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
+4-2Lines changed: 4 additions & 2 deletions
Original file line number
Diff line number
Diff line change
@@ -30,8 +30,8 @@ The computation of the log determinant with its derivatives for positive definit
30
30
31
31
1. Various kernel types, such as square-exponential, Matérn, and spectral mixture.
32
32
2. Both Gaussian and non-Gaussian likelihood.
33
-
3.High-dimensional feature space.
34
-
4.Previously prohibitive data size.
33
+
3.Previously prohibitive data size.
34
+
<!---4. High-dimensional feature space.--->
35
35
36
36
37
37
## Setup
@@ -41,6 +41,8 @@ The computation of the log determinant with its derivatives for positive definit
41
41
42
42
## Usage
43
43
44
+
The code is built with Matlab. In particular, the Lanczos algorithm with reorthogonalization uses ARPACK which comes with Matlab by default. If you are using Octave and require reorthogonalization, replace *lanczos_arpack* with *lanczos_full* in *cov/apx.m*. Fast Lanczos implementation without reorthogonalization is also available by setting *ldB2_maxit = -niter*. Please check the comments in demos for example.
0 commit comments