I am considering the addition of periodic boundary conditions support to the BD simulation. I guess it would apply to any optimizer in IMP - what would be the technical complications? I guess naively,what is needed is mainly to reformulate the restraints in a periodic way, but I wonder if it affects any sort of score evaluator optimizations (ClosePairs, etc. - the latter is probably easy to handle as it mainly affects particles at the boundary). In the future it will also allow periodically optimized restraints (such as ones using FFT).
Barak
p.s. in my particular case, I thought of PBC over the x,y plane only, if that matter for anything, which I believe it doesn't
Actually that is already doable using the IMP::core::TransformationSymmetry score state (or particle state?). For your XYZ box, copy your system into 8 clones, and apply a different TransformationSymmetry to each of them, corresponding to the different translation. Just note that to make the calculation efficient, the score should not include the inter clone score terms. Also the optimizer/samplers must be applied only to the original copy.
On Fri, Nov 15, 2013 at 2:06 PM, Barak Raveh barak.raveh@gmail.com wrote:
> I am considering the addition of periodic boundary conditions support to > the BD simulation. I guess it would apply to any optimizer in IMP - what > would be the technical complications? I guess naively,what is needed is > mainly to reformulate the restraints in a periodic way, but I wonder if it > affects any sort of score evaluator optimizations (ClosePairs, etc. - the > latter is probably easy to handle as it mainly affects particles at the > boundary). In the future it will also allow periodically optimized > restraints (such as ones using FFT). > > Barak > > p.s. in my particular case, I thought of PBC over the x,y plane only, if > that matter for anything, which I believe it doesn't > > _______________________________________________ > IMP-dev mailing list > IMP-dev@salilab.org > https://salilab.org/mailman/listinfo/imp-dev > >
On 11/15/13 2:16 PM, Riccardo Pellarin wrote: > Actually that is already doable using the > IMP::core::TransformationSymmetry score state (or particle state?). > For your XYZ box, copy your system into 8 clones, and apply a different > TransformationSymmetry to each of them,
Right, but that's not PBC, it's periodic images. It would achieve the same result though...
Ben
Yes... And sorry for the typo, you need 26 clones if you want to have a 3D periodic system (8 clones is for 2D).
On Fri, Nov 15, 2013 at 2:20 PM, Ben Webb ben@salilab.org wrote:
> On 11/15/13 2:16 PM, Riccardo Pellarin wrote: > >> Actually that is already doable using the >> IMP::core::TransformationSymmetry score state (or particle state?). >> For your XYZ box, copy your system into 8 clones, and apply a different >> TransformationSymmetry to each of them, >> > > Right, but that's not PBC, it's periodic images. It would achieve the same > result though... > > Ben > -- > ben@salilab.org http://salilab.org/~ben/ > "It is a capital mistake to theorize before one has data." > - Sir Arthur Conan Doyle > > _______________________________________________ > IMP-dev mailing list > IMP-dev@salilab.org > https://salilab.org/mailman/listinfo/imp-dev >
Thanks! So, how much do I "pay" for the periodicity? Is the memory times 8? What about restraint evaluation?
Also, do you know how they do it in popular MD softwares? Do they keep 28 images, or just evaluate the energy function periodically?
On Fri, Nov 15, 2013 at 2:46 PM, Riccardo Pellarin < pellarin.riccardo@gmail.com> wrote:
> Yes... And sorry for the typo, you need 26 clones if you want to have a 3D > periodic system (8 clones is for 2D). > > > On Fri, Nov 15, 2013 at 2:20 PM, Ben Webb ben@salilab.org wrote: > >> On 11/15/13 2:16 PM, Riccardo Pellarin wrote: >> >>> Actually that is already doable using the >>> IMP::core::TransformationSymmetry score state (or particle state?). >>> For your XYZ box, copy your system into 8 clones, and apply a different >>> TransformationSymmetry to each of them, >>> >> >> Right, but that's not PBC, it's periodic images. It would achieve the >> same result though... >> >> Ben >> -- >> ben@salilab.org http://salilab.org/~ben/ >> "It is a capital mistake to theorize before one has data." >> - Sir Arthur Conan Doyle >> >> _______________________________________________ >> IMP-dev mailing list >> IMP-dev@salilab.org >> https://salilab.org/mailman/listinfo/imp-dev >> > >
It's a bit pricey memory-wise, if you system is big. Score evaluation is a bit less efficient, but not so much if you are careful in selecting non-redundant score terms.
