Crhmc preparation

[iakoviid]

trafficstars

Implemented the crhmc polytope preparation process into volesti.

[codecov[bot]]

Codecov Report

Merging #238 (e6700fa) into develop (2e6572d) will decrease coverage by 1.45%. The diff coverage is 42.53%.

@@             Coverage Diff             @@
##           develop     #238      +/-   ##
===========================================
- Coverage    55.24%   53.79%   -1.46%     
===========================================
  Files           85       92       +7     
  Lines         4802     5556     +754     
  Branches      2108     2573     +465     
===========================================
+ Hits          2653     2989     +336     
- Misses         896     1030     +134     
- Partials      1253     1537     +284     

Impacted Files	Coverage Δ
include/preprocess/crhmc/opts.h	0.00% <0.00%> (ø)
include/preprocess/crhmc/lewis_center.h	15.29% <15.29%> (ø)
include/preprocess/crhmc/analytic_center.h	18.18% <18.18%> (ø)
include/sos/barriers/TwoSidedBarrier.h	42.22% <42.22%> (ø)
include/preprocess/crhmc/crhmc_problem.h	44.11% <44.11%> (ø)
include/ode_solvers/oracle_functors.hpp	29.78% <50.00%> (+5.54%)	:arrow_up:
include/preprocess/crhmc/crhmc_utils.h	59.75% <59.75%> (ø)
include/preprocess/crhmc/crhmc_input.h	76.00% <76.00%> (ø)
include/generators/known_polytope_generators.h	80.00% <0.00%> (-0.59%)	:arrow_down:
include/convex_bodies/hpolytope.h	79.48% <0.00%> (+2.36%)	:arrow_up:
... and 2 more

[iakoviid]

Thank you for the guidance!! Regarding the qd library: -PackedCSparse does not use quaddouble but only doubledouble. And now that I have a used it more often I can say that I think I can draw qd out of PackedCSparse. But I have to make some heavy changes to PackedCSparse and tests the constants of the preprocessing module, maybe increasing some tolerances.

Regarding the PackedCSparse library: For the functions to implemented efficiently with Eigen we would need a similar low level treatment. And also without the use of PackedCSparse I don't know how to implement the SIMD capabilities that I explore at the moment.

[vissarion]

Thank you for the guidance!! Regarding the qd library: -PackedCSparse does not use quaddouble but only doubledouble. And now that I have a used it more often I can say that I think I can draw qd out of PackedCSparse. But I have to make some heavy changes to PackedCSparse and tests the constants of the preprocessing module, maybe increasing some tolerances.

Regarding the PackedCSparse library: For the functions to implemented efficiently with Eigen we would need a similar low level treatment. And also without the use of PackedCSparse I don't know how to implement the SIMD capabilities that I explore at the moment.

Sure, I didn't say not to use them but whether it is possible to get them using cmake on build time as with eigen and not including them in external.