The Rate of Convergence of the Least Square Estimator in a Nonlinear Regression with φ-mixing Errors

For nonlinear model X_n＝y_n（θ）＋ε_n, we obtain the a.s. convergence rate of the LS estimatorθn, provided that ε_n is an φ-mixing errors with Eε_i=θ, sup_iE|ε_i|^t＜∞（for some t＞2）,\sum_i^\infty \Phi^1 / 2(i)<\inftyand other suitabe conditions, which generalized and improved Prakasa Rao’sresult （J. Mult. Anal. 14（1984）,315-322）.