-
For sparse symmetric 10Kx10K A and B (B is diagonal), generalized problem is solved for the 15 smallest eigenvalues via
+ SymGEigsSolver in 20 secs
+ converting A v = lambda B v into the standard …
-
From Ben Goodrich on stan-dev:
OK. I invented a new rule (locally): You are allowed to assign anything to a matrix with 0 rows and 0 columns. So, this now works:
```
model {
matrix[0,0] X[2];
ve…
-
To not depend on external libraries and to provide correct, fast, and simple examples of LA usage with quantities we need simple vector and tensor types (up to 3 dimensions) together with the followin…
mpusz updated
3 months ago
-
The quantities cvdrift and gbdrift output by the vmec_fieldlines routine must always satisfy:
cvdrift >= gbdrift
These quantities are defined in Dr. Edmund Highcock's thesis in section 3.7 [here](ht…
-
Spatial convergence tests throw an unknown exception during the Poisson pre-step of the second cycle. Furthermore `GuermondNeumannBC.cc` also fails in the Poisson pre-step but in the first cycle of th…
-
Currently, BFPack uses a few linear algebra primitives including:
* `FINDinv` (inverse of square matrix)
* `spofa` (factorization of a real symmetric positive definite matrix)
These can be repl…
-
##### at 2006-01-31 11:11:09 Jan Noordam reported:
Implement improved LM solver strategy suggested by WNB
-
Erika kindly extracted the data into a usable format. Might do some quick editing/culling first:
```
English,Français
collinearity,colinéarité
combinatorial method,méthode combinatoire
comparis…
-
We should build a method to allow users to find the top eigenvector of a linear operator defined by a tensor network.
My API proposal is as such
```
tensor = tn.eigensolution(in_edges=list_of_e…
-
benchmarks/newton_solver_benchmark_set/spiegelman_et_al_2016 diverges
```
*** Timestep 1: t=5.86784e+12 seconds, dt=5.86784e+12 seconds
Skipping temperature solve because RHS is zero.
Sol…