Add x + conj(x) rule

[amilsted]

Not sure if this is the right thing to do, but a PR seemed like the easiest way to bring this up.

[github-actions]

Benchmark Results

	master	40c837f...	t[master]/t[40c837f...]
overhead/acrule/a+2	1.03 ± 0.22 μs	1.31 ± 0.51 μs	0.787
overhead/acrule/a+2+b	1.08 ± 0.33 μs	1.24 ± 0.42 μs	0.869
overhead/acrule/a+b	0.352 ± 0.056 μs	0.487 ± 0.11 μs	0.723
overhead/acrule/noop:Int	22.9 ± 2.1 ns	25.8 ± 2.5 ns	0.89
overhead/acrule/noop:Sym	0.0467 ± 0.0096 μs	0.0496 ± 0.0096 μs	0.941
overhead/rule/noop:Int	0.0366 ± 0.0047 μs	0.0474 ± 0.015 μs	0.771
overhead/rule/noop:Sym	0.0666 ± 0.013 μs	0.0824 ± 0.022 μs	0.809
overhead/rule/noop:Term	0.0678 ± 0.014 μs	0.0773 ± 0.02 μs	0.877
overhead/ruleset/noop:Int	0.143 ± 0.02 μs	0.162 ± 0.032 μs	0.884
overhead/ruleset/noop:Sym	0.177 ± 0.025 μs	0.212 ± 0.041 μs	0.836
overhead/ruleset/noop:Term	5.11 ± 0.89 μs	5.5 ± 0.9 μs	0.93
overhead/simplify/noop:Int	0.191 ± 0.027 μs	0.238 ± 0.054 μs	0.803
overhead/simplify/noop:Sym	0.223 ± 0.029 μs	0.307 ± 0.073 μs	0.725
overhead/simplify/noop:Term	0.0731 ± 0.01 ms	0.104 ± 0.021 ms	0.702
overhead/simplify/randterm (+, *):serial	0.185 ± 0.0095 s	0.223 ± 0.012 s	0.829
overhead/simplify/randterm (+, *):thread	0.114 ± 0.015 s	0.145 ± 0.02 s	0.787
overhead/simplify/randterm (/, *):serial	0.407 ± 0.062 ms	0.54 ± 0.11 ms	0.753
overhead/simplify/randterm (/, *):thread	0.472 ± 0.064 ms	0.613 ± 0.11 ms	0.769
overhead/substitute/a	0.109 ± 0.023 ms	0.108 ± 0.019 ms	1.01
overhead/substitute/a,b	0.0927 ± 0.015 ms	0.104 ± 0.02 ms	0.888
overhead/substitute/a,b,c	26.4 ± 4.4 μs	29.6 ± 5.4 μs	0.892
polyform/easy_iszero	0.0617 ± 0.0081 ms	0.0607 ± 0.011 ms	1.02
polyform/isone	2.6 ± 0.2 ns	3 ± 0.2 ns	0.867
polyform/iszero	2.15 ± 0.36 ms	2.3 ± 0.4 ms	0.937
polyform/simplify_fractions	2.95 ± 0.43 ms	3.01 ± 0.39 ms	0.981
time_to_load	5.72 ± 0.23 s	5.59 ± 0.12 s	1.02

Benchmark Plots

A plot of the benchmark results have been uploaded as an artifact to the workflow run for this PR.
Go to "Actions"->"Benchmark a pull request"->[the most recent run]->"Artifacts" (at the bottom).

[shashi]

Fine with this, could you add a test or two?

[codecov-commenter]

Codecov Report

Merging #530 (8171356) into master (e4519eb) will not change coverage.
The diff coverage is n/a.

❗ Your organization is not using the GitHub App Integration. As a result you may experience degraded service beginning May 15th. Please install the Github App Integration for your organization. Read more.

@@           Coverage Diff           @@
##           master     #530   +/-   ##
=======================================
  Coverage   81.19%   81.19%           
=======================================
  Files          15       15           
  Lines        1856     1856           
=======================================
  Hits         1507     1507           
  Misses        349      349           

Impacted Files	Coverage Δ
src/simplify_rules.jl	93.75% <ø> (ø)

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[amilsted]

Fine with this, could you add a test or two?

Done. Also added a couple more rules (correct ones this time 😅).

[MilesCranmer]

Just curious @shashi is there an ::_iscomplex?

[shashi]

@MilesCranmer No I don't think so, but it could be added.

[amilsted]

Looks like the tests I added are passing now. I guess the segfault is a more general issue?

[MilesCranmer]

@shashi The reason I asked about the ::_iscomplex is that it looks like adding this rule would hurt the performance of simplification by about 20% across the board (#530 (comment)) even though I think some of those tests are real numbers only. So maybe having that _iscomplex check would reduce the impact of this on non-complex simplification problems?

[MilesCranmer]

(It could just be measurement uncertainty in the benchmarks though?? Not sure. Probably worth a local benchmark to be sure)