space::irbem::TotalFieldMead
cheatah-space v0.1.0-alpha — Biome Standard 0.6.5-alpha
IGRF plus Mead & Fairfield (1975), as a single field.
The same shape as TotalFieldT89, for the same reason: a particle experiences one field, and trace_invariant and make_lstar follow whatever satisfies GeoFieldModel without knowing there are two of them. IGRF is defined in GEO and the external model in GSM; the rotation between them depends only on the epoch and is built once in Rotations.
There is no device trace lane for this model — traces through it run on the host through the generic trace_invariant, which is where a model this cheap belongs: the external term adds ~50 flops to each of ~600 IGRF evaluations of a trace, so a dedicated kernel would buy almost nothing over irbem_trace_i_f32's measured curve.
NMAX | the internal field's truncation degree. 10 reproduces IRBEM's own choice. |
IrbemMead.TotalFieldSuperposesInternalAndExternalFunctions
constexpr TotalFieldMead(const Igrf< NMAX > &internal, const Rotations &rotations, double kp_times_ten)
#
internal | the internal field, already built for the epoch. |
rotations | the epoch's frame rotations — built once, reused for every point. |
kp_times_ten | Kp in IRBEM's slot-1 scaling (Kp x 10, nominally 0..90). Binned by mead_kp_bin: values inside one group give identical fields by construction. |
O(1) — three pointers and a double.
none.
IrbemMead.TotalFieldSuperposesInternalAndExternalThe total field at a geographic point.
p | the position, GEO, Earth radii. |
B_internal + B_external, in GEO, nT. When the external model refuses the point — a non-finite input or a point inside the Earth — the INTERNAL field is returned alone rather than a zero or a NaN, exactly as TotalFieldT89 does; an out-of-validity external field is still added, because extrapolating the fit is the caller's decision and external_status is how they learn it was made.
One IGRF evaluation, one Mead-Fairfield evaluation, two 3x3 rotations.
none.
IrbemMead.TotalFieldSuperposesInternalAndExternalWhether the external model answered at p, and if not, why.
p | the position, GEO, Earth radii. |
the external model's status; Status::Ok when it contributed without caveat.
One Mead-Fairfield evaluation and one rotation.
none.
IrbemMead.TotalFieldReportsWhenTheExternalModelDeclinesThe activity level this field was built for, in IRBEM's Kp x 10 scaling.
the value passed to the constructor.
O(1).
none.
IrbemMead.TotalFieldSuperposesInternalAndExternalThe epoch's frame rotations.
the rotations this field was built with.
O(1).
none.
IrbemMead.TotalFieldSuperposesInternalAndExternalThe internal part's Gauss coefficient g(n, m), in nT — the INTERNAL field's, because a superposition with a non-potential external term has no expansion of its own; see TotalFieldT89::g for the full argument.
n | the degree. |
m | the order. |
the internal part's coefficient.
O(1).
none.
IrbemMead.TotalFieldSuperposesInternalAndExternalThe internal part's h(n, m), in nT — see g for why this forwards.
n | the degree. |
m | the order. |
the internal part's coefficient.
O(1).
none.
IrbemMead.TotalFieldSuperposesInternalAndExternalThe internal field alone — what dipole_moment and any staging need.
the internal model.
O(1).
none.
IrbemMead.TotalFieldSuperposesInternalAndExternalConstants & variables
The internal part's truncation degree — what generic staging and buffer sizing read.