On Fri, Nov 15, 2013 at 4:24 PM, Barak Raveh barak.raveh@gmail.com wrote:
> Thanks! So, how much do I "pay" for the periodicity? Is the memory times > 8? What about restraint evaluation? > > Also, do you know how they do it in popular MD softwares? Do they keep 28 > images, or just evaluate the energy function periodically? > > > > On Fri, Nov 15, 2013 at 2:46 PM, Riccardo Pellarin < > pellarin.riccardo@gmail.com> wrote: > >> Yes... And sorry for the typo, you need 26 clones if you want to have a >> 3D periodic system (8 clones is for 2D). >> >> >> On Fri, Nov 15, 2013 at 2:20 PM, Ben Webb ben@salilab.org wrote: >> >>> On 11/15/13 2:16 PM, Riccardo Pellarin wrote: >>> >>>> Actually that is already doable using the >>>> IMP::core::TransformationSymmetry score state (or particle state?). >>>> For your XYZ box, copy your system into 8 clones, and apply a different >>>> TransformationSymmetry to each of them, >>>> >>> >>> Right, but that's not PBC, it's periodic images. It would achieve the >>> same result though... >>> >>> Ben >>> -- >>> ben@salilab.org http://salilab.org/~ben/ >>> "It is a capital mistake to theorize before one has data." >>> - Sir Arthur Conan Doyle >>> >>> _______________________________________________ >>> IMP-dev mailing list >>> IMP-dev@salilab.org >>> https://salilab.org/mailman/listinfo/imp-dev >>> >> >> > > > -- > Barak >
> Also, do you know how they do it in popular MD softwares? Do they keep 28 images, or just evaluate the energy function periodically?
charmm can do both. However periodic images is very general and can be used for any weird symmetry, and PBC only for boxes.
On Fri, Nov 15, 2013 at 4:27 PM, Riccardo Pellarin < pellarin.riccardo@gmail.com> wrote:
> It's a bit pricey memory-wise, if you system is big. Score evaluation is a > bit less efficient, > but not so much if you are careful in selecting non-redundant score terms. > > > On Fri, Nov 15, 2013 at 4:24 PM, Barak Raveh barak.raveh@gmail.comwrote: > >> Thanks! So, how much do I "pay" for the periodicity? Is the memory times >> 8? What about restraint evaluation? >> >> Also, do you know how they do it in popular MD softwares? Do they keep 28 >> images, or just evaluate the energy function periodically? >> >> >> >> On Fri, Nov 15, 2013 at 2:46 PM, Riccardo Pellarin < >> pellarin.riccardo@gmail.com> wrote: >> >>> Yes... And sorry for the typo, you need 26 clones if you want to have a >>> 3D periodic system (8 clones is for 2D). >>> >>> >>> On Fri, Nov 15, 2013 at 2:20 PM, Ben Webb ben@salilab.org wrote: >>> >>>> On 11/15/13 2:16 PM, Riccardo Pellarin wrote: >>>> >>>>> Actually that is already doable using the >>>>> IMP::core::TransformationSymmetry score state (or particle state?). >>>>> For your XYZ box, copy your system into 8 clones, and apply a different >>>>> TransformationSymmetry to each of them, >>>>> >>>> >>>> Right, but that's not PBC, it's periodic images. It would achieve the >>>> same result though... >>>> >>>> Ben >>>> -- >>>> ben@salilab.org http://salilab.org/~ben/ >>>> "It is a capital mistake to theorize before one has data." >>>> - Sir Arthur Conan Doyle >>>> >>>> _______________________________________________ >>>> IMP-dev mailing list >>>> IMP-dev@salilab.org >>>> https://salilab.org/mailman/listinfo/imp-dev >>>> >>> >>> >> >> >> -- >> Barak >> > >
-
Barak Raveh -
Ben Webb -
Riccardo Pellarin