|
We present a full step feasible interior-point algorithm for circular cone optimization us-
ing Euclidean Jordan algebras. The speci city of our method is to use a transformation similar to
that introduced by Darvay and Takacs for the centering equations of the central path of the linear
optimization. The Nesterov and Todd symmetrization scheme is used to derive the search directions.
The theoretical complexity bound of the algorithm coincides with the best-known iteration bound for
small-update methods.
|