Maximum Likelihood estimation on the GPU using c++
This version is extremely preliminary. It is very likely that there are typos in the code. However, the optimization routine is working properly and I will leave it as it is for the moment.
This requires gcc version 4.9.x and above. See here for instructions on compiling gcc locally on tesla.
You also need to install the NLopt optimization library.
See here for instructions on installing NLopt locally on Tesla.
Run the example
Compile and run the example with
make parallel ./CudaT
The code will generate a csv file called PARAMETERSFOUND.csv with the ML estimates. The output of the file compiled will be stored in nohupPARALLR.out