Actions

Difference between revisions of "VLD"

From Online Dictionary of Crystallography

m (Style edits to align with printed edition)
 
(5 intermediate revisions by 3 users not shown)
Line 1: Line 1:
 
==Definition==
 
==Definition==
  
VLD (Vive la Différence) is a phasing approach based on the properties of the Fourier transforms.  Probabilistic techniques  lead to the difference electron density with Fourier coefficients
+
VLD ("Vive la Différence") is a phasing approach based on the properties of the Fourier transforms.  Probabilistic techniques  lead to the difference electron density with Fourier coefficients
 
 
<math>F_q\approx \bigg[(m|F|-D|F_p|)-|F_p|(1-D) \bigg(  \frac{e-\sigma^2_A}{1-\sigma^2_A} \bigg) \bigg]exp(\textrm{i}\varphi_p)~~~~~(1)</math>
 
  
 +
<center>
 +
<math>F_q\approx \bigg[(m|F|-D|F_p|)-|F_p|(1-D) \bigg(  \frac{e-\sigma^2_A}{1-\sigma^2_A} \bigg) \bigg]\exp(\textrm{i}\varphi_p).~~~~~(1)</math>
 +
</center>
  
 
<math>|F|</math>and <math>|F_p|</math>are the target and the model structure factor amplitudes, respectively, <math>\varphi</math> and <math>\varphi_p</math> the associated phases, <math>\textit{e}</math>  is related to the experimental error,   
 
<math>|F|</math>and <math>|F_p|</math>are the target and the model structure factor amplitudes, respectively, <math>\varphi</math> and <math>\varphi_p</math> the associated phases, <math>\textit{e}</math>  is related to the experimental error,   
  
<math>\sigma_A=\Bigg(\frac{\sum_p}{\sum_N}\Bigg)^\frac{1}{2}D</math>
+
<center>
 +
<math>\sigma_A=\Bigg(\frac{\sum_p}{\sum_N}\Bigg)^{1/2}D,</math>
 +
</center>
  
<math>\sum N</math> and <math>\sum p</math> are  the scattering powers of the target and of the model structure respectively. <math>D</math>  is related to the misfit between model and target structure via the relation  <math>D=\langle\cos(2\pi \mathbf{h}\Delta \mathbf{r})\rangle</math>
+
<math>\sum</math><sub>''N'' </sub> and <math>\sum</math><sub>''p'' </sub> are  the scattering powers of the target and of the model structure respectively. <math>D</math>  is related to the misfit between model and target structure ''via'' the relation  <math>D=\langle\cos(2\pi \mathbf{h}\Delta \mathbf{r})\rangle</math>.
  
  
The difference electron density calculated via coefficients (1)  is well correlated with  the true difference density even when the model is uncorrelated with the target, and suggests the following cyclic algorithm :
+
The difference electron density calculated ''via'' coefficients (1)  is well correlated with  the true difference density even when the model is uncorrelated with the target, and suggests the following cyclic algorithm:
  
  
a) Assign random phases to  the target structure factors and calculate the  corresponding target  electron density  map;
+
(''a'') Assign random phases to  the target structure factors and calculate the  corresponding target  electron density  map.
  
b) Select  a model electron density and calculate the difference electron density via the coefficients (1). Once modified and  Fourier inverted it will provide <math>F_q</math> coefficients;
+
(''b'') Select  a model electron density and calculate the difference electron density ''via'' the coefficients (1). Once modified and  Fourier inverted it will provide <math>F_q</math> coefficients.
  
c) A new electron density map may be obtained by summing model and difference electron densities via the tangent formula
+
(''c'') A new electron density map may be obtained by summing model and difference electron densities ''via'' the tangent formula
<math>\tan\varphi=\frac{R_p\sin\varphi_p+w_qR_q\sin\varphi_q}{R_p\cos\varphi_p+w_qR_q\cos\varphi_q}~~~~~(2)</math><br>
+
<center>
 +
<math>\tan\varphi=\frac{R_p\sin\varphi_p+w_qR_q\sin\varphi_q}{R_p\cos\varphi_p+w_qR_q\cos\varphi_q}~~~~~(2)</math></center><br>
 
where <math>w_q=\sqrt{2(1-\sigma_A)}</math>.
 
where <math>w_q=\sqrt{2(1-\sigma_A)}</math>.
  
d) Calculate the new electron density map for the target  by using phases given by (2) and return to b).
+
(''d'') Calculate the new electron density map for the target  by using phases given by (2) and return to (''b'').
 
