1: \begin{abstract}

2: 

3: In usual (nonstochastic) tensor network calculations, the truncated singular value decomposition is often used for approximating a tensor,

4: and it causes systematic errors.

5: By introducing stochastic noise in the approximation, however, 

6: one can avoid such systematic errors at the expense of statistical errors which can be straightforwardly controlled.

7: Therefore in principle, exact results can be obtained even at finite bond dimension up to the statistical errors.

8: A previous study of the unbiased method implemented in tensor renormalization group algorithm, however, showed that

9: the statistical errors for physical quantity are not negligible, and

10: furthermore the computational cost is linearly proportional to a system volume.

11: In this paper, we introduce a new way of stochastic noise such that the statistical error is suppressed, and

12: moreover, in order to reduce the computational cost we propose common noise method whose cost is proportional to the logarithm of volume.

13: We find that the method provides better accuracy for the free energy compared with 

14: the truncated singular value decomposition when applying to tensor renormalization group for Ising model on square lattice.

15: Although the common noise method introduces systematic error originated from a correlation of noises,

16: we show that the error can be described by a simple functional form in terms of the number of noises, thus

17: the error can be straightforwardly controlled in an actual analysis.

18: We also apply the method to 

19: the graph independent local truncation

20: algorithm and show that the accuracy is further improved.

21: 

22: \end{abstract}

23: