From a541800da558125510c87d40b2e871b0cb512523 Mon Sep 17 00:00:00 2001 From: Chad Hanna <chad.hanna@ligo.org> Date: Tue, 14 Aug 2018 13:37:32 -0400 Subject: [PATCH] update makefile for bns svs bank --- .../share/O3/sept_opa/Makefile.bns_svd | 72 +++++++++++++++++++ 1 file changed, 72 insertions(+) create mode 100644 gstlal-inspiral/share/O3/sept_opa/Makefile.bns_svd diff --git a/gstlal-inspiral/share/O3/sept_opa/Makefile.bns_svd b/gstlal-inspiral/share/O3/sept_opa/Makefile.bns_svd new file mode 100644 index 0000000000..ec2985e28e --- /dev/null +++ b/gstlal-inspiral/share/O3/sept_opa/Makefile.bns_svd @@ -0,0 +1,72 @@ +############################ +# Template bank parameters # +############################ + +# The frequency at which to begin generating a template +LOW_FREQUENCY_CUTOFF = 15.0 + +# The frequency at which to stop generating a template +HIGH_FREQUENCY_CUTOFF = 1024 + +# The sample rate, note that this is usually exactly = 2 * HIGH_FREQUENCY_CUTOFF +SAMPLE_RATE = 2048 + +# The number of templaes that go into each sub bank. The total number of templates in a file is given by NUM_SPLIT_TEMPLATES * NUMBANKS +NUM_SPLIT_TEMPLATES = 500 + +# How many effective spin bins to use. If this is a nonspinning bank, then use +# 1. For a fully spinning bank 20 is a good number. In this example we have +# spin magnitudes up to 0.05 so we choose 3. +NUM_CHI_BINS = 3 + +# This determines how many overlap tempates are used the boundaries of each sub +# bank. NOTE that the overlap templates are only used to help compute the SVD, +# they are discarded in the final single inspiral table. The analysis does not +# contain duplicate templates at the end. +OVERLAP = 50 + +# NOTE 1000 is used in place of "infinty" it is possible to specify additiona +# approximantes. The format is (chirp mass start):approximant:(chirp mass end) +# which allows the user to specify different chirp masses for each approximant +APPROXIMANT1 = 0.00:1000:TaylorF2 + +# This determines how many sub banks go into each file. The interpretation is +# that the number of times you specify a given integer in this list will +# determin the frequency at which than number of sub banks is put into a file. +# The order matters, it goes from lowest to highest chirp mass. Examples: +# +# NUMBANKS = 3,4,5 +# +# Approximately 1/3 of the files would contain 3 sub banks, 1/3 would contain 4 sub banks and 1/3 would contain 5 sub banks. The lowest mass templates would be grouped into 3 sub banks and the highest mass templates would be grouped into 5 sub banks. +# +# NUMBANKS = 3,4,4,5 +# +# Here 1/4 would have 3 sub banks, 1/2 would have 4 and 1/4 would have 5. Again, the lowest mass templates would have three sub banks. The in-the-middle mass templates would have 4 and the highest mass templates would have 5. +# +NUMBANKS = 4,5 + +# The reference PSD +PSD = ../../banks/bns/H1L1V1-REFERENCE_PSD-1186624818-687900.xml.gz + +# The template bank file that you want to split up and decompose. +BANK = ../../banks/bns/SBANK_COMBINED-GSTLAL_BNS.xml.gz + +# Condor information +ACCOUNTING_GROUP=ligo.dev.o3.cbc.em.gstlalonline +ACCOUNTING_USER=chad.hanna + +all : bank.dag + +%_split_bank.cache : $(BANK) + mkdir -p $*_split_bank + gstlal_bank_splitter --f-low $(LOW_FREQUENCY_CUTOFF) --group-by-chi $(NUM_CHI_BINS) --output-path $*_split_bank --approximant $(APPROXIMANT1) --output-cache $@ --overlap $(OVERLAP) --instrument $* --n $(NUM_SPLIT_TEMPLATES) --sort-by mchirp --max-f-final $(HIGH_FREQUENCY_CUTOFF) $(BANK) + +%_bank.dag : %_split_bank.cache $(PSD) + gstlal_inspiral_svd_bank_pipe --autocorrelation-length 351 --instrument $* --reference-psd $(PSD) --bank-cache $< --overlap $(OVERLAP) --flow $(LOW_FREQUENCY_CUTOFF) --output-name $@ --num-banks $(NUMBANKS) --samples-max-256 2048 --samples-min 2048 --condor-command 'accounting_group = $(ACCOUNTING_GROUP)' --condor-command 'accounting_group = $(ACCOUNTING_USER)' + +bank.dag : H1_bank.dag L1_bank.dag V1_bank.dag + cat $^ > bank.dag + rm -f $^ + +clean : + rm -rf *.sub* *.dag* *.cache *.sh logs gstlal_svd_bank* *split_bank* -- GitLab