[Pw_forum] scaling of turbo_lanczos.x

[Simon Binnie]

Hello Eduardo,

The relevant Equation is Eq. 33 and the definition of \alpha_{ij}(\omega)
at the start of Section 2. 'ipol' refers to the axis long which the
perturbing electric field is polarized for the calculation. If only one
direction is considered (1 =< ipol =< 3) then only the appropriate
diagonal part of \alpha_{ij}(\omega) is calculated, so you get
\alpha_{ipol,ipol}(\omega). Since this only depends on ^{m}z_{ipol ipol}
the off diagonal parts, ^{m}z_{i ipol} (for i \= ipol) are not calculated.

In the case where ipol=4 the programme runs three times with ipol=1,2,3.
In this case the off-diagonal parts of \alpha_{ij}(\omega) are also
calculated which in turn require the off-diagonal components of z, leading
to three z components at each step (^{m}z_{i ipol} for i = 1,2 3).

Brgrds,

Simon

>> From: Eduardo Ariel Menendez Proupin <eariel99 at gmail.com>
>> Date: January 12, 2012 12:28:53 PM GMT+01:00
>> To: pw_forum at pwscf.org
>> Subject: Re: [Pw_forum] scaling of turbo_lanczos.x
>> Reply-To: PWSCF Forum <pw_forum at pwscf.org>
>>
>>
>> I have one more question ? What is nipol ? The manual reads
>>
>>         nipol           1 if ipol < 4; 3 if ipol=4
>>                 Determines the number of zeta coefficients to be  
>> calculated for
>>                  a given polarization direction.
>> I think this z is the $^{m}z_{ij}$ present in Eqs. 33, 34-36 of the  
>> article included in the Doc directory. What parameter in the Eq.  
>> correspond to nipol ?
>>

-- 
Simon Binnie | Post Doc, Condensed Matter Sector
Scuola Internazionale di Studi Avanzati (SISSA)
Via Bonomea 256 | 34100 Trieste | sbinnie at sissa.it