I think you could answer this by performing the following thought experiment:

1. Refine the structure to convergence using strict NCS constraints.

2. Switch to using the equivalent 'infinite-in-the-limit weight'
restraints, keeping everything else as is & continue refinement of the
output from step 1 (assume limitations of finite precision in the code
have been overcome by re-programming using whatever precision
arithmetic is necessary).

3. Does there exist some finite value of the weight such that the
structure changes by less than the experimental error (say by < 0.2
Ang RMSD) at step 2 (and if not then why not?).

If the answer to (3) is yes, then there's no significant difference in
effect between constraints & 'infinite weight' restraints.

Cheers

-- Ian

On Thu, Sep 23, 2010 at 5:06 PM, MARTYN SYMMONS
<[email protected]> wrote:
> Dear All
>        one thing I remembered from what Gerard pointed out was the difference 
> in the XPLOR/CNS formalism between strict and restrained which is not a 
> continuum. Restrained was obviously when you had multiple copies and they 
> were restrained with a weight (which was like a force constant) to be similar 
> when superimposed. So if you increase the force constant then they can move 
> during refinement but they all try to move together when they move.
>
> And the other extreme is strict where there was no force applied at all but 
> only a single copy of the chain and the ASU is built by applying the NCS 
> symmetry. The atoms are free to move but, unlike the case with restrained 
> where there is superimposition on the fly, in the strict case there is no 
> automatic update of the superimposition matrices. So every move gets 
> religiously copied to all the chains when the ASU is made. At this point I 
> guess the copies can bump and so apply a force on each other but that is a 
> local, and likely to be perturbing, force.
>
> best wishes
>  Martyn
>
> Martyn Symmons
> Cambridge
>
>
>
> --- On Thu, 23/9/10, Ian Tickle <[email protected]> wrote:
>
>> From: Ian Tickle <[email protected]>
>> Subject: Re: [ccp4bb] Effect of NCS on estimate of data:parameter ratio
>> To: [email protected]
>> Date: Thursday, 23 September, 2010, 11:21
>> Hi Gerard & Pavel
>>
>> Isn't this the proviso I was referring to, that one cannot
>> in practice
>> use an infinite weight because of rounding errors in the
>> target
>> function.  The weight just has to be 'big enough' such
>> that the
>> restraint residual becomes sufficiently small that it's no
>> longer
>> significant.
>>
>> In numerical constrained optimisation the method of
>> increasing the
>> constraint weights (a.k.a. 'penalty coefficients') until
>> the
>> constraint violations are sufficiently small is called the
>> 'penalty
>> method', see http://en.wikipedia.org/wiki/Penalty_method .  The
>> method
>> where you substitute some of the parameters using the
>> constraint
>> equations is called (you guessed it!) the 'substitution
>> method', see
>> http://people.ucsc.edu/~rgil/Optimization.pdf .
>> There are several
>> other methods, e.g. the 'augmented Lagrangian method' is
>> very popular,
>> see 
>> http://www.ualberta.ca/CNS/RESEARCH/NAG/FastfloDoc/Tutorial/html/node112.html
>> .  As in the penalty method, the AL method adds
>> additional parameters
>> to be determined (the Lagrange multipliers, one per
>> constraint)
>> instead of eliminating some parameters using the constraint
>> equations;
>> however the advantage is that it removes the requirement
>> that the
>> penalty coefficient be very big.
>>
>> The point about all these methods of constrained
>> optimisation is that
>> they are in principle only different ways of achieving the
>> same
>> result, at least that's what the textbooks say!
>>
>> And now after the penalties and substitutions it's time to
>> blow the whistle ...
>>
>> Cheers
>>
>> -- Ian
>>
>> On Wed, Sep 22, 2010 at 10:00 PM, Pavel Afonine <[email protected]>
>> wrote:
>> >  I agree with Gerard. Example: it's unlikely to
>> achieve a result of
>> > rigid-body refinement (when you refine six
>> rotation/translation parameters)
>> > by replacing it with refining individual coordinates
>> using infinitely large
>> > weights for restraints.
>> > Pavel.
>> >
>> >
>> > On 9/22/10 1:46 PM, Gerard DVD Kleywegt wrote:
>> >>
>> >> Hi Ian,
>> >>
>> >>> First, constraints are just a special case of
>> restraints in the limit
>> >>> of infinite weights, in fact one way of
>> getting constraints is simply
>> >>> to use restraints with very large weights
>> (though not too large that
>> >>> you get rounding problems). These
>> 'pseudo-constraints' will be
>> >>> indistinguishable in effect from the 'real
>> thing'.  So why treat
>> >>> restraints and constraints differently as far
>> as the statistics are
>> >>> concerned: the difference is purely one of
>> implementation.
>> >>
>> >> In practice this is not true, of course. If you
>> impose "infinitely strong"
>> >> NCS restraints, any change to a thusly restrained
>> parameter by the
>> >> refinement program will make the target function
>> infinite, so effectively
>> >> your model will never change. This is very
>> different from the behaviour
>> >> under NCS constraints and the resulting models in
>> these two cases will in
>> >> fact be very easily distinguishable.
>> >>
>> >> --Gerard
>> >>
>> >>
>> ******************************************************************
>> >>                           Gerard J.
>>  Kleywegt
>> >>   Dept. of Cell & Molecular Biology
>>  University of Uppsala
>> >>                   Biomedical Centre  Box
>> 596
>> >>                   SE-751 24 Uppsala
>>  SWEDEN
>> >>
>> >>    http://xray.bmc.uu.se/gerard/  mailto:[email protected]
>> >>
>> ******************************************************************
>> >>   The opinions in this message are fictional.
>>  Any similarity
>> >>   to actual opinions, living or dead, is purely
>> coincidental.
>> >>
>> ******************************************************************
>> >
>>
>
>
>

Reply via email to