This is the code repository for "Learning measurement-induced phase transition using attention".

Due to the large volume, Data_src folder is not present in this repository. It is provided upon reasonable request.

• 1-StateDistinguishing: Training and testing results for state distinguishing task (learnability transition.)
• 2-PhaseRecognition: Training and testing results for phase recognition task. Folder includes inter-trajectory attention score data extraction.
• Data_out folder contains training outcomes. Fig_*.ipynb files analyze results and generate figure for the main and supplementary text.

model_name ranges from QuAN4_Tbt2 (full model), SMLP_Tbt2 (temporal attention only), SMLP_Tbm2 (no attention). System size are even number between 8 and 20. Sample size should be smaller or equal to 6,000,000. Monitoring strength is elaborated in the main text of the paper.

bash train_statedistinguishing.sh "model_name" "system size L" "set size N" "sample size M" "monitoring strength gamma"

bash test_statedistinguishing.sh "model_name" "system size L" "set size N" "sample size M" "monitoring strength gamma"

model_name ranges from QuAN4_Tbt2 (full model), SMLP_Tbt2 (temporal attention only), SMLP_Tbm2 (no attention). System size are even number between 8 and 20. Sample size should be smaller or equal to 40,000. Monitoring strength is elaborated in the main text of the paper.

bash train_phaserecognition.sh "model_name" "system size L" "set size N" "sample size M"

bash test_phaserecognition.sh "model_name" "system size L" "set size N" "sample size M"
bash test_attn.sh "model_name" "system size L" "set size N" "sample size M"