Marcus P S

Sets may not have an ordering convention, but the enumeration of the elements of a set (or iteration over them) is quite often done in (left-to-right) lexicographical order. Although I...

Sure thing. On Feb 7, 2015 7:24 AM, "Tim Holy" [email protected] wrote: > I agree with @simonster https://github.com/simonster; even if you don't > instantiate the array, the returned product can...

I haven't checked. I assumed that it would be negligible, but I can look into it. I am curious about what you mean by the type parameter encoding of the...

One way to address the potential performance issue is to have different `Product` types for different orders (which I assume is functionally equivalent to what Blake suggested). As for the...

Here is a shot at a different API, based on a suggestion by @blakejohnson There is no (runtime) branching on `next()` anymore, but there is one in `product()`. There are...

Sounds good @blakejohnson On Mon, Nov 9, 2015, 9:41 PM Blake Johnson [email protected] wrote: > @marcusps https://github.com/marcusps maybe you and I should set aside > some time this week to...

@ChrisRackauckas @acroy Expmv.jl a norm 1 estimation function in a branch, just hasn't been fully tested. Development has stopped, true -- I just don't have the bandwidth at the moment.

I did a line by line translation of Al-Mohy and Higham's MATLAB implementation of their algorithm, but I am missing their fast 1-norm estimation routine. They also describe that algorithm...

It's a BSD license.

The improvement comes not from the symmetrization per se, but the fact the symmetrized errors in readout are easier to correct for. The correct, IMHO, is OK to be performed...