查看“︁NNPDF”︁的源代码

{{Expert needed|物理}}

{{Infobox software
| name                   = NNPDF
| logo                   = LogoNNPDF.png
| developer              = The NNPDF Collaboration
| latest release version = 3.1
| genre                  = [[粒子物理]]
| website                = {{URL|http://nnpdf.hepforge.org}}
}}

'''NNPDF'''是用於識別[[部分子]]分佈函數（{{lang-en|parton distribution functions}}）的首字母縮寫詞。 

{{TransH}}

'''NNPDF''' is the acronym used to identify the [[parton distribution functions]] from the NNPDF Collaboration. NNPDF parton densities are 
extracted from global fits to data based on a combination of a [[Monte Carlo method]] for uncertainty estimation and the use of 
[[neural networks]] as basic interpolating functions.

== Methodology ==
[[File:General-strategy-evol.jpg|thumb|The NNPDF Collaboration strategy is summarized in this diagram.]]

NNPDF途徑可以分為四個主要步驟：
* The generation of a large sample of Monte Carlo replicas of the original experimental data, in a way that central values, errors and correlations are reproduced with enough accuracy.
* The training (minimization of the <math>\chi^2</math>) of a set of PDFs parametrized by [[artificial neural network|neural network]]s on each of the above MC replicas of the data. PDFs are parametrized at the initial evolution scale <math>Q^{2}_{0}</math> and then evolved to the experimental data scale <math>Q^2</math> by means of the [[DGLAP]] equations. Since the PDF parametrization is redundant, the minimization strategy is based in [[genetic algorithm]]s as well as gradient descent based minimizers.
* The neural network training is stopped dynamically before entering into the overlearning regime, that is, so that the PDFs learn the physical laws which underlie experimental data without fitting simultaneously statistical noise.
* Once the training of the MC replicas has been completed, a set of statistical estimators can be applied to the set of PDFs, in order to assess the statistical consistency of the results. For example, the stability with respect PDF parametrization can be explicitly verified.

The set of <math>N_{rep}</math> PDF sets (trained neural networks) provides a representation of the underlying PDF probability density, from which   any statistical estimator can be computed.

{{TransF}}

== 示例 ==

<gallery widths=444px heights=300px>
File:Gluon log ref.jpg|NNPDF1.0 膠子
</gallery>

<!--The image below shows the [[gluon]] at small-x from  [https://arxiv.org/abs/0808.1231 the NNPDF1.0 analysis], available
through [http://projects.hepforge.org/lhapdf/ the LHAPDF interface]-->

== 版本 ==
NNPDF各版本：
{| class="wikitable centered sortable" style="text-align: center;"
|-
! PDF set
! DIS 數據
! Drell-Yan 數據
! Jet 數據
! LHC 數據
! 獨立<math>s</math>和<math>\bar{s}</math>參數
! 重夸克質量
! NNLO
|-
| [http://projects.hepforge.org/lhapdf/ '''NNPDF3.1'''] {{Wayback|url=http://projects.hepforge.org/lhapdf/ |date=20120214044927 }}
| {{Yes}}
| {{Yes}}
| {{Yes}}
| {{Yes}}
| {{Yes}}
| {{Yes}}
| {{Yes}}
|-
|-
| [http://projects.hepforge.org/lhapdf/ '''NNPDF3.0'''] {{Wayback|url=http://projects.hepforge.org/lhapdf/ |date=20120214044927 }}
| {{Yes}}
| {{Yes}}
| {{Yes}}
| {{Yes}}
| {{Yes}}
| {{Yes}}
| {{Yes}}
|-
| [http://projects.hepforge.org/lhapdf/ '''NNPDF2.3'''] {{Wayback|url=http://projects.hepforge.org/lhapdf/ |date=20120214044927 }}
| {{Yes}}
| {{Yes}}
| {{Yes}}
| {{Yes}}
| {{Yes}}
| {{Yes}}
| {{Yes}}
|-
| [http://projects.hepforge.org/lhapdf/ '''NNPDF2.2'''] {{Wayback|url=http://projects.hepforge.org/lhapdf/ |date=20120214044927 }}
| {{Yes}}
| {{Yes}}
| {{Yes}}
| {{Yes}}
| {{Yes}}
| {{Yes}}
| {{Yes}}
|-
| [http://projects.hepforge.org/lhapdf/ '''NNPDF2.1'''] {{Wayback|url=http://projects.hepforge.org/lhapdf/ |date=20120214044927 }}
| {{Yes}}
| {{Yes}}
| {{Yes}}
| {{No}}
| {{Yes}}
| {{Yes}}
| {{Yes}}
|-
| [http://projects.hepforge.org/lhapdf/ '''NNPDF2.0'''] {{Wayback|url=http://projects.hepforge.org/lhapdf/ |date=20120214044927 }}
| {{Yes}}
| {{Yes}}
| {{Yes}}
| {{No}}
| {{Yes}}
| {{No}}
| {{No}}
|-
| [http://projects.hepforge.org/lhapdf/ '''NNPDF1.2'''] {{Wayback|url=http://projects.hepforge.org/lhapdf/ |date=20120214044927 }}
| {{Yes}}
| {{No}}
| {{No}}
| {{No}}
| {{Yes}}
| {{No}}
| {{No}}
|-
| [http://projects.hepforge.org/lhapdf/ '''NNPDF1.0'''] {{Wayback|url=http://projects.hepforge.org/lhapdf/ |date=20120214044927 }}
| {{Yes}}
| {{No}}
| {{No}}
| {{No}}
| {{No}}
| {{No}}
| {{No}}
|-
|}

所有PDF集都可通過LHAPDF界面和[http://nnpdf.mi.infn.it/for-users/ NNPDF網頁] {{Wayback|url=http://nnpdf.mi.infn.it/for-users/ |date=20220820093531 }}獲得。

== 外部鏈接 ==
*[http://nnpdf.hepforge.org The NNPDF Collaboration 主頁] {{Wayback|url=http://nnpdf.hepforge.org/ |date=20220709152836 }}
*[http://sophia.ecm.ub.es/nnpdf/nnpdf-driver.htm Download NNPDF Parton Distribution sets]{{Dead link}}
*[https://arxiv.org/abs/0808.1231 The NNPDF1.0 analysis] {{Wayback|url=https://arxiv.org/abs/0808.1231 |date=20220820093543 }}
*[https://arxiv.org/abs/hep-ph/0701127 The NNPDF Non-Singlet analysis] {{Wayback|url=https://arxiv.org/abs/hep-ph/0701127 |date=20220820093548 }}
*[https://link.springer.com/article/10.1140%2Fepjc%2Fs10052-017-5199-5 NNPDF3.1 release] {{Wayback|url=https://link.springer.com/article/10.1140%2Fepjc%2Fs10052-017-5199-5 |date=20220511082031 }}
*[https://link.springer.com/article/10.1140%2Fepjc%2Fs10052-019-7197-2 NNPDF latest fitting code] {{Wayback|url=https://link.springer.com/article/10.1140%2Fepjc%2Fs10052-019-7197-2 |date=20210729041445 }}
*[http://projects.hepforge.org/lhapdf/ The LHAPDF interface] {{Wayback|url=http://projects.hepforge.org/lhapdf/ |date=20120214044927 }}

[[Category:物理学软件]]
[[Category:计算物理学]]