Closed ErikRZH closed 2 years ago
When analysing this data: https://gist.github.com/ErikRZH/506130f5098f99cb92d5b0e2806acd87 with this threshold: https://gist.github.com/ErikRZH/d55faca48103942b4ab89ff944753175
And in RQA.cpp setting:
RQA.cpp
bool CPU_RQA = true; bool CPU_RQA_RR = true; bool CPU_RQA_DET = true; bool CPU_RQA_LAM = true; bool GPU_RQA = true; bool GPU_RQA_RR = true; bool GPU_RQA_DET = true; bool GPU_RQA_LAM = true; bool GPU_RQA_ALL = false;
Using ./RQA.exe rossler_data_2.txt thresholdsimple.txt 2 1 2 2
./RQA.exe rossler_data_2.txt thresholdsimple.txt 2 1 2 2
The GPU and CPU versions of DET do not agree: GPU_DET = 0.813348, CPU_DET= 0.817957
PyRQA with the same data and settings show that the correct value is the CPU_DET value (0.8179573291189254).
This issue was due to exclusion of the diagonal line in the calculation of the DET. The issue is now fixed.
When analysing this data: https://gist.github.com/ErikRZH/506130f5098f99cb92d5b0e2806acd87 with this threshold: https://gist.github.com/ErikRZH/d55faca48103942b4ab89ff944753175
And in
RQA.cpp
setting:Using
./RQA.exe rossler_data_2.txt thresholdsimple.txt 2 1 2 2
The GPU and CPU versions of DET do not agree: GPU_DET = 0.813348, CPU_DET= 0.817957
PyRQA with the same data and settings show that the correct value is the CPU_DET value (0.8179573291189254).