Skip to main content
Revision 1b619a75151e3d5d95d32a2fb732e1208ec0cd33 authored by Konrad Werys on 01 November 2019, 11:50:53 UTC, committed by Konrad Werys on 01 November 2019, 11:50:53 UTC
ci: big changes to the builds v11
1 parent e9d12cf
Raw File Download
notepad.md 
# NOTES

## Different results from the configurations

parameter_to_map: T1_SHMOLLI  
functions_type: **FunctionsThreeParams/FunctionsShmolli**  
fitting_method: LevMarVnl  
max_function_evals: 4000  
fTolerance: 1e-12  
sign_calc_method: MagPhase  
start_point_calc_method: Default  

FunctionsThreeParams: 10s and seemingly correct results  
FunctionsShmolli: 2.2s and wrong results  

parameter_to_map: T1_SHMOLLI  
functions_type: FunctionsShmolli  
fitting_method: LevMarVnl  
max_function_evals: 4000  
fTolerance: 1e-12  
sign_calc_method: MagPhase  
start_point_calc_method: **Default/StartPointSHMOLLI**  

Default: 2.2s and wrong results  
StartPointSHMOLLI: 10s and seemingly correct results

## 5 vs 7 samples

5 samples are calculated way slower. Maybe playing with fmax and maxIter would reduce the time?