Current Problem
Currently, MaxGap may apply auxiliary variables to the Ising equation. This could potentially lead to double counting a feasible state.
(ex. One feasible state, S1, may have only one combination of auxiliary variables that lead to ground, while another feasible state, S2, could have two combinations of auxiliary variables that lead to ground. If S1 and S2 are the only feasible states, then in the case specified here, the S2 state is twice as likely to be sampled than S1.)
Proposed Solution
After MaxGap is run and the necessary auxiliary variables are found, run LP to refine the coefficients so that there is a uniform distribution among the feasible states.
(cont'd previous example: one of the S2 + auxiliary states is made to exceed the ground energy level. In this way S1 and S2 are equally likely to be sampled).
Current Problem
Currently, MaxGap may apply auxiliary variables to the Ising equation. This could potentially lead to double counting a feasible state.
(ex. One feasible state, S1, may have only one combination of auxiliary variables that lead to ground, while another feasible state, S2, could have two combinations of auxiliary variables that lead to ground. If S1 and S2 are the only feasible states, then in the case specified here, the S2 state is twice as likely to be sampled than S1.)
Proposed Solution
After MaxGap is run and the necessary auxiliary variables are found, run LP to refine the coefficients so that there is a uniform distribution among the feasible states.
(cont'd previous example: one of the S2 + auxiliary states is made to exceed the ground energy level. In this way S1 and S2 are equally likely to be sampled).