Adding the first implicit GLEE method based on extrapolated backward Euler

TS,TSGLEE: declared the TSGetAuxSolution in header file, and fixed its implementation for TSGLEE

TSGLEE: moved out the TSGLEE implementation from the explicit directory since it's not purely explicit anymore

TS, TSGLEE: created an interface for a "starting method". Set it to a default starting method for TSGLEE.

TS, TSGLEE: added interface for getting "auxiliary" solutions, which in the case of TSGLEE, is the y~ or the eps

TSGLEE: moved the "starting method" (initialization of auxiliary solution) to TSSetUp_GLEE, instead of the hack-y way in which it was placed in TSStep_GLEE

TSGLEE: extended the implementation of TSGLEE to diagonally-implicit GLEE methods

TSGLEE: Implemented the General Linear methods with Error Estimation

TSPostEvaluate: TSPostEvaluate() should also be called when the step is rolled back.

TSPostEvaluate: Support for setting a custom post evaluate function (Only added to TSRK for now)

TSPostEvaluate can be used for modifying the solution after the step evaluation. Having TSPostEvaluat

TSPostEvaluate: Support for setting a custom post evaluate function (Only added to TSRK for now)

TSPostEvaluate can be used for modifying the solution after the step evaluation. Having TSPostEvaluate enables
the user to solve DAEs with PETSc's explicit methods using a partitioned approach in which the differential
variables are updated by the time-stepping solver while the algebraic variables are updated through user
callbacks in TSPostStage() and TSPostEvaluate(). Another use of TSPostEvaluate is to modify the solution such as
via some filtering. Typically such filtering operations (from Emil) are done in TSPostStep(). However, TSEvent can
modify the solution if it detects an event thus making the action taken in TSPostStep() invalid. So, instead of
doing the filtering in TSPostStep(), the users should rather do it in TSPostEvaluate().

show more ...

Refactored ./configures handling of package requirements for precision. Now each package indicates
what it supports with
self.precisions = ['single','double','__float128']
Updated the package files t

Refactored ./configures handling of package requirements for precision. Now each package indicates
what it supports with
self.precisions = ['single','double','__float128']
Updated the package files to indicate what they support
Removed unneeded #requires arguments from makefiles for external packages.

Time: 2 hours
Reported-by: Patrick Sanan <patrick.sanan@gmail.com>

show more ...

fix PetscFormatRealArray() to cast PetscReal in print statement to double so works for __float128 precision

Since __float128 results in identically zero KSP residual norm thus producing an KSP conve

fix PetscFormatRealArray() to cast PetscReal in print statement to double so works for __float128 precision

Since __float128 results in identically zero KSP residual norm thus producing an KSP convergence reason
of ATOL instead of the RTOL for double precision

Reported-by: nightly tests

show more ...

TSEIMEX: doc fix

TSGL rename source files and folders to glle

TS: Changed TSGL to TSGLLE because TSGL is too general and we want to emphasize that TSGLLE is a type of GL with local error estimation.

Merge branch 'maint'

always need to cast PetscReal variables in print statements with double

otherwise garbage is printed for quad precision
Reported-by: nightly tests

Merge branch 'barry/housekeeping-2'

adding missing dashes for help menu
Contributed-by: Patrick Sanan <patrick.sanan@gmail.com>

replace uses of PetscMalloc() with PetscNew() or PetscMalloc1() where appropriate