We perform a dynamical analysis of the recently updated set of the radial velocity ( RV ) measurements of the HD 160691 ( \mu Arae ) .
The purely kinematic , 2-Keplerian model of the measurements leads to the best-fit solution in which the eccentricity of the outer planet is about 0.7 and its semi-major axis is about 4 AU .
The parameters of the inner planet are well determined .
The eccentricity is about 0.3 and the semi-major axis is about 1.65 AU .
In such circumstances , the best 2-Keplerian model leads to a catastrophically unstable configuration , which is self disrupting in less than 20,000 yr .
In order to derive dynamically stable configurations which are simultaneously consistent with the RV data , we use the so called GAMP ( Genetic Algorithm with MEGNO Penalty ) which incorporates the genetic fitting algorithms and a verification of the stability by the fast indicator of the dynamics ( MEGNO ) .
Using this method , we derive meaningful limits on the parameters of the outer planet which also provide a stable behavior of the system .
It appears , that the best-fit solutions are located in a shallow valley of { ( \chi ^ { 2 } _ { \nu } ) ^ { 1 / 2 } } , in the ( a _ { { \mbox { \scriptsize c } } } ,e _ { { \mbox { \scriptsize c } } } ) -plane , extending over 2 AU ( for the formal 1 \sigma confidence interval of the best fit ) .
We find two equally good best-fit solutions leading to the qualitatively different orbital configurations .
One of them corresponds to the center of the 5:1 mean motion resonance ( MMR ) and the second one describes a configuration between the 6:1 and 17:2 MMRs .
Generally , the HD 160691 system can be found in the zone confined to other low-order MMRs of the type p :1 with p > 5 .
Our results support the classification of the \mu Ara as a hierarchical planetary system , dynamically similar to other known multi-planet systems around HD 12661 and HD 169830 .
The results of the GAMP analysis on the extended data set are fully consistent with our previous conclusions concerning a much shorter observational window .
It constitutes a valuable confirmation that the applied method is well suited for the analysis of the RV data series which only partially cover the longest orbital period .