[Rafael-SOWNet]

A complex number's parts are decimals in the precision context, whose exponents end at 1e1000. A product or a whole power of finite numbers that had a part pass that end had oo - oo in it and came out NaN, which says the value does not exist. Simplify returned the same NaN. 2.6.0 does the same. A compilesweep follow-up found it; no issue was filed.

	2.6.0 and master	now
"(3+4i)^2000".EvalNumerical()	NaN	+oo + +ooi (each part the sign of the exact part)
"(1+i)^(10^6)".EvalNumerical()	NaN	+oo
"(1e600+1e600i)*(1e600+1e600i)".EvalNumerical()	NaN	+ooi
"(3+4i)^2000".Simplify()	NaN	(3 + 4i) ^ 2000
(3+4i)^1400, (3+4i)^(-2000)	finite, 0	unchanged

• Product. When the parts of a complex product of finite factors leave the range, the parts are formed with unbounded exponents and each is rounded into the context. The bounded result is used whenever it is finite, so a value in range takes no new path.
• Power. A complex power that leaves the range on the way is raised the same way.
• Division needs nothing. Smith's method only adds terms of like sign, and the near-one route ends in the product above. (1e600+1e600i)/(1e-600+1e-600i) is still NaN, because 1e-600 is below the context's exponent floor and the divisor evaluates to 0. That is a separate, known limit.
• Simplification. Evaluate an exact power too large to build as the decimal #2347's guard against an infinity that only rounding made now covers any sum, product or power of exact finite operands (a power's base not zero). Their value is exact and finite. An operand that is already a decimal is not covered. The integrator reads a decimal sample past the range as +oo, and the first version of this guard took that away: one Rubi antiderivative came back 23,030 characters long.

Allocation. The kernel gate failed SolveHard at +5.1% on the first commit. #2347 had already moved it +2.6%, inside the 3% limit. Both halves came from reading IsFinite, a cache built on each entity's first read: #2347's guard read it on every exact value, and this branch's power check on every power built. The second commit reads finiteness from the parts instead, and computes #2347's size bound in longs. SolveHard allocates 9,549,209 B, against 9,546,434 B before #2347.

Tests: AComplexOverflowIsNotNaNTest, 12 cases. For each Gaussian power, the test checks the sign of every part against the exact Gaussian integer. 10 fail on master; the 2 controls pass on both. The full suite, AOT and the nine harnesses pass. There is a BREAKING-CHANGES entry, measured on 2.6.0.

🤖 Generated with Claude Code

[codecov]

Codecov Report

❌ Patch coverage is 97.22222% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 86.76%. Comparing base (fb99ffe) to head (dc07fd1).

Files with missing lines	Patch %	Lines
...uriMath/Functions/Evaluation/Evaluation.Classes.cs	80.00%	0 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #2348      +/-   ##
==========================================
- Coverage   86.77%   86.76%   -0.01%     
==========================================
  Files         312      312              
  Lines       68645    68664      +19     
  Branches    19980    19987       +7     
==========================================
+ Hits        59567    59577      +10     
- Misses       4910     4918       +8     
- Partials     4168     4169       +1     

Flag	Coverage Δ
cpp	37.77% <ø> (ø)
csharp	87.30% <97.22%> (-0.01%)	⬇️
fsharp	32.97% <ø> (ø)
interactive	40.72% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
...uriMath/Core/Entity/Continuous/Number/Operators.cs	90.87% <100.00%> (+0.35%)	⬆️
...uriMath/Functions/Evaluation/Evaluation.Classes.cs	84.90% <80.00%> (-0.48%)	⬇️

... and 7 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.