interesting result on hadamard product coefficient to Mx<=1 form

[smilesun]

In this version, I mistakenly ignored $w_t$ in $st=(A+BK^s)s{t-1}+w_t$, which result in the following wrong derivation:

https://github.com/smilesun/tube_mpc/blob/f419c6b23e5496d090bb471daec861f509ef3ef2/tmpc/constraint_tightening_z_terminal.py#L43

where c_1 c_2 are two different vectors taking Hadamard product on the form Mx<=ones(nrow, 1). One is able to reduce this to

M'x<=1 M'Ax<=1 M'A^sx<=1 form

[smilesun]

unfortunately, here, with frac{c_1}{c_2}, one can not ensure the form A', A'^2, A'^3 for all subsequent steps https://github.com/smilesun/tube_mpc/blob/f419c6b23e5496d090bb471daec861f509ef3ef2/tmpc/constraint_tightening_z_terminal.py#L52

However, we do not really need a stardard form to calculate positive invariant set,
M'x<=1 M'A_1x<=1 M'A_1A_2x<=1 M'A_1A_2A_3x<=1

might also be possible?