Merge pull request #1008 from RosettaCommons/vmullig/extrares_issue Tweaking 4-phenyl-phenylalanine.params to use tyrosine rotamer library. (Actually, added a second params file, with three-letter code BIF, that uses the Tyr rotamer library. B03 still uses the B03 rotamer library.) No test changes expected.

• Summary

Fixed a Rosetta@home specific issue where jobs with errors continue on rather than exiting and calling boinc_exit(). This should only affect the boinc build.

• Summary

Merge pull request #1001 from RosettaCommons/rhiju/solve_challenges Rhiju/solve challenges + the few integration test changes are expected (from introduction of BlockStack patches or changes to behavior o PartitionContactScreener in stepwise) + srlz unit tests were broken previously.

• Summary

Merge pull request #1004 from RosettaCommons/aleaverfay/remove_multithread_jd2_job_dist Scrapping attempted multithreaded job distributor for JD2 JD2 cannot accommodate a multithreaded job distributor. This feature will come online with JD3. Hopefully, this fixes the multithreaded unit tests.

• Summary

Merge pull request #1006 from RosettaCommons/kalekundert/explicit-python2 Explicitly call python2 when necessary.

• Summary

Merge pull request #1005 from RosettaCommons/rfalford12/revise_memb_etable Similar to the LK term for soluble proteins, the Lazaridis membrane solvation energy relies on an Etable. The membrane etable, called MembEtable, was originally implemented as a copy of Etable, creating several unused variables and redundant declarations of constants (amonst other challenges). I'm working on a new membrane solvation energy that requires modifying this class, so now is a good time for this refactoring. In this version, MembEtable is a subclass of Etable (previously, MembEtable inherited from Etable but never took advantage of this hierarchy). Below is a summary of the adjustments: No duplicate private members (constants) between Etable and MembEtable Private members of Etable are accessed in MembEtable through getters (added necessary getters) Customizable functions (e.g. solv1, dsolv1) are now virtual in Etable Removed extraneous comments, commented out blocks of code I also cleaned up the Etable - several of the private member variables were named without a trailing underscore which collides when you declare an accessor for it. Unit & Integration Test Changes: = Cpp Check errors complain 3D Farrays not initialized in the constructor, really they are just not initialized in the initialization list but in the body of the constructor = Integration & profile test failures existed before this pull request

• Summary

Merge pull request #998 from RosettaCommons/tlinsky/tomponents Update to latest denovo classes

• Summary

Merge pull request #999 from RosettaCommons/rhiju/make_new_move_selector_default Rhiju/make new_move_selector default for stepwise + Includes several (but not quite all) bugfixes for -new_move_selector running with `-submotif_moves`. will put that in a separate pull request, if we think it should be default. + integration tests that change are swm (stepwise monte carlo). + saw a broken windows pyrosetta build on test server (Haraku), but then requeued and it compiled fine.

• Summary

Add a new pilot app dna_motif_matcher

• Summary

Updating tests states in benchmark scripts

• Summary

Merge pull request #980 from RosettaCommons/rhiju/unit_test_mania Rhiju/unit test mania redux Finally getting all these unit tests commited! No changes to regression tests except unstable ones; the .srlz/cxx11thread build was broken before.

• Summary

Merge pull request #995 from RosettaCommons/vmullig/iterated_linmin Add a simple gradient-descent minimization algorithm The Hessian approximations in Rosetta really trip me up when I'm trying to use the minimizer for anything other than minimizing energy in torsion space. This pull request adds a simple Hessian-free iterative gradient descent algorithm for the minimizer (basically, iterated linmin calls with a convergence check after each iteration). Simple gradient descent (first-derivative based minimization) is generally less efficient than approaches that use the Hessian or an approximation thereof (first- and second-derivative based minimization), but there is no guarantee that the DFP or LBFGS approximations of the Hessian matrix work for all functions -- indeed they are poor approximations in some cases that mathematicians have characterized but the rest of us don't think about. Having the option of doing plain, vanilla gradient descent is therefore useful, at least for checking whether it's the Hessian approximation that's tripping up a minimization protocol. Tasks: - Add linmin_iterated and linmin_iterated_atol minimization flavours. - Add a function to do iterated gradient descent with convergence check. - Add a unit test. - Update documentation. - Add new minimization type to options_rosetta.py in branch aleaverfay/lbfgs_as_default (pull request #327). I'd suggest that we start using linmin_iterated or linmin_iterated_atol in minimizer and FastRelax benchmarks, sort of as our negative control (minimization with no Hessian approximation). It SHOULD converge more slowly (i.e. with more iterations) than dfpmin or lbfgs. If anything else converges more slowly, though, that should be a big red flag for that other algorithm.

• Summary

Merge pull request #992 from RosettaCommons/rhiju/fix_stepwise_packer_interface_res_determination interface residues in StepWisePacker swa/swm integration tests are expected -- slightly worse energies due to residues far from interface not getting packed.

• Summary

Merge pull request #975 from RosettaCommons/rhiju/check_new_move_selector Rhiju/check new move selector + integration test for swm protein loop sampler changes, apparently due to change in random number generated. other tests are 'usual' ones that change in different hardware setups. + cppcheck and linux .srlz.unit remain broken; former was just fixed by andrew i think.

• Summary

Merge pull request #993 from RosettaCommons/aleaverfay/make_assignment_ops_return_nonconst_refs Fixing ~60 assignment operators that returned const-references Assignment operators in c++ are, by convention, supposed to return non-const references to the underlying type. All of the assignment operators I have written up until now have returned const-references, however. My bad. This is, to my surprise, a bad thing. It violates the "principle of least surprise" (not to be confused with the principle of least surprise for Andrew) in making a class behave differently from a primitive. In C++ it's legal to take two ints, x and y and do this: ( x = y )++; This commit will fix the broken cppcheck test, and removes many complaints (that don't register as failures) that cppcheck makes about our assignment operators.

• Summary

Merge pull request #983 from RosettaCommons/aleaverfay/document_xsd_classes Adding promised documentation to the XML-schema-generation classes. Also documenting the ResidueSelector XSD generation code. Finally, removing "return" statements from several residue-selector-creator's::provide_selector_xsd functions that were legacy from an early implementation.

• Summary

Legend: