Closed thomas-dresselhaus closed 3 years ago
Need to add
#include <cassert>
in gaussquad.cpp to fix build failure. You can either do this locally and update this pull request (preferred to keep merges granular), or I can include it in my next update - currently working on issue #21 has delayed things slightly
Need to add
#include <cassert>
in gaussquad.cpp to fix build failure. You can either do this locally and update this pull request (preferred to keep merges granular), or I can include it in my next update - currently working on issue #21 has delayed things slightly
Done. Sorry about that.
Merging #22 (59d0909) into master (fa8df43) will decrease coverage by
0.44%
. The diff coverage is87.80%
.
@@ Coverage Diff @@
## master #22 +/- ##
==========================================
- Coverage 90.07% 89.62% -0.45%
==========================================
Files 19 19
Lines 1964 1956 -8
==========================================
- Hits 1769 1753 -16
- Misses 195 203 +8
Impacted Files | Coverage Δ | |
---|---|---|
include/libecpint/ecpint.hpp | 100.00% <ø> (ø) |
|
include/libecpint/gaussquad.hpp | 100.00% <ø> (ø) |
|
src/lib/radial_quad.cpp | 82.27% <77.27%> (+1.14%) |
:arrow_up: |
src/lib/bessel.cpp | 97.95% <100.00%> (ø) |
|
src/lib/ecpint.cpp | 100.00% <100.00%> (ø) |
|
src/lib/gaussquad.cpp | 89.79% <100.00%> (-10.21%) |
:arrow_down: |
src/lib/radial_gen.cpp | 81.46% <100.00%> (-0.08%) |
:arrow_down: |
include/libecpint/multiarr.hpp | 95.34% <0.00%> (-0.11%) |
: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 fa8df43...59d0909. Read the comment docs.
This PR essentially marks the functions
compute_shell_pair
and related functions for derivatives asconst
. As a result, no members of the involved classes change during these function calls thus making these functions thread-safe.This enables codes using libecpint to instantiate
ECPIntegral
only once and use it in all threads, whereas before one instance per thread was required.Quite a number of changes were required to achieve the above. A small number of these changes may in principle affect performance, so it should be tested that performance is not degraded by the contents of this PR.