Improve power of complex intervals#745
Improve power of complex intervals#745OlivierHnt wants to merge 4 commits intoJuliaIntervals:masterfrom
Conversation
Benchmark Results (Julia v1)Time benchmarks
Memory benchmarks
|
|
This looks good to me. I would just add
|
|
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## master #745 +/- ##
==========================================
- Coverage 76.37% 76.18% -0.19%
==========================================
Files 31 31
Lines 2988 3019 +31
==========================================
+ Hits 2282 2300 +18
- Misses 706 719 +13 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
|
I could not simply have a test that Instead, I check that we do have an improved real part: x = interval(-0.5) + interval(im) * interval(-1e-14, 1e-14)
y = interval(1.5, 2.5)
res = x^y
ref = exp(y*log(x))
# tighter on the real part (right half-plane)
@test inf(real(ref)) < -0.3 < -1e-13 < inf(real(res))Let me know if you have any ideas for tests before we merge the PR. |
Since we use the formula
exp(y * log(x))to evaluatex^ywhenx::Complex{<:Interval},y::Complex{<:Interval}, we get the following undesirable overestimation:yet
This PR implements the following definition for complex powers:
We compute
We compute
With this PR