[tests] t20-qfunction.c revert debug traces

[occa] okl OpenCL hack to make all the tests run

[occa] qfunction + okl and fortran __FILE__ (< 72 characters!)

[occa] operator + qfunction alloc

[t20] okl offsets

[active/passive] t20 without operator

[MFEM][NEK] Converting example BPs to active-passive

Rename CEED_QDATA_NONE to CEED_VECTOR_ACTIVE and CEED_VECTOR_NONE

fortran: ceed_restriction_identity, ceed_basis_colocated, ceed_qdata_none

style, updated ex1

added CEED_RESTRICTION_IDENTITY, CEED_BASIS_COLOCATION, CEED_QDATA_NONE

WIP passive fields

Fleshed out OperatorSetup, built comments for frame of OperatorApply. Need to finish OperatorApply and deallocate arrays in OperatorDestroy

WIP, attemped to flesh out OpApply

de

WIP passive fields

Fleshed out OperatorSetup, built comments for frame of OperatorApply. Need to finish OperatorApply and deallocate arrays in OperatorDestroy

WIP, attemped to flesh out OpApply

debugging

more debugging

modified Fortran interface, need to add new operator/qfunction functions

pointer fix

CeedElemRestrictionCreateVector added, currently assumes one comp

Dedugged OpSetup

Debugging OperatorApply

rebuilt code, error at QFuncApply

pointer errors

my pointers are a mess

malloc issues

error in OpApply, in output RestrictApply

minor fixes, still broken

both known errors identified

moved edata, indata, outdata to opref

error only when no print statement in code

possible location of mem issue IDed

Impliment OperatorApply

Fortran interface improvements started

Started rewriting Fortran tests

started Fortran conversion, ex1-vol-ann working in 1D ONLY

debugging ex1

ex1-vol-ann works

t20 fortran working, not happy with it though

cleanup

Fortran QFunc Apply with 32 arguments

t30 runs

cleanup

show more ...

Makefile: group prove-allexamples and follow patterns

* Speed up tests by exposing parallelism to prove
* Use TESTARGS in examples rather than pattern matching in tap.sh
* Use shallow clone and para

Makefile: group prove-allexamples and follow patterns

* Speed up tests by exposing parallelism to prove
* Use TESTARGS in examples rather than pattern matching in tap.sh
* Use shallow clone and parallel make to build MFEM
* Use MFEM_CXXFLAGS=-O to speed up build
* Use pattern rules to build MFEM examples in OBJDIR
* Use non-phony dependencies to avoid overzealous building

show more ...

Using main MAKEFILE, added make mfem and make prove-mfem

tests: activate backends from Makefile; can also use command line

[Opt] Add first version of opt backend, use colocated grad

Add t17-basis to test BasisApply for GRAD and TRANSPOSE

CeedQRFactorization: simplify/tighten implementation

Fix small error and t15 output.. Note that this Householder
factorization uses the usual sign convention to choose which
reflection to choose. T

CeedQRFactorization: simplify/tighten implementation

Fix small error and t15 output.. Note that this Householder
factorization uses the usual sign convention to choose which
reflection to choose. This sometimes chooses a different reflector
than the previous implementation. Also add a non-square t16-basis.

Thanks-to: Jeremy Thompson

show more ...

Add HouseholderApply function, modify arguments to QRFactorization and t15

added t16-basis-f

Update Fortran interface, add t15-basis-f

Add QR Factorization of row-major matricess, CeedBasisGetColocatedGrad, and t16

Add template backend, requested in issue #38

Split Ref into multiple files, update style

Fix a bug in the t30-operator.c test: the vector 'U' was not
initialized.