- TrainSum.linsolver(rhs, *maps, method='dmrg', solver=GMRES(nsteps=10, subspace=10, eps=1e-08, solver=<trainsum.lstsqsolver.LstsqSolver object>), decomposition=SVDecomposition(max_rank=50, cutoff=0.0), strategy=SweepingStrategy(ncores=2, mode='connected', nsweeps=10, min_size=None), optimizer='greedy')
Variational linear solver for quantics tensor trains. Note that the meaning of the decomposition differs for the DMRG and AMEn algorithms. For DMRG it is used to recover single cores in multi-core sweeping strategies, thereby defining the rank. For AMEn it is used for the enrichment step, so the rank grows with 2*max_rank in each step.
- Parameters:
rhs (TensorTrain)
maps (LinearMap)
method (Literal['dmrg', 'amen'])
solver (LocalLinSolver)
strategy (SweepingStrategy)
optimizer (Literal['optimal', 'dp', 'greedy', 'random-greedy', 'random-greedy-128', 'branch-all', 'branch-2', 'auto', 'auto-hq'])
- Return type: