On adaptive information with varying cardinality for linear problems with elliptically contoured measures (Q1263942)

The paper deals with a generalization of author's earlier results related to the comparison of power of adaptive and nonadaptive information. The earlier results [e.g., Numer. Math. 44, 169-190 (1984; Zbl 0555.65030), J. Complexity 2, 204-228 (1986; Zbl 0615.94004)] indicate that adaptive information is not more powerful than nonadaptive information whether information cardinality is fixed (in the case of linear problems with elliptically contoured measures) or allowed to vary (with Gaussian measure). The current attempt is to generalize the results to a class of elliptically contoured measures for which the answer still remains negative. The paper begins with the definition of various terminologies (e.g., nonadaptive and adaptive information, fixed and varying cardinalities, etc.) in a separable Banach space. The main theorem is established assuming the probability measure to be elliptically contoured.