Talk:Biconjugate gradient method

Scalars and Vectors

I've implemented this algorithm, and in the presentation, someone needs to more clearly differentiate between the terms that are scalars and those that aren't. For instance, ${\alpha }_{k}$ and ${\beta }_{k}$ are scalars and the rest are not.CFDFEM (talk) 20:39, 1 September 2009 (UTC)

I suppose it is more or less a convention that Greek letters are usually used for scalars, lowercase letters for vectors and uppercase letters for matrices, though personally I would prefer further having vectors and matrices in bold type for better distinction.Kxx (talk) 05:13, 3 September 2009 (UTC)

The wiki page for a similar method (https://en.wiki.x.io/wiki/Conjugate_gradient_method) distinguishes between vectors and scalars by writing vectors in bold — Preceding unsigned comment added by 2600:1702:3C50:A520:A90D:B199:2070:C1B9 (talk) 04:02, 21 June 2022 (UTC)

Additional citations needed

The only source cited by this article is the R. Fletcher's paper that originally proposed BiCG. Fletcher's paper describes only the unpreconditioned BiCG for real systems (evidenced by use of simple transpose instead of conjugate transpose) and contains limited discussion of properties of BiCG (biorthogonality, biconjugacy and finiteness only). Additional citations are needed to support contents beyond what is in Fletcher's paper.Kxx (talk) 07:28, 8 January 2010 (UTC)

Kxx, in 2011, ServiceAT added a reference. I think it addresses this problem, but inline citations are still desired. Gryllida 05:49, 11 September 2013 (UTC)

I do not have a copy of Numerical Recipes to verify that, especially whether the complex formulation of BiCG is in the book. Kxx (talk | contribs) 01:00, 12 September 2013 (UTC)

Can we omit the dual solution?

In the posted algorithms we include the dual solution, although it is not used and not a desired output of the algorithm. Perhaps we should omit it?--Frozenport (talk) 09:24, 8 May 2015 (UTC)