def forward(ctx, twiddle, input, increasing_stride=True):
"""Experimental in-place implementation that does not store intermediate results.
Instead, the intermediate results are computed from the output during the backward pass.
Parameters:
twiddle: (n - 1, 2, 2) if real or (n - 1, 2, 2, 2) if complex
input: (batch_size, n) if real or (batch_size, n, 2) if complex
increasing_stride: whether to multiply with increasing stride (e.g. 2, 4, ..., n/2) or
decreasing stride (e.g., n/2, n/4, ..., 2).
Note that this only changes the order of multiplication, not how twiddle is stored.
In other words, twiddle[@log_stride] always stores the twiddle for @stride.
Returns:
output: (batch_size, n) if real or (batch_size, n, 2) if complex
"""
assert increasing_stride, 'Decreasing stride not implemented'
output = butterfly_multiply_inplace(twiddle, input)
ctx.save_for_backward(twiddle, output)
return output
[dst(x) for _ in range(ntrial)]
end = timer()
dst_times[idx_n] = (end-start) / ntrial
# BP
start = timer()
[BP_mul_cy_inplace(ABCDs, perm, x) for _ in range(ntrial)]
end = timer()
bp_times[idx_n] = (end-start) / ntrial
# BP_inplace_all
twiddles = twiddle_list_concat(B)
x_torch = torch.tensor(x).unsqueeze(0)
# perm_torch = torch.tensor(perm)
start = timer()
[butterfly_multiply_inplace(twiddles, x_torch) for _ in range(ntrial)]
end = timer()
bp_all_times[idx_n] = (end-start) / ntrial
print(dense_times)
print(fft_times)
print(scipyfft_times)
print(dct_times)
print(dst_times)
print(bp_times)
print(bp_all_times)
# print(bp_times / fft_times)
# print(bp_times / dct_times)
# print(bp_times / dst_times)
# print(bp_times / bp_all_times)
def forward(ctx, twiddle, input):
output = butterfly_multiply_inplace(twiddle, input)
ctx.save_for_backward(twiddle, output)
return output