ERA returns NaN with ForwardDiff
Closed this issue · 1 comments
MicheleCeresoli commented
When trying to compute a derivative with ForwardDiff, the iers_era
function returns a NaN type, which does not allow the AD computation of the GCRF-to-ITRF rotation
julia> using IERSConventions
julia> using JSMDUtils.Math: D¹
julia> D¹(t->iers_era(iers2010a, t), 1)
NaN
MicheleCeresoli commented
This is caused by the usage of the mod
function:
julia> ForwardDiff.derivative(t->mod(t, 1), 2)
NaN