Mean-trend-removal accuracy

[Oseltamivir]

Hi,

Just wanted to let the team know about a small error I noticed in mean trend removal.

For the SCC zero-stage-generation branch, this was being used:

if ((j % 2) == 0) {
    matrix[idx] = std::sin(angle);
} else {
    matrix[idx] = std::cos(angle);
}

But it should be

if ((j % 2) == 0) {
    matrix[idx] = std::cos(angle);
} else {
    matrix[idx] = std::sin(angle);
}

Attached are the images showing the difference. I am not sure if current main still has this error. I noticed this error due to the obvious phase shift when the temperatures are plotted.

[YanSong996]

Hi @Oseltamivir, thanks!

Could you please share the code you used to add the mean trend back?

[Oseltamivir]

Hi,

Unfortunately, my code is in a cluster being shipped back from SC.

But calculations was done simply by:

1. Generating design matrix with columns defined in equation 2 of the paper, using forcing_new.csv and params.csv
2. Calculate mean trend using design matrix
3. Add back the residuals using the z*.csv files