Adaptive truncation

By default every inner bond of the output has dimension chi_out, even if the product has lower rank across that cut. Set cutoff to trim each bond to its effective rank:

import quimb.tensor as qtn

from src_method import apply

H = qtn.MPO_identity(8, phys_dim=2)
psi = qtn.MPS_rand_state(8, bond_dim=4, seed=0)

fixed = apply(H.arrays, psi.arrays, chi_out=32, seed=1)
trimmed = apply(H.arrays, psi.arrays, chi_out=32, cutoff=1e-10, seed=1)

print([a.shape[0] for a in fixed[1:]])
print([a.shape[0] for a in trimmed[1:]])

During the right-to-left sweep, the QR factorisation at each site is followed by an SVD of its small (chi_out, chi_out) triangular factor, and singular values below cutoff * sigma_max are discarded. The extra cost is negligible next to the contractions.

• cutoff must lie in [0.0, 1.0); 0.0 (the default) disables trimming.
• chi_out remains the upper bound: the cutoff only ever lowers a bond.
• Two-site trains use the exact fallback, which ignores cutoff and truncates to chi_out.