Errata for "G. Harcos, P. Michel, The subconvexity problem for Rankin-Selberg L-functions and equidistribution of Heegner points. II, Invent. Math. 163 (2006), 581-655."

The following correction was kindly communicated to me by Valentin Blomer.

Page 647, third display: This formula needs three fixes. First, on the left hand side, psi should be conjugated as 
in [Z01b,Th.1.3.2] and [Z04,Th.14.2]. Second, on the right hand side, the L-function L(...) should be replaced by
the completed L-function Lambda(...) as in [Z01b,Th.1.3.2] and [Z04,Th.14.2]. Alternatively, the L-function L(...)
can be left unchanged, but then <g~,g~> needs to be replaced by L(1,sym^2 g). Third, on the right hand side, one 
should insert the factor (w/2)^2, where w is the number of units in K. It seems that this subtle error is also 
present in [Z01b] and [Z04]. The point is that, for an open compact subgroup U of T(A_f), the measure space 
T(F)\T(A)/U can be identified with h copies of T(F_infty)/S, where h=|T(F)\T(A_f)/U| is a generalized class number, 
and S=T(F_infty)\cap T(F)U is the image of a generalized unit group in T(F_infty). Hence the measure induced on 
T(F_infty)-invariant functions is not the counting measure on T(F)\T(A_f)/U, but the counting measure multiplied by 
the volume of T(F_infty)/S. The last volume equals 1/|S| by the normalization of [Z01b] and [Z04], hence the resulting
measure on T(F)\T(A_f)/U is the counting measure divided by |S|. In the current situation, T(F_infty)=C*/R*
is the unit circle modulo {+-1}, so S is isomorphic to the unit group modulo {+-1}.

Page 649, first display: On the right hand side, the order of the unit group should be squared, and the first 
denominator 2 should be deleted. Note that E(z,s) here is 2zeta(s) times the E(z,s) in [GZ86,p.248].