Add IREGI

recipes/iregi/makefile_ML5_lib.patch

@@ -0,0 +1,56 @@
+From e19ac63f03c4e5103f1456803b5e1db8169b9040 Mon Sep 17 00:00:00 2001
+From: Matthew Feickert <matthew.feickert@cern.ch>
+Date: Sun, 9 Feb 2025 03:10:29 -0700
+Subject: [PATCH] fix: Make makefile_ML5_lib compatible with precompiled
+ libraries
+
+* Let makefile_ML5_lib build IREGI standalone against existing
+  compiled versions of the libraries.
+---
+ vendor/IREGI/src/makefile_ML5_lib | 17 +++--------------
+ 1 file changed, 3 insertions(+), 14 deletions(-)
+
+diff --git a/vendor/IREGI/src/makefile_ML5_lib b/vendor/IREGI/src/makefile_ML5_lib
+index fff9dc8..7319db6 100644
+--- a/vendor/IREGI/src/makefile_ML5_lib
++++ b/vendor/IREGI/src/makefile_ML5_lib
+@@ -6,14 +6,9 @@ FFDIR=$(qcdlooppath)/ff
+ onelooppath=$(PWD)/oneloop
+ #FFLAGS=-O -g -ggdb -fno-automatic -fbounds-check -I$(onelooppath) # -fbounds-check is very useful in checking mermory error,-g is for debuging,-ggdb is for debuging special in gdb
+ FFLAGS=-O -g -fPIC -I$(onelooppath)
+-obj= global.o funlib.o linear_algebra.o kinematics.o matrices.o matrix_base.o cmatrix_base.o mis_warp.o special_fun.o pave_reduce.o cpave_reduce.o si_reduce.o csi_reduce.o ti_reduce.o cti_reduce.o gti_reduce.o binary_tree.o IREGI4ML5_interface.o avh_olo_foriregi.o
+-olomod= avh_olo_foriregi.mod avh_olo_foriregi_dp_box.mod    avh_olo_foriregi_dp_olog.mod   avh_olo_foriregi_units.mod \
+-        avh_olo_foriregi_dp.mod        avh_olo_foriregi_dp_boxc.mod   avh_olo_foriregi_dp_prec.mod   avh_olo_foriregi_version.mod \
+-        avh_olo_foriregi_dp_arrays.mod avh_olo_foriregi_dp_bub.mod    avh_olo_foriregi_dp_print.mod \
+-        avh_olo_foriregi_dp_auxfun.mod avh_olo_foriregi_dp_dilog.mod  avh_olo_foriregi_dp_qmplx.mod \
+-        avh_olo_foriregi_dp_bnlog.mod  avh_olo_foriregi_dp_kinds.mod  avh_olo_foriregi_dp_tri.mod
+-mod= global.mod funlib.mod linear_algebra.mod kinematics.mod matrices.mod matrix_base.mod cmatrix_base.mod mis_warp.mod special_fun.mod pave_reduce.mod cpave_reduce.mod si_reduce.mod csi_reduce.mod ti_reduce.mod cti_reduce.mod gti_reduce.mod $(olomod)
+-libiregi.a: make_ql $(obj) $(mod)
++obj= global.o funlib.o linear_algebra.o kinematics.o matrices.o matrix_base.o cmatrix_base.o mis_warp.o special_fun.o pave_reduce.o cpave_reduce.o si_reduce.o csi_reduce.o ti_reduce.o cti_reduce.o gti_reduce.o binary_tree.o IREGI4ML5_interface.o
++mod= global.mod funlib.mod linear_algebra.mod kinematics.mod matrices.mod matrix_base.mod cmatrix_base.mod mis_warp.mod special_fun.mod pave_reduce.mod cpave_reduce.mod si_reduce.mod csi_reduce.mod ti_reduce.mod cti_reduce.mod gti_reduce.mod
++libiregi.a: $(obj) $(mod)
+ 	@rm -f libiregi.a
+ 	@ar csr libiregi.a  $(obj)
+ #$(QLDIR)libqcdloop.a $(FFDIR)libff.a $(onelooppath)libavh_olo.a
+@@ -39,8 +34,6 @@ matrix_base.o matrix_base.mod : matrix_base.f90
+ 	$(FC) -c $< $(FFLAGS)
+ cmatrix_base.o cmatrix_base.mod : cmatrix_base.f90
+ 	$(FC) -c $< $(FFLAGS)
+-avh_olo_foriregi.mod : oneloop/avh_olo_foriregi.f90
+-	$(FC) -c $< $(FFLAGS)
+ mis_warp.o mis_warp.mod : mis_warp.f90 kinematics.mod global.mod $(olomod)
+ 	$(FC) -c $< $(FFLAGS)
+ special_fun.o special_fun.mod : special_fun.f90
+@@ -64,9 +57,5 @@ binary_tree.o binary_tree.mod : binary_tree.f90 global.mod
+ IREGI4ML5_interface.o : IREGI4ML5_interface.f90 global.mod funlib.mod ti_reduce.mod binary_tree.mod cti_reduce.mod
+ 	$(FC) -c $< $(FFLAGS)
+
+-make_ql:
+-	(make -f makefile_qcdloop)
+-
+ clean:
+ 	@rm -rf $(obj) $(exe) *.mod *~ *.a
+-	@(make clean -f makefile_qcdloop)
+-- 
+2.47.1
+

