1: \begin{abstract}	{The `` Inertial Forward-Backward algorithm '' (IFB) is a powerful tool for convex nonsmooth minimization problems, and under the local error bound condition, the $R$-linear convergence rates for the sequences of objective values and iterates have been proved if the inertial parameter $\gamma_k$ satisfies ${\sup _k}{\gamma _k} < 1.$ However, the convergence result for ${\sup _k}{\gamma _k} = 1$ is not know. In this paper, based on the local error bound condition, we exploit a new assumption condition for the important parameter $t_k$ in IFB, which implies that ${\lim _{k \to \infty }}{\gamma _k} = 1,$ and establish the convergence rate of function values and strong convergence of the iterates generated by the IFB algorithms with six $t_k$ satisfying the above assumption condition in Hilbert space. It is remarkable that, under the local error bound condition, we show that the IFB algorithms with some $t_k$ can achieve sublinear convergence rate of $o\left( {\frac{1}{{{k^p}}}} \right)$ for any positive integer $p>1$. In addition, we propose a class of Inertial Forward-Backward algorithm with adaptive modification and show it has same convergence results as IFB under the error bound condition. Some numerical experiments are conducted to illustrate our results.

2: % Body of abstract:

3: }

4: % Keywords:

5: {Inertial Forward-Backward algorithm; local error bound condition; rate of convergence.}

6: \end{abstract}