Closed dlcole3 closed 2 years ago
Merging #35 (2e58f2a) into main (c910182) will decrease coverage by
0.48%
. The diff coverage is84.48%
.
@@ Coverage Diff @@
## main #35 +/- ##
==========================================
- Coverage 98.04% 97.56% -0.49%
==========================================
Files 1 1
Lines 1025 1068 +43
==========================================
+ Hits 1005 1042 +37
- Misses 20 26 +6
Impacted Files | Coverage Δ | |
---|---|---|
src/DynamicNLPModels.jl | 97.56% <84.48%> (-0.49%) |
:arrow_down: |
Continue to review full report at Codecov.
Legend - Click here to learn more
Δ = absolute <relative> (impact)
,ø = not affected
,? = missing data
Powered by Codecov. Last update c910182...2e58f2a. Read the comment docs.
one last thing, jac_op
should be extended
https://github.com/JuliaSmoothOptimizers/NLPModels.jl/blob/51841a1f092bd2fad0bfd1b9c7ec74c3cc94ace0/src/nlp/api.jl#L599-L605
Added
add_jtsj!
to the source code to calculate $J^T \Sigma J$ from aLQJacobianOperator
and a vector, $\Sigma$, and adds this value to a matrix $H$. The code is similar in construct to the_build_H_blocks
function, but somewhat less efficient. This is because a new block matrix must be created and multiplied every iteration of the for loop, whereas_build_H_blocks
could reuse the same matrix several times (e.g.,_build_H_blocks
could calculate $QB$ one time and add it N times to $H$, whereasadd_jtsj!
must calculate separate $\Sigma_i B, \forall i = 1,...,N$).Also updated
LQJacobianOperator
and removed theScaled_Jac
attribute (unnecessary) and changed theJ1B
,J2B
, andJ3B
matrices toSJ1
,SJ2
, andSJ3
for storing the scaled values of the Jacobian. Also added a function toruntest.jl
to testadd_jtsj!
and a function to test themul
functions with theLQJacobianOperator
. This latter function was just to simplify the file.