SYSML21 — ISQ/SI libraries: unit expressions contradict declared quantity dimensions (Hartree energy is energy cubed, dose rates typed as dose, alternative units multiplied together)

Found by reducing every unit expression in SI.sysml to its dimension and comparing with the quantityPowerFactors of the declared ...Unit type (SysML-v2-Release commit fb97b754). Line numbers are in SI.sysml unless stated.

1. Alternative units from one ISO table cell multiplied into a single product

Line 136 is the most serious because the error is also in the generated ISQ model: ISQAtomicNuclear.sysml lines 259-263 declare HartreeEnergyUnit with exponents L 6, M 3, T -6, and the documentation (line 247) says "quantity dimension: L^6*M^3*T^-6, measurement unit(s): eV*J*kg*m^2*s^-2" for a quantity defined as an energy, E_H = e^2/(4 pi eps_0 a_0). The three alternative units eV, J, kg*m^2*s^-2 were joined into a product and the dimension was derived from that product. It should be L^2 M T^-2. For line 163 the documentation at ISQCondensedMatter.sysml line 554 repeats the product; for line 247 the documentation at ISQCondensedMatter.sysml line 626 lists the alternatives correctly as "m^3/C, m^3*s^-1*A^-1". Each alternative should become its own unit declaration. Since the ISQ files are generated, the generator's handling of multi-unit cells probably needs the fix, and other quantities may be affected in ways an expression-versus-type comparison cannot see.

2. Exponent sign

Line 137: TotalMassStoppingPowerUnit = eV*m^-2/kg gives T^-2; the type declares L^4 T^-2. Should be eV*m^2/kg (compare line 147, J*m^2/kg).

3. Dose equivalent rate typed as dose equivalent

Lines 286, 299, 239: 'sievert per second' : DoseEquivalentUnit = Sv/s, 'watt per kilogram' : DoseEquivalentUnit = W/kg, 'metre squared second to the power minus 3' : DoseEquivalentUnit = m^2*s^-3. Each expression is L^2 T^-3; DoseEquivalentUnit (ISQAtomicNuclear.sysml line 2520) declares L^2 T^-2. The origin is ISQAtomicNuclear.sysml line 2527, attribute doseEquivalentRate: DoseEquivalentValue, whose own documentation says "quantity dimension: L^2*T^-3". The rate needs its own value and unit types; the "sievert per second" present since the 2020-09 release shows this is long-standing.

4. Why this is not detected

MeasurementReferences.sysml line 276 asserts hasValidUnitPowerFactors : VerifyUnitPowerFactors, but the constraint definition (lines 517-525) has only inputs and documentation, no Boolean expression, and MeasurementRefCalculations.sysml declares unit '*', '/', '^' as signatures without dimension arithmetic. Suggest giving the constraint a body, or adding a library regression check: expression-versus-type dimension comparison, plus anchor values for a few conversion factors (which would also have caught SYSML21-405).