Commit Graph
64 Commits
Author SHA1 Message Date
FredApps a60db4f1c7 Merge pull request #3323 from vangel76/fix/linest-qr
LINEST, LOGEST, TREND, GROWTH: solve by Householder QR (#3322)
2026-11-12 17:59:31 +01:00
vangel76andClaude Opus 5.5 86ed3d93b4 LINEST: treat only identical raw values as a constant predictor
The 64-ulp spread threshold still classified distinct representable
predictors such as [1e15, 1e15 + 1, 1e15 + 2] as constant and dropped
them. A column is now constant only when every raw value is the same
double; the 1e15 case is added to the offset regression test.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-08 12:19:34 +01:00
vangel76andClaude Sonnet 5.5 ed15db4101 Merge upstream/main into fix/linest-qr
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-11-07 18:37:48 +01:00
vangel76andClaude Sonnet 5.5 8e5bcf1575 Merge upstream/main into fix/binom-poisson-accuracy
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-11-07 18:37:33 +01:00
vangel76andClaude Sonnet 5.5 12f69f99bf BINOM.DIST: keep small-k masses whose complement power underflows
The small-k product also needs (1 - p)^(n - k) to stay a normal number;
otherwise it returned 0 for BINOM.DIST(15,1100,0.5,FALSE) (exact
2.1367646606291324e-298) and the cumulative form built on it. Those cases
now use the logarithmic formula. Reference grid gets k = 10..25 at n = 1100.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-11-07 17:20:53 +01:00
vangel76andClaude Sonnet 5.5 583c8a4819 LINEST: keep predictors that vary by little next to a large offset
A centred column was dropped when its length fell under 1e-12 of the raw
column, which also removed real variation such as [1e13, 1e13+1, 1e13+2].
Constant columns are now recognised on the raw values, within a few ulps.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-11-07 17:01:20 +01:00
FredApps cbd765332c Merge pull request #2686 from vangel76/perf/stat-select-2679
perf: select instead of sort in MEDIAN, PERCENTILE, QUARTILE, LARGE, SMALL
2026-11-07 16:51:02 +01:00
vangel76andClaude Sonnet 5.5 756d5981b1 Merge upstream/main into fix/binom-poisson-accuracy
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-11-05 01:41:42 +01:00
Dylan M. Taylor 065f05dbec Security: MODE counts through an index, FREQUENCY bins by binary search (#3198)
* functions: MODE counts through an index, FREQUENCY bins by binary search

MODE/MODE.SNGL/MODE.MULT searched the list of distinct numbers seen so far
for every input, and FREQUENCY scanned the sorted bins for every data value,
so both were quadratic in computed arrays. MODE now keys counts by value
(-0 folded into 0, NaN never merged) and keeps first-occurrence order;
FREQUENCY finds the first bin at or above each value with partition_point
over the order it already sorts. Results are unchanged.

* address review: FREQUENCY keeps NaN data in the overflow bin and NaN bins empty
2026-11-03 22:51:58 -05:00
vangel76andClaude Sonnet 5.5 a5c74edfec linest test: use f64::from directly (clippy)
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-11-04 00:18:00 +01:00
vangel76andClaude Sonnet 5.5 40e75c7912 binom/poisson reference test: fail on errors, NaN and non-zero results for zero references (review point)
CodeRabbit pointed out that an error or NaN result passed the accuracy test and
that a zero reference allowed any finite value (the bound was infinite). Errors
now count as infinite error and a zero reference allows 1e-300.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-11-04 00:05:58 +01:00
vangel76andClaude Sonnet 5.5 d0b40c3b30 linest: drop a constant column that centres to rounding noise (review point)
CodeRabbit pointed out that the drop test used the centred column length, so a
constant column such as [0.1; 3] (centred to about -1.4e-17) was scaled to a unit
vector and kept, giving a meaningless coefficient and wrong df. A centred column
shorter than 1e-12 of its uncentred length is now treated as collinear with the
intercept. The suggested 1e-6 threshold was not used: it would also drop columns
that sit far from zero, such as 1e9 + i.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-11-04 00:05:38 +01:00
vangel76andClaude Sonnet 5.5 7e8b94fc39 BINOM.DIST: keep tiny masses and huge n out of the product form
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-11-03 23:54:53 +01:00
vangel76andClaude Sonnet 5.5 22c06b7a4a BINOM.DIST, POISSON.DIST: Loader mass, recurrence sums, exact prefactors (#3324)
The mass came from ln_gamma differences (relative error 1.1e-13 at n = 100,
1.5e-11 at n = 10,000) and the cumulative forms summed up to 10,000 of them.