recipes/iregi/meta.yaml

@@ -0,0 +1,98 @@
+{% set name = "iregi" %}
+{% set version = "1.1.0" %}
+
+package:
+  name: {{ name }}
+  version: {{ version }}
+
+source:
+  - url: https://github.com/hep-packaging-coordination/mg5amcnlo-release-mirror/releases/download/3.5.7/MG5_aMC_v3.5.7.tar.gz
+    sha256: ca3e027f078438318bd4143a102aebc2d51358f537b5d52bbc574fe49a6a901e
+    patches:
+      # FIXME: makefile_ML5_lib tries to build everything by default
+      - makefile_ML5_lib.patch
+      # FIXME: mg5amcnlo uses 'avh_olo_forIREGI' instead of 'avh_olo' is mis_warp.f90
+      - use-avh_olo.patch
+
+build:
+  number: 0
+
+outputs:
+  - name: {{ name }}-static
+
+    build:
+      skip: true  # [win]
+      script:
+        - cd vendor/IREGI/src/
+
+        # Remove vendored code that is on conda-forge
+        - rm -rf oneloop
+        - rm -rf qcdloop
+        - rm makefile
+        - rm makefile_oneloop
+        - rm makefile_qcdloop
+
+        # Build
+        # Enable extended-source with '-ffixed-line-length-none' for long lines
+        # c.f. https://gcc.gnu.org/onlinedocs/gfortran/Fortran-Dialect-Options.html#index-ffixed-line-length-n
+        # The build is not parallel safe, so must compile single threaded
+        - make -f makefile_ML5_lib FC=$FC FFLAGS="$FFLAGS -std=legacy -Wall -ffixed-line-length-none -I$PREFIX/include/oneloop"
+
+        # Install
+        - mv libiregi.a $PREFIX/lib/
+        - mkdir -p $PREFIX/include/iregi
+        - mv *.mod $PREFIX/include/iregi/
+
+    requirements:
+      build:
+        - {{ stdlib('c') }}
+        - {{ compiler('fortran') }}
+        - make
+        - libtool
+      host:
+        - oneloop
+
+    test:
+      commands:
+        - test -f $PREFIX/lib/libiregi.a
+        - test -f $PREFIX/include/iregi/binary_tree.mod
+        - test -f $PREFIX/include/iregi/cmatrix_base.mod
+        - test -f $PREFIX/include/iregi/cpave_reduce.mod
+        - test -f $PREFIX/include/iregi/csi_reduce.mod
+        - test -f $PREFIX/include/iregi/cti_reduce.mod
+        - test -f $PREFIX/include/iregi/funlib.mod
+        - test -f $PREFIX/include/iregi/global.mod
+        - test -f $PREFIX/include/iregi/gti_reduce.mod
+        - test -f $PREFIX/include/iregi/kinematics.mod
+        - test -f $PREFIX/include/iregi/linear_algebra.mod
+        - test -f $PREFIX/include/iregi/matrices.mod
+        - test -f $PREFIX/include/iregi/matrix_base.mod
+        - test -f $PREFIX/include/iregi/mis_warp.mod
+        - test -f $PREFIX/include/iregi/pave_reduce.mod
+        - test -f $PREFIX/include/iregi/si_reduce.mod
+        - test -f $PREFIX/include/iregi/special_fun.mod
+        - test -f $PREFIX/include/iregi/ti_reduce.mod
+
+about:
+  home: https://github.com/mg5amcnlo/mg5amcnlo/tree/3.x/vendor/IREGI/src
+  summary: 'IREGI: one-loop tensor Integral REduction with General propagator Indices'
+  description: |
+    IREGI is a package for one-loop tensor Integral REduction with General
+    propagator Indices.
+
+    The algorithm was proposed by G. Duplancic, B. Nizic
+    (DOI: [10.1140/epjc/s2004-01723-7](https://doi.org/10.1140/epjc/s2004-01723-7)).
+
+    This implementation for MADLOOP5 in MadGraph5_aMC@NLO was originally
+    authored by Hua-Sheng Shao.
+  # IREGI source only exists in the MG5_aMC@NLO repo and so has the same license
+  # modified University of Illinois/NCSA license
+  license: LicenseRef-NCSA
+  license_family: OTHER
+  license_file: LICENSE
+  dev_url: https://github.com/mg5amcnlo/mg5amcnlo/tree/3.x/vendor/IREGI/src
+
+extra:
+  feedstock-name: iregi
+  recipe-maintainers:
+    - matthewfeickert

recipes/iregi/use-avh_olo.patch

@@ -0,0 +1,26 @@
+From 751683b8e56e6116338d8b0d2e50eed9cfb5be53 Mon Sep 17 00:00:00 2001
+From: Matthew Feickert <matthew.feickert@cern.ch>
+Date: Sun, 9 Feb 2025 03:14:54 -0700
+Subject: [PATCH] fix: Use avh_olo as standalone build
+
+* Use avh_olo from OneLOop.
+---
+ vendor/IREGI/src/mis_warp.f90 | 2 +-
+ 1 file changed, 1 insertion(+), 1 deletion(-)
+
+diff --git a/vendor/IREGI/src/mis_warp.f90 b/vendor/IREGI/src/mis_warp.f90
+index 3f3b93b..3c972ef 100644
+--- a/vendor/IREGI/src/mis_warp.f90
++++ b/vendor/IREGI/src/mis_warp.f90
+@@ -1,7 +1,7 @@
+ MODULE mis_warp
+ USE global
+ USE kinematics
+-USE avh_olo_forIREGI
++USE avh_olo
+ IMPLICIT NONE
+ LOGICAL::first=.TRUE.
+ SAVE first
+-- 
+2.47.1
+