Thank you for this great package!
It would be great if there was a flang plugin (similar to clang), that does all the proper preprocessing steps, to make it easier to use. More documentation on just the steps required to compile fortran-based code with Enzyme already would help.
Moreover, some more documentation on using vector mode AD would be amazing. I have a function that solves an ODE and returns an array of shape (N_TIME_STEPS, N_COLUMNS), and would like to get the derivative of all columns, at any timestep, resulting in an array of (N_ARG, N_TIME_STEPS, N_COLUMNS). The documentation I could find was not really clear on this.
Thank you for this great package!
It would be great if there was a flang plugin (similar to clang), that does all the proper preprocessing steps, to make it easier to use. More documentation on just the steps required to compile fortran-based code with Enzyme already would help.
Moreover, some more documentation on using vector mode AD would be amazing. I have a function that solves an ODE and returns an array of shape
(N_TIME_STEPS, N_COLUMNS), and would like to get the derivative of all columns, at any timestep, resulting in an array of(N_ARG, N_TIME_STEPS, N_COLUMNS). The documentation I could find was not really clear on this.