We can borrow an approach from DMRG, where a density matrix for a large system is stored sparsely and only a fixed a number of eigenvalues, sorted by largest to smallest, is kept. We can implement a similar approach in Pauli Propagation: instead of truncating all coefficients below a certain threshold, we can truncate all but the largest $k$ coefficients in a given PauliSum. This basically just requires sorting the terms by absolute value of coefficients and then truncating the smallest until there are only $k$ terms left in the sum. This is kind of like a CoefficientThreshold where the threshold is adaptive during the propagation.
We can borrow an approach from DMRG, where a density matrix for a large system is stored sparsely and only a fixed a number of eigenvalues, sorted by largest to smallest, is kept. We can implement a similar approach in Pauli Propagation: instead of truncating all coefficients below a certain threshold, we can truncate all but the largest$k$ coefficients in a given $k$ terms left in the sum. This is kind of like a
PauliSum. This basically just requires sorting the terms by absolute value of coefficients and then truncating the smallest until there are onlyCoefficientThresholdwhere the threshold is adaptive during the propagation.