D> TensorBase<S, D>
where
    D: DimAPI,
{
    pub fn fill_f<T>(&mut self, b: T) -> Result<()>
    where
        Self: TensorFillAPI<T>,
    {
        fill_f(self, b)
    }

    pub fn fill<T>(&mut self, b: T)
    where
        Self: TensorFillAPI<T>,
    {
        fill(self, b)
    }
}

/* #endregion */

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn test_assign_with_cast() {
        let mut device = DeviceCpuSerial::default();
        device.set_default_order(RowMajor);
        let mut a: Tensor<f32, _> = zeros(([2, 3], &device));
        let b = arange((6i32, &device)).into_shape((2, 3));
        a.assign(&b);
        assert_eq!(a.raw(), &vec![0.0f32, 1.0, 2.0, 3.0, 4.0, 5.0]);

        let c: i32 = 10;
        a.fill(c);
        assert_eq!(a.raw(), &vec![10.0f32; 6]);
    }

    #[test]
    #[cfg(feature = "faer")]
    fn test_assign_with_cast_faer() {
        let mut device = DeviceFaer::default();
        device.set_default_order(RowMajor);
        let mut a: Tensor<f32, _> = zeros(([2, 3], &device));
        let b = arange((6i32, &device)).into_shape((2, 3));
        a.assign(&b);
        assert_eq!(a.raw(), &vec![0.0f32, 1.0, 2.0, 3.0, 4.0, 5.0]);

        let c: i32 = 10;
        a.fill(c);
        assert_eq!(a.raw(), &vec![10.0f32; 6]);
    }
}