You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: README.md
+1-39Lines changed: 1 addition & 39 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -29,39 +29,8 @@ Currently, it supports the following calculations:
29
29
30
30
# Installation
31
31
32
-
## Runtime Dependencies
32
+
## Install from PyPI
33
33
34
-
The only Python runtime dependency is
35
-
[numpy](https://numpy.org/). It will be automatically installed as the requirements.
36
-
Moreover, the library optionally depends on
37
-
[Intel Math Kernel Library (mkl)](https://software.intel.com/content/www/us/en/develop/tools/oneapi/components/onemkl.html),
38
-
for its [PARDISO](https://www.intel.com/content/www/us/en/develop/documentation/onemkl-developer-reference-c/top/sparse-solver-routines/onemkl-pardiso-parallel-direct-sparse-solver-iface.html) sparse solver.
39
-
It is recommended to install `mkl` because it gives huge performance boosts.
40
-
41
-
The easiest way to install `mkl` is using `pip` or `conda`:
42
-
43
-
```
44
-
pip install mkl
45
-
```
46
-
47
-
or
48
-
49
-
```
50
-
conda install -c conda-forge mkl
51
-
```
52
-
53
-
You need to add the path to the `mkl` runtime file `libmkl_rt.so` or `mkl_rt.dll` the environment variable
54
-
`LD_LIBRARY_PATH` in Linux or `Path` in Windows (`conda` does this automatically in the environment).
55
-
If the library can find `mkl` runtime, it uses it as the sparse solver.
56
-
It is recommended to set the environment variable `MKL_THREADING_LAYER` to `SEQUENTIAL`,
57
-
as multi-threading is handled in a higher level.
58
-
If the library cannot find `mkl` runtime, it will fall back to an internally built-in (and much slower)
0 commit comments