Skip to content

Fix farfield rE dimensionalization - #936

Merged
hughcars merged 2 commits into
mainfrom
hughcars/fix-farfield-units
Sep 17, 2026
Merged

hughcars merged 2 commits into
mainfrom
hughcars/fix-farfield-units

Conversation

@hughcars

Copy link
Copy Markdown
Collaborator

Fixes #935

@hughcars
hughcars force-pushed the hughcars/fix-farfield-units branch from 48e9a55 to ebfd5ea Compare September 16, 2026 23:21
@hughcars
hughcars requested a review from Sbozzolo September 16, 2026 23:23
The radiation-zone field r E_inf written to farfield-rE.csv is a field
times a distance, i.e. a voltage, but Measurement::Dimensionalize converted
it with Nondimensionalize<FIELD_E> (and the reload path applied the same
conversion again). The written values were rE / E_c instead of rE * V_c,
off by the factor L_c / Z_0 and hence dependent on the internal
characteristic length: for the dipole antenna example (L_c = 12 m) the
reported far field was ~31x too small, while a run of the same model with
L_c = 1 m gave values 12x larger with every other output unchanged. The
radiated power integrated from the written far field was (L_c / Z_0)^2 of
the accepted port power; with the correct voltage scale it agrees to 0.3 %.
Normalized radiation patterns and the Stratton-Chu kernel (tested against
the analytic dipole with the voltage scale) were unaffected.

Use the VOLTAGE scale in both conversion directions. The idempotent
PostOperator measurement test now carries far-field samples and checks that
the conversion round-trips and that |rE| dimensionalizes as the probe field
(V/m) times the length scale (m); both checks fail on the previous code.
Multiply the six r*E columns of the five farfield-rE.csv references by
Z_0 / L_c, the exact factor of the unit fix in the previous commit
(L_c = 12 m for the antenna cases, x31.394193; L_c = 4.5 mm for the CPW
lumped cases, x83717.847), leaving all other columns byte-identical.
Regenerating the files with the fixed code reproduces the transformed
values to <= 6e-6 (MPI partition noise), and the five cases pass the
regression suite against them.
@hughcars
hughcars force-pushed the hughcars/fix-farfield-units branch from ebfd5ea to 1167e1a Compare September 17, 2026 13:00

@Sbozzolo Sbozzolo left a comment

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thank you!

@hughcars
hughcars merged commit 2cb89b6 into main Sep 17, 2026
58 of 61 checks passed
@hughcars
hughcars deleted the hughcars/fix-farfield-units branch September 17, 2026 16:04
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

farfield-rE.csv values depend on Lc

2 participants