The ABINIT Users Mailing List ( CLOSED )

["Adam Sorini" <asorini@gmail.com>]

On Mon, Oct 20, 2008 at 6:41 PM, Allan, Douglas C Dr <AllanDC@corning.com> wrote:

Adam,

 

The word "columnwise" in this context means that the first 3 numbers you enter will fill out the first column and so forth.  You never enter numbers columnwise when you are reading files line by line, you just enter numbers.  It is not a misprint or typo.

Hi Doug,

Thanks for your clarification.

I think the OP's point is that the doc is unclear. At the very least, we can say that the doc was unclear to at least one user. ;)

Just because a document contains no typos or misprints, that doesn't mean that it is clearly written. As an example we could consider the very first sentence of the description of rprim:

"Give, in columnwise entry, the three dimensionless primitive translations in real space."

The correct interpretation (knowing that this phrase is a description of rprim and not an order to the user) is that the (set of 9) rprim(i,j) (they) give the primitive translation vectors as columns of the matrix rprim. But, to an uninitiated person reading the doc the word "give" in the first sentence might sound like it is refering to "you". As in "[You] give [via the input file in columns]... the three dimensionless primitive translations in real space." This is the danger of having implied subjects in the documentation of one's computer program... ;)

Oh well. Cheers,

Adam

 

Also do not overlook the role of the length scale, which is defined independently of rprim.  Rprim gets normalized after input so it represents directions that are of unit length.

 

Cheers,

Doug

 

---

From: Adam Sorini [mailto:asorini@gmail.com]
Sent: Monday, October 20, 2008 6:28 PM
To: forum@abinit.org
Subject: Re: [abinit-forum] Definition of rprim

 

It's my understanding that

rprim a b c
        d e f
        g h i

mean that the primative vectors are R1 = (a,b,c), R2 = (d,e,f), R3 = (g,h,i) and the matrix rprim(i,j) is
rprim(i,j) = Rj(i)

So, I think that's what they mean by "columnwise". Just that the first index of array rprim(i,j) gives the components of the primative lattice vector. The use of the word "columnwise" in the first sentence of the description of the basic input variable rprim then appears to be a mistake since we don't write the vectors in the input file "columnwise". Cheers,

Adam

On Mon, Oct 20, 2008 at 2:17 PM, Anindya Roy <royanin@gmail.com> wrote:

Hi all,

I'm having a little trouble to interpret the basic definition of
'rprim' as stated in documentation of version 5.5.4 (and similar
ones): rprim is defined to be  "...in columnwise entry, the three
dimensionless primitive translations in real space."

Later the example given for hexagonal lattice:

"rprim  0.86602540378  0.5  0.0
     -0.86602540378  0.5  0.0
      0.0            0.0  1.0

that can be avoided thanks to angdeg:

angdeg 90 90 120
"

It looks to me that new basis vectors are in row wise entry, rather
than column wise.

So if we assume that the new vector R_1 is [rprim(11)\hat{x} +
rprim(12) \hat{y} + rprim(13) \hat{z}] , then it's internally
consistent how xred and xcart are converted among themselves. Am I
missing something obvious? Your thought will be greatly appreciated.

Thanks,
Anindya Roy
Graduate student
Rutgers University, USA