Another NAG out of range

[Juergen Reuter]

Running with the following SINDARIN:

process eemm = e1, E1 => e2, E2
sqrts= 250 GeV
beams = e1, E1 => isr, none
integrate (eemm)

leads to the following 'out of range' error (with NAG with the Jenkins flags):

Runtime Error: sf_mappings.f90, line 505: Subscript 1 of R (value 0) is out of range (1:3)
Program terminated by fatal error
sf_mappings.f90, line 505: Error occurred in SF_MAPPINGS:SF_RES_MAPPING_INVERSE
sf_base.f90, line 1375: Called by SF_BASE:SF_CHAIN_INSTANCE_COMPUTE_KINEMATICS
processes.f90, line 3671: Called by PROCESSES:KINEMATICS_COMPUTE_SELECTED_CHANNEL
processes.f90, line 3923: Called by PROCESSES:COMPONENT_INSTANCE_COMPUTE_SEED_KINEMATICS
processes.f90, line 5866: Called by PROCESSES:PROCESS_INSTANCE_COMPUTE_SEED_KINEMATICS
processes.f90, line 6330: Called by PROCESSES:PROCESS_INSTANCE_EVALUATE_SQME
processes.f90, line 6373: Called by PROCESSES:PROCESS_INSTANCE_EVALUATE
mci_base.f90, line 646: Called by MCI_BASE:MCI_INSTANCE_EVALUATE
mci_vamp.f90, line 1838: Called by MCI_VAMP:VAMP_SAMPLING_FUNCTION
vamp.f90, line 738: Called by VAMP_REST:VAMP_SAMPLE_GRID0
vamp.f90, line 1941: Called by VAMP_REST:VAMP_SAMPLE_GRIDS
mci_vamp.f90, line 1722: Called by MCI_VAMP:MCI_VAMP_INSTANCE_SAMPLE_GRIDS
mci_vamp.f90, line 1130: Called by MCI_VAMP:MCI_VAMP_INTEGRATE
processes.f90, line 2676: Called by PROCESSES:PROCESS_MCI_ENTRY_INTEGRATE
processes.f90, line 1468: Called by PROCESSES:PROCESS_INTEGRATE
integrations.f90, line 487: Called by INTEGRATIONS:INTEGRATION_EVALUATE
integrations.f90, line 671: Called by INTEGRATIONS:INTEGRATION_INTEGRATE
integrations.f90, line 814: Called by INTEGRATIONS:INTEGRATE_PROCESS
commands.f90, line 3022: Called by COMMANDS:CMD_INTEGRATE_EXECUTE
commands.f90, line 5762: Called by COMMANDS:COMMAND_LIST_EXECUTE
whizard.f90, line 357: Called by WHIZARD:WHIZARD_PROCESS_STREAM
whizard.f90, line 332: Called by WHIZARD:WHIZARD_PROCESS_FILE
main.f90, line 416: Called by MAIN
Aborted (core dumped)

[Juergen Reuter]

Milestone renamed

[Juergen Reuter]

WK, any update on that one?

[cweiss]

Shall we move this ticket to gitlab?

[kilian]

Update. First, there were sanity checks missing in the structure function setup. In the current case, the resonance mapping requires two parameters, while the structure function setup provides only one. This case now throws a BUG fatal error.

Second, it should be straightforward to implement a single-parameter resonance mapping. The fail case (single ISR in e+e-) is somewhat academic. However, e- nuebar -> mu- numubar is a physically meaningful test with a W resonance. I have added this to the functional tests (which throws the BUG error) and will try to set up an appropriate mapping for this.

[kilian]

Fixed this 2017/02/24.

The structure-function setups ISR/none or non/ISR now are supported also with resonances in the s channel. Actually, I added appropriate mapping modes both for a resonance and for an on-shell particle (2->1). The test results appear consistent - to be most confident in the results, somebody would have to perform an analytic calculation for comparison.

Some more complex (likewise academic) setups might still run into problems; at least, they should produce proper errors now.

[Juergen Reuter]

Milestone renamed