Xtensor append values to bottom of tensor?

Suppose I'm trying to shift (not bitwise) a tensor of values up by n and then add NaNs at the bottom to fill the shift offset.

int n = 2;

original {{ 1., 2., 3., 4. } }

xt::xtensor_fixed<float, xt::xshape<1, 4>, xt::layout_type::column_major> original = {{ 1., 2., 3., 4. } };

values to keep following shift of n {{3., 4. } }

auto shiftvalues = xt::view(original, xt::all(), xt::range(n, original.size()));

after inserting nans to match original size {{3., 4. , nan, nan} }

(?)shiftvalues.appendBottom(std::numeric_limits<double>::quiet_NaN());
(?)shiftvalues.appendBottom(std::numeric_limits<double>::quiet_NaN());

int n = 2;
xt::xtensor_fixed<float, xt::xshape<1, 4>, xt::layout_type::column_major> original = {{ 1., 2., 3., 4. } };
auto shiftvalues = xt::view(original, xt::all(), xt::range(n, original.size()));

How would I go about inserting the NaN values into the tensor following steps 1 and 2? I haven't been able to find any methods in the documentation to append values to the bottom of a tensor.

Or would it be better to create a tensor of the same size as original, fill it with NaNs, then set the values at appropriate indexes to shiftvalues?

Thanks

Solution

Nevermind figured it out

int n = 2;
xt::xtensor_fixed<float, xt::xshape<1, 4>, xt::layout_type::column_major> original = {{ 1., 2., 3., 4. } };
xt::xtensor<float, 2>::shape_type sh1 = {1, n};

auto offsetfill = xt::full_like(xt::empty<float>(sh1), std::numeric_limits<double>::quiet_NaN());
auto shiftvalues = xt::view(original, xt::all(), xt::range(n, original.size()));

auto result = xt::concatenate(xt::xtuple(shiftvalues, offsetfill), 1);
std::cout << result << std::endl;

{{ 3., 4., nan., nan.}}

Its a bit of a workaround I suppose, there's no appending values in the traditional way.

Create a shape with specified offset size n
Put shape into temp tensor. Fill the temp tensor with value of choice.
Slice original tensor to keep values you want.
Concatenate 3 and 2 to get desired result