Nix: init

[nim65s]

No description provided.

[nim65s]

>>> for dim in range(2, 101):
...     A = rng.random((dim, dim))
...     A = (A + A.T) * 0.5 + np.diag(10.0 + rng.random(dim))
...     print(dim, nanoeigenpy.BDCSVD(A, options).info())
...     
2 ComputationInfo.Success
3 ComputationInfo.Success
4 ComputationInfo.Success
5 ComputationInfo.Success
6 ComputationInfo.Success
7 ComputationInfo.Success
8 ComputationInfo.Success
9 ComputationInfo.Success
10 ComputationInfo.Success
11 ComputationInfo.Success
12 ComputationInfo.Success
13 ComputationInfo.Success
14 ComputationInfo.Success
15 ComputationInfo.Success
16 ComputationInfo.NumericalIssue
17 ComputationInfo.NumericalIssue
18 ComputationInfo.NumericalIssue
19 ComputationInfo.NumericalIssue
20 ComputationInfo.NumericalIssue
21 ComputationInfo.NumericalIssue
22 ComputationInfo.NumericalIssue
23 ComputationInfo.NumericalIssue
24 ComputationInfo.NumericalIssue
25 ComputationInfo.NumericalIssue
26 ComputationInfo.NumericalIssue
27 ComputationInfo.NumericalIssue
28 ComputationInfo.NumericalIssue
29 ComputationInfo.NumericalIssue
30 ComputationInfo.NumericalIssue
31 ComputationInfo.NumericalIssue
32 ComputationInfo.NumericalIssue
33 ComputationInfo.NumericalIssue
34 ComputationInfo.NumericalIssue
35 ComputationInfo.NumericalIssue
36 ComputationInfo.NumericalIssue
37 ComputationInfo.NumericalIssue
38 ComputationInfo.NumericalIssue
39 ComputationInfo.NumericalIssue
40 ComputationInfo.NumericalIssue
41 ComputationInfo.NumericalIssue
42 ComputationInfo.NumericalIssue
43 ComputationInfo.NumericalIssue
44 ComputationInfo.NumericalIssue
45 ComputationInfo.NumericalIssue
46 ComputationInfo.NumericalIssue
47 ComputationInfo.NumericalIssue
48 ComputationInfo.NumericalIssue
49 ComputationInfo.NumericalIssue
50 ComputationInfo.NumericalIssue
51 ComputationInfo.NumericalIssue
52 ComputationInfo.NumericalIssue
53 ComputationInfo.NumericalIssue
54 ComputationInfo.NumericalIssue
55 ComputationInfo.NumericalIssue
56 ComputationInfo.NumericalIssue
57 ComputationInfo.NumericalIssue
58 ComputationInfo.NumericalIssue
59 ComputationInfo.NumericalIssue
60 ComputationInfo.NumericalIssue
61 ComputationInfo.NumericalIssue
62 ComputationInfo.NumericalIssue
63 ComputationInfo.NumericalIssue
64 ComputationInfo.NumericalIssue
65 ComputationInfo.NumericalIssue
66 ComputationInfo.NumericalIssue
67 ComputationInfo.NumericalIssue
68 ComputationInfo.NumericalIssue
69 ComputationInfo.NumericalIssue
70 ComputationInfo.NumericalIssue
71 ComputationInfo.NumericalIssue
72 ComputationInfo.NumericalIssue
73 ComputationInfo.NumericalIssue
74 ComputationInfo.NumericalIssue
75 ComputationInfo.NumericalIssue
76 ComputationInfo.NumericalIssue
77 ComputationInfo.NumericalIssue
78 ComputationInfo.NumericalIssue
79 ComputationInfo.NumericalIssue
80 ComputationInfo.NumericalIssue
81 ComputationInfo.NumericalIssue
82 ComputationInfo.NumericalIssue
83 ComputationInfo.NumericalIssue
84 ComputationInfo.NumericalIssue
85 ComputationInfo.NumericalIssue
86 ComputationInfo.NumericalIssue
87 ComputationInfo.NumericalIssue
88 ComputationInfo.NumericalIssue
89 ComputationInfo.NumericalIssue
90 ComputationInfo.NumericalIssue
91 ComputationInfo.NumericalIssue
92 ComputationInfo.NumericalIssue
93 ComputationInfo.NumericalIssue
94 ComputationInfo.NumericalIssue
95 ComputationInfo.NumericalIssue
96 ComputationInfo.NumericalIssue
97 ComputationInfo.NumericalIssue
98 ComputationInfo.NumericalIssue
99 ComputationInfo.NumericalIssue
100 ComputationInfo.NumericalIssue

[nim65s]

Ok, the issue comes from eigen version. The bug occur on e7248b26a1ed53fa030c5c459f7ea095dfd276ac but not v3.4.0.

Tried with eigen master, but:

python3.13-nanoeigenpy> /build/source/include/nanoeigenpy/decompositions/sparse/sparse-lu.hpp:67:23: error: ‘MappedSparseMatrix’ in namespace ‘Eigen’ does not name a template type; did you mean ‘SparseMatrix’?
python3.13-nanoeigenpy>    67 |       typename Eigen::MappedSparseMatrix<Scalar, Eigen::ColMajor, StorageIndex>;
python3.13-nanoeigenpy>       |                       ^~~~~~~~~~~~~~~~~~
python3.13-nanoeigenpy>       |                       SparseMatrix
python3.13-nanoeigenpy> /build/source/include/nanoeigenpy/decompositions/sparse/sparse-lu.hpp:69:60: error: ‘MappedSparseMatrix’ was not declared in this scope

I'll bissect.

[ManifoldFR]

Should this be merged? It looks alright to me, so I'll approve in the meantime.

[nim65s]

Hum, maybe now that NixOS/nixpkgs#431482 has been merged into stagging, we should wait for it to be available on nixos-unstable, and remove the patch from here: https://nixpkgs-tracker.ocfox.me/?pr=431482