A minimal template implementation of n-dimentional array. An element unit is 0-dim tensor.
Usage example:
Tensor<float> t1 = Tensor<float>({5, 5, 5}); // create new tensor with shape 5x5x5
Tensor<float> t2 = t1(3, 4); // shape 1x5x5, shares data with t1
Tensor<float> t3 = t2; // shape 1x5x5, doesn’t share any data with t1 and t2 as copied
Tensor<float> t4; // empty tensor
t4 = t1(3); // shape 5x5, shares data with t1 and t2, move semantics here
Tensor<float> t5 = t1(3)(4)(2, 5); // shape 3, shares data with t1, t2, t4
t5[{0}] = 100.0; // t1[{3, 4, 3}], t2[{0, 4, 3}], t4[{4, 3}] are 100.0 as well, but t3[{0, 4, 3}] is 0.0
t1[{3, 4, 4}] = -100.0; // t2[{0, 4, 4}], t4[{4, 4}], t5[{2}] are -100.0 as well, but t3[{0, 4, 4}] is 0.0License: WTFPL