1: \begin{abstract}

2: 	 In this paper we proposed two new quasi-boundary value methods for regularizing the ill-posed backward heat conduction problems. With a standard finite difference discretization in space and time, the obtained all-at-once nonsymmetric sparse linear systems have the desired block $\omega$-circulant structure, which can be utilized to design an efficient parallel-in-time (PinT) direct solver that built upon an explicit FFT-based diagonalization of the time discretization matrix. 

3: 	 Convergence analysis is presented to justify the optimal choice of the regularization parameter.

4: 	 Numerical examples are reported to validate our analysis and illustrate the superior computational efficiency of our proposed PinT methods.

5: 	  

6: 	\end{abstract}

7: