Robin Schmidt

i think, i will first try to implement my own pentadiagonal solver based on that paper: https://arxiv.org/pdf/1409.4802.pdf the pseudocode looks straightforward enough to implement. i hope, it will work out...

> the pseudocode looks straightforward enough to implement. pfff...it was a mess and there were even obvious errors in the paper (the solution vector on rhs of the matrix equation...

puuh - that would have to be a wild guess. i translated the main high-level routine that is relevant for this particular solver and the two subroutines, it calls -...

my task just became sooo much easier - there is already a c translation available: https://www.netlib.org/clapack/index.html fortunately, i did not yet translate too much fortran code myself. and having done...

> this netlib site is a real treasure trove of math-code...but much of it is written in fortran for example - this: https://en.wikipedia.org/wiki/SLATEC !!! ...and all in the public domain...

> my task just became sooo much easier - there is already a c translation available hmm.. actually, to be honest, if the f2c translation works well, this additional step...

34 routines translated, so far. i already can use a simple version of the banddiagonal-solver (gbsv) since yesterday - and it returns a correct (but numerically rather imprecise) result. so,...

if you want to experiment with zeroing out bins in the spectrogram, look at the function `spectrogramSine()` in the test project. this code creates a sinewave and sends this signal...

it's probably obvious, but just in case: if you use zero-padding, the number of bins that you want to zero out per harmonic should be scaled up by the zero-padding...

i have just finished the translation work for the band-diagonal solvers. boy - was this a big pile of grunt work! i'll file this under preparations/preliminaries so i didn't write...