At present, the only way to fix the magnetization in the TEBD code for the Hubbard model is to add a magnetic field. Otherwise, the magnetization is set as you described.
Is there any way to fix the no. of up and down spins in the TEBD code for the fermion hubbard model? It seems like I'm only allowed to conserve total N and the up and down are randomly assigned and varies from run to run.