Failure to compute quantile of InverseGaussian

v0.25.62 according to ]status Distributions. The quantile function seems to fail for seemingly arbitrary numbers, e.g.

μ   = 1.0
λ   = 0.25
quantile(InverseGaussian(μ,λ),0.99)
quantile(InverseGaussian(μ,λ),0.999)
quantile(InverseGaussian(μ,λ),0.9999)
quantile(InverseGaussian(μ,λ),0.99999)
quantile(InverseGaussian(μ,λ),0.999999)

All work successfully. But

quantile(InverseGaussian(μ,λ),0.999996)

leads to no result (it just gets stuck trying to compute it). 0.999995 and 0.999997 work instantly, too.

Newton's method is entering an oscillation:

0.46351854580739754
1.1822190665191383
2.5171241967844167
4.663152461244044
7.681537906525332
11.504627262164904
16.00183230199545
21.035352263873737
26.481726330380177
32.2266130798376
38.135513473013404
43.98404551213666
49.3182977047776
53.31471534484445
55.194990126949854
55.52269267456251
55.53113502401951
55.53114044440781
55.53114044459374
55.53114044440781
55.53114044459374
55.53114044440781
...

- `quantile_bisect` now uses isapprox to check whether lx and rx are …

Use Roots defaults with a bracketing fallback for quantile root-finding

Simplify quantile root-finding to Roots defaults (#2070)

Fixed by #2070

Navigation Menu

Repository navigation

Failure to compute quantile of InverseGaussian #1571

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Failure to compute quantile of InverseGaussian #1571

Description

Activity

simsurace commented on Jun 26, 2022

devmotion commented on Jun 25, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions