[jkang-pf]

reactivity computes ⟨σv⟩ with er, which holds reaction energies (18.3e6, 4.03e6 and 0.82e6 eV) for D+He3→He4, D+D→T and D+D→He3. The Bosch-Hale fit needs the reduced-mass energy m_r c². The correct values are already defined as mc2 in the same function but are never used. ⟨σv⟩ scales as 1/√(m_r c²), so at every temperature D+D→He3 was 6.9 % too high, D+D→T 52 % too low and D+He3→He4 75 % too low. D+T→He4 was correct.

The fix uses mc2 in the formula, defines the D-T value (1124656) under the same name and removes the er lines. Nothing else changes.

The new test/runtests_nuclear.jl pins 24 values of Table VIII of Bosch & Hale 1992 (thermal reactivities at 1–50 keV, as reprinted in the erratum, Nucl. Fusion 33, 1919) at rtol = 1e-3. 18 of them fail before the fix and all pass after it. I ran this test file with the package (Pkg.test with runtests_nuclear.jl), not the rest of the suite.

This changes the D-D and D+He3 source terms and powers wherever those reactions are enabled. #311 asks about the ½ factor for identical D-D reactants; that is a separate issue and not part of this PR.

🤖 Generated with Claude Code

[jmcclena]

@jkang-pf As long as you're digging into this. Do you have an opinion on whether the factor of 0.5 should be there for #311 ?

[jkang-pf]

Thanks for merging this so quickly, @jmcclena

On #311: yes, I think the factor of 1/2 should be there. The Bosch-Hale is per pair, so for
identical reactants the reaction rate density is R = 1/2 n_D^2 (n_D^2 counts every pair twice).

In nuclear.jl, D_D_to_He3_reactions and D_D_to_T_reactions return n_D^2 without the 1/2:

• D_D_to_He3_source! uses it for both the heating and the particle source (D consumed as -2R, He3
  produced as +R), so that branch is 2x too high.
• D_D_to_T_source! is inconsistent: the T and H particle sources use reactivity / 2.0, which is
  right, but the heating uses the full reactivity, which is 2x too high.
  So the D-D heating and fusion_power are 2x high. A consistent fix is to put the 1/2 in the two
  *_reactions functions and drop the / 2.0 in D_D_to_T_source!. We did that in our IMAS3d (halved the
  identical-particle rate) and pin it with a test, so I can send the same as a PR if you want it.

One thing worth a release note: this PR changes D+D->He3, D+D->T and D+He3->He4 by constant factors of
0.935, 2.07 and 4.03 (D+T->He4 is unchanged), so any D-D or D-3He result in FUSE moves.