Repeated cycles of the above algorithms lead from random phases to the correct values.
 
Repeated cycles of the above algorithms lead from random phases to the correct values.
The method may be generalized by replacing the difference electron density by hybrid electron densities. Furthermore , VLD may be used for extending and refining models provided by non ab initio approaches.
+
The method may be generalized by replacing the difference electron density by hybrid electron densities. Furthermore, VLD may be used for extending and refining models provided by non ab initio approaches.
  
 
== References ==
 
== References ==
  
  
*Burla, M.C., Caliandro, C. ''et al.'' (2010). ''The difference electron density: a probabilistic reformulation'', Acta Cryst. '''A66''', 347-361.
+
*Burla, M. C., Caliandro, C. ''et al.'' (2010). ''Acta Cryst.'' A'''66''', 347-361. ''The difference electron density: a probabilistic reformulation''.
*Burla, M.C., Giacovazzo, C. ''et al.'' (2010). ''The VLD algorithm''. J. Appl. Cryst. (2010). '''43''', 825–836.
+
*Burla, M. C., Carrozzini, B. ''et al.'' (2012). ''J. Appl. Cryst.'' '''45''', 1287-1294. ''VLD algorithm and hybrid Fourier syntheses''.  
 +
 
 +
[[Category: Structure determination]]

Latest revision as of 18:10, 17 May 2017

Definition

VLD ("Vive la Différence") is a phasing approach based on the properties of the Fourier transforms. Probabilistic techniques lead to the difference electron density with Fourier coefficients

[math]F_q\approx \bigg[(m|F|-D|F_p|)-|F_p|(1-D) \bigg( \frac{e-\sigma^2_A}{1-\sigma^2_A} \bigg) \bigg]\exp(\textrm{i}\varphi_p).~~~~~(1)[/math]

[math]|F|[/math]and [math]|F_p|[/math]are the target and the model structure factor amplitudes, respectively, [math]\varphi[/math] and [math]\varphi_p[/math] the associated phases, [math]\textit{e}[/math] is related to the experimental error,

[math]\sigma_A=\Bigg(\frac{\sum_p}{\sum_N}\Bigg)^{1/2}D,[/math]

[math]\sum[/math]N and [math]\sum[/math]p are the scattering powers of the target and of the model structure respectively. [math]D[/math] is related to the misfit between model and target structure via the relation [math]D=\langle\cos(2\pi \mathbf{h}\Delta \mathbf{r})\rangle[/math].


The difference electron density calculated via coefficients (1) is well correlated with the true difference density even when the model is uncorrelated with the target, and suggests the following cyclic algorithm:


(a) Assign random phases to the target structure factors and calculate the corresponding target electron density map.

(b) Select a model electron density and calculate the difference electron density via the coefficients (1). Once modified and Fourier inverted it will provide [math]F_q[/math] coefficients.

(c) A new electron density map may be obtained by summing model and difference electron densities via the tangent formula

[math]\tan\varphi=\frac{R_p\sin\varphi_p+w_qR_q\sin\varphi_q}{R_p\cos\varphi_p+w_qR_q\cos\varphi_q}~~~~~(2)[/math]

where [math]w_q=\sqrt{2(1-\sigma_A)}[/math].

(d) Calculate the new electron density map for the target by using phases given by (2) and return to (b). Repeated cycles of the above algorithms lead from random phases to the correct values. The method may be generalized by replacing the difference electron density by hybrid electron densities. Furthermore, VLD may be used for extending and refining models provided by non ab initio approaches.

References

  • Burla, M. C., Caliandro, C. et al. (2010). Acta Cryst. A66, 347-361. The difference electron density: a probabilistic reformulation.
  • Burla, M. C., Carrozzini, B. et al. (2012). J. Appl. Cryst. 45, 1287-1294. VLD algorithm and hybrid Fourier syntheses.