Measurement-induced phase transitions by matrix product states scaling

Guillaume Cecile,Hugo Lóio,Jacopo De Nardis
2024-03-13
Abstract:We study the time evolution of long quantum spin chains subjected to continuous monitoring via matrix product states (MPS) at fixed bond dimension, with the Time-Dependent Variational Principle (TDVP) algorithm. The latter gives an effective classical non-linear evolution with a conserved charge, which approximates the real quantum evolution up to an error. We show that the error rate displays a phase transition in the monitoring strength, which can be well detected by scaling analysis with relatively low values of bond dimensions. The method allows for an efficient numerical determination of the critical measurement-induced phase transition parameters in many-body quantum systems. Moreover, in the presence of U(1) global spin charge, we show the existence of a charge-sharpening transition well separated from the entanglement transition which we detect by studying the charge fluctuations of a local sub-part of the system at very large times. Our work substantiates the TDVP time evolution as a method to identify measured-induced phase transitions in systems of arbitrary dimensions and sizes.
Statistical Mechanics,Disordered Systems and Neural Networks,Quantum Physics
What problem does this paper attempt to address?