Martin R. Albrecht

`MatGSOGram` is for quadratic forms, i.e. it directly works on objects `G = B \cdot B^T` where `B` is a basis. The option `GSO_INT_GRAM` is for the internal representation of...

LLL would cost way more than O(n^2). From memory, the regular calls to LLL should help with numerical stability.

NumPy only does double precision, right? If that's all we need then I think it makes sense to call it (given that it is shipped with Sage anyways)

Okay, it seems this can be closed then. Re-open as needed.

You need to multiply left by the basis, something like: `g6k.M.B.multiply_left()`, see e.g. here https://github.com/malb/bdd-predicate/blob/9ff858b65ca42dffd46cca9c549e1cdcc97e616f/usvp.py#L517-L522

Where does this formula come from? I'd be surprised if `(0.2570 * beta + 16.4 + log(self.svp_repeat(beta, d), 2))` was correct.

Mhhh, now that I think about it, I don't think we should cost Quantum BKZ, i.e. change `LaaMosPol14`.

I'm struggling to reproduce this. Let's try to establish a baseline, where we don't restrict the number of samples at all: ```python sage: lwe = LWE.Parameters(n=3584, q=549824583172097, Xs=ND.DiscreteGaussian(2410.67), Xe=ND.DiscreteGaussian(2389.33)) sage:...

I'm using ```diff (mu4e--compose-setup 'edit (lambda (parent) - (find-file (plist-get parent :path)) + (let ((buf (find-file-noselect (plist-get parent :path)))) + (with-current-buffer buf + (mu4e-compose-mode) + (display-buffer buf))) ``` but I'm...

FWIW, I still have to patch this one to ```emacs-lisp (defun mu4e-compose-edit() "Edit an existing draft message." (interactive) (let* ((msg (mu4e-message-at-point))) (unless (member 'draft (mu4e-message-field msg :flags)) (mu4e-warn "Cannot edit...