The mass now comes from Loader's saddle-point formula, the cumulative sum
walks away from the mode by the ratio of neighbouring terms and stops when
terms stop mattering, and the incomplete beta / gamma branches take the same
mass as their prefactor.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-11-03 23:49:52 +01:00
vangel76andClaude Sonnet 5.5 f74d95542f LINEST: solve by Householder QR instead of the normal equations (#3322)
The normal equations square the condition number, so a degree-5 polynomial on
x = 1..12 came back with errors up to 2.6e-6 and x, x^2 near 1e6 gave slope
2.99976 and intercept 251 for y = 3x + 7. LINEST, LOGEST, TREND and GROWTH now
factor the centred, column-scaled data with Householder QR. Collinear columns
are still dropped under the same relative tolerance.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-11-03 23:38:55 +01:00
Brandon Thomas 59fd17ae56 Merge PR #2980
# Conflicts:
#	docs/function-parity.md
2026-10-31 10:17:09 -04:00
Brandon Thomas 0db7d97f2e Merge PR #3008
# Conflicts:
#	docs/function-parity.md
2026-10-30 23:06:47 -04:00
Brandon Thomas 4d1233d0a2 Merge PR #3007
# Conflicts:
#	docs/function-parity.md
2026-10-30 23:06:47 -04:00
Brandon Thomas 07c8201786 Merge PR #3006
# Conflicts:
#	docs/function-parity.md
2026-10-30 23:06:47 -04:00
Brandon Thomas 31d5cc5b41 Merge PR #3003
# Conflicts:
#	crates/functions/src/stat/tests.rs
#	docs/function-parity.md
2026-10-30 23:06:47 -04:00
Brandon Thomas c0fcc7de16 Merge PR #3002
# Conflicts:
#	docs/function-parity.md
2026-10-30 23:06:47 -04:00
Brandon Thomas 5390e230d8 Merge PR #3001
# Conflicts:
#	crates/functions/src/stat/tests.rs
#	docs/function-parity.md
2026-10-30 23:06:46 -04:00
Brandon Thomas c9eb797d8b Merge PR #3000
# Conflicts:
#	crates/functions/src/stat/tests.rs
#	docs/function-parity.md
2026-10-30 23:06:46 -04:00
Brandon Thomas 5441691ca3 Merge PR #2999
# Conflicts:
#	crates/functions/src/stat/tests.rs
#	docs/function-parity.md
2026-10-30 23:04:26 -04:00
Harrison Booth 006b7962bd Preserve F distribution tail accuracy 2026-10-30 21:14:19 -05:00
Harrison Booth 7ea3a4c2c3 Merge upstream main into harmonic mean branch 2026-10-30 20:59:56 -05:00
Harrison Booth cae0c9abdb Merge main into function updates 2026-10-30 20:59:41 -05:00
Harrison Booth ed81d7e34e Merge upstream main into binomial distribution branch 2026-10-30 20:59:39 -05:00
Harrison Booth 8f7dff0d04 Merge main into function updates 2026-10-30 20:59:31 -05:00
Harrison Booth 7b7e734524 Merge main into function updates 2026-10-30 20:59:20 -05:00
Harrison Booth 6d5ab57f84 Merge upstream main into Weibull distribution branch 2026-10-30 20:59:18 -05:00
Harrison Booth a04629576d Merge main into function updates 2026-10-30 20:59:12 -05:00
Harrison Booth 322298b202 Merge upstream main into gamma distribution branch 2026-10-30 20:58:48 -05:00
Harrison Booth 45939a8a41 Merge remote-tracking branch 'upstream/main' into gc-build-g718x
# Conflicts:
#	docs/function-parity.md
2026-10-30 20:56:57 -05:00
Harrison Booth 70da8bb3ae Cover strict upper tail in F inverse 2026-10-30 20:53:15 -05:00
Harrison Booth a0da53f576 Match T DIST zero degree errors by mode 2026-10-30 20:53:00 -05:00
Harrison Booth ca44a5d68f Avoid overflow in t density normalization 2026-10-30 20:52:05 -05:00
Harrison Booth 3b61d660e9 Return NUM for invalid T distribution degrees 2026-10-30 20:51:47 -05:00
Harrison Booth cca8a58bfa Stabilize F distribution CDF at extreme degrees 2026-10-30 20:51:07 -05:00
Harrison Booth d3cd393dba G719: return NA for empty harmonic mean input 2026-10-30 20:25:09 -05:00
Harrison Booth 8985685833 G718: match beta density endpoint errors 2026-10-30 20:24:21 -05:00
Harrison Booth 962a5df841 G717: cap binomial trial counts 2026-10-30 20:23:32 -05:00
Harrison Booth fea4c03ca2 G709: use stable tail at F median 2026-10-30 20:18:56 -05:00
Harrison Booth 1d9096e5ec G709: stabilize F upper tail at extreme degrees 2026-10-30 20:05:47 -05:00
Harrison Booth a9bacdca4c G711: keep T.DIST validation isolated 2026-10-30 19:56:36 -05:00
Harrison Booth 6aa2cb8223 G709: invert upper F probabilities through complementary tail 2026-10-30 19:49:32 -05:00
Harrison Booth 1b1224c67a G713: simplify zero-shape boundary 2026-10-30 19:42:50 -05:00
Harrison Booth 4e5c86d842 G711: preserve small density term at huge degrees 2026-10-30 19:39:05 -05:00
Harrison Booth ad69d4cbdd G710: update zero-degree test error 2026-10-30 19:39:00 -05:00
Harrison Booth 48c6b57060 G714: return zero Weibull density at zero 2026-10-30 19:34:59 -05:00