diff --git a/_analysis/ananotes.md b/_analysis/ananotes.md index 0f46a84..5bf439e 100644 --- a/_analysis/ananotes.md +++ b/_analysis/ananotes.md @@ -5,4 +5,5 @@ layout: newbase {% include layouts/find_title.md name=page.name %} PHENIX Analysis Notes are expected to follow the rules outlined in the official template, and adhere to its structure and layout. +The links below contain the suggested template for the analysis notes in both PDF and TeX formats. {%- include documents/doc.md tag='ananote_template' -%} diff --git a/_data/documents.yml b/_data/documents.yml index 31770d0..54f18d6 100644 --- a/_data/documents.yml +++ b/_data/documents.yml @@ -1756,7 +1756,7 @@ ### Analysis notes - name: ananote_template_final.pdf - title: The Analysis Note Template + title: The Analysis Note Template (PDF) tags: [ananote_template,] author: G.David venue: @@ -1765,6 +1765,16 @@ format: pdf date: +- name: ananote_template_final.tex + title: The Analysis Note Template (TeX) + tags: [ananote_template,] + author: G.David + venue: + type: document + category: physics + format: tex + date: + ## -- ATTIC -- # Replaced local "asset" file with a reference to Zenodo (created for this purpose) diff --git a/assets/documents/ananote_template_final.tex b/assets/documents/ananote_template_final.tex new file mode 100755 index 0000000..d9c28d8 --- /dev/null +++ b/assets/documents/ananote_template_final.tex @@ -0,0 +1,669 @@ +\documentclass[12pt]{article} +\usepackage{graphicx} +\topmargin = -15mm +\textheight = 230mm +\textwidth = 155mm +\evensidemargin = 0mm +\oddsidemargin = 0mm +\setcounter{topnumber}{300} +\setcounter{bottomnumber}{300} +\setcounter{totalnumber}{300} + +\begin{document} + +\begin{titlepage} +\begin{center} + {\LARGE Template for PHENIX Analysis Notes + } \\ +\vspace{15mm} +Gabor David$^{\mbox{a}}$, Yasuyuki Akiba$^{\mbox{b}}$ + +\vspace{1cm} + +$^{\mbox{a}}$ SBU / BNL \\ +$^{\mbox{b}}$ RIKEN \\ +\end{center} + +\vspace{2cm} +\date{today} + + + +\vspace{2cm} + +%\begin{document} + +%\maketitle + +\begin{abstract} + +This write-up gives some guidelines/suggestions how to +structure analysis notes, what information to include and to what +depth. It is not meant to be a fill-out form or checklist, the more +so, since no two analyses are similar. Instead, it's trying to convey +the spirit in which notes should be written. Namely, always ask +yourself: ``Had I not been doing this analysis myself, would I be able +to reproduce it and come to the exact same results based solely upon +what's written in the analysis note?'' If the answer is negative, +please keep working on it. A positive answer means at the very +least three things: 1/ all inputs (data, calibrations, recalibrators, +deadmaps, ect.) are clearly defined (timestamp, version) or stored +2/ all codes, macros, scripts etc. used in the analysis +are preserved safely, long-term (not in a volatile private directory) +and 3/ the flow of the analysis is clearly described along with all +auxiliary information needed to repeat it. + +\end{abstract} + +\end{titlepage} + + +\tableofcontents + +\newpage + +\vspace{1.0in} +{\large {\bf General guidelines}} +\vspace{0.5in} + +The overwhelming majority of PHENIX Analysis Notes are accompanying +new physics results extracted from data. Before making them public, +as preliminary or final, the submission of a comprehensive analysis +note is a formal requirement, too. In reality, the amount and depth +of information in the actual notes is quite disparate. While there's +no way (and no point) to squeeze notes of the most diverse content +into a single ``checklist'', or ``fill-out form'', we think some +minimum should be adhered to. This template is attempt to guide you +what this minimum should be (more information is always welcome!). +Also, while writing your note, you should adhere to the +following principles. + +\begin{itemize} +\item{{\bf Reproducibility.} + The purpose of an analysis note is to ensure that our published + results are reproducible by {\bf any} reasonably knowledgable PHENIX + person. + Reproducibility means that at the very least you give a full + list of runs analyzed, production tag, analysis (train) code + location and tag, analysis macros with location, QA criteria, cuts + applied and procedures described comprehensibly. Err on the side + of giving too much detail. {\bf Omit} information referring to an earlier + note {\bf only} if that analysis did + the {\bf exact} same thing as you {\bf and} if the description in it is + truly comprehensive. Even a + {\bf non-expert in your topic} should in principle be able to + {\bf repeat} your analysis and come up with the {\bf same result}.} +\item{{\bf Provide your own systematic error estimates.} + Unless you are analyzing the exact same data with the exact same + code, cuts and procedures for the exact same physics quantity (but + then, why would you do that???) your systematic errors are {\bf not} + the same as your buddy's. If you estimated them independently and + came up with the same number - that's fine, but show (prove) it! + Statements like ``X got this number so this is my error, too'' are not + acceptable.} +\item{{\bf Explain differences from previous analyses.} + If a similar analysis has already been done before, and now you + come up with different results or systematic errors, include a + section in which you analyze/explain + why did things change (hopefully improve...)} +\item{{\bf Point out problems.} + Be honest. Saying that you didn't find the solution for a + particular problem (and actually pointing out that problem instead + of glossing over it) is both honest, ethical - and useful for the + collaboration. Pretending that it doesn't exist, or that you solved + it, even when you yourself know that's not true is the exact + opposite. - If you found a mistake in an earlier analysis note, + point it out, politely!} +\item{{\bf Start early.} + Ideally analysis notes are written {\bf in parallel with the + actual analysis}, or at least when you are no more than halfway + through. Yes, that means sometimes re-writing text, + correcting numbers a dozen times, but at least at the end what you + wrote down will be correct - which is not necessarily the case when you + just finished your analysis on time to meet a deadline but for + approval you still have to submit an analysis note...} +\item{{\bf Make your note readable}. + Give meaningful section (subsection...) titles and + make a table of contents. Some otherwise excellent notes are + lacking this ``feature''. Ever got frustrated trying to find two + relevant sentences or a crucial number in 30-60-170+ pages of + unstructured and un-indexed text? + Also, since analysis notes are often long and sometimes hard to + navigate, it is a good practice to give the exact location (pages) + {\it within} a reference (``as described on pp. n-m of ANxxx''). + You actually looked it up, right? Then this shouldn't be a burden.} +\item{{\bf Catch the spirit.} Analyses are quite complex, and it is + hard, if not impossible, to write a set of formal rules, a + checklist, which make sure that your analysis will indeed become + fully reproducible. If in your case something more is needed than + described here, please, please, add it! {\bf Science is + reproducibility}, and you are a scientist, aren't you? So please, + act like one.} + +\end{itemize} + + +\vspace{1.0in} +{\bf Selected ``no-no''-s} + +To avoid being beaten up or lawsuits, here's the disclaimer: none of +these examples are taken from actual notes, I just made them up. +Unfortunately similar things can sometimes be found in actual notes. + +\begin{itemize} +\item{``We evaluated error X by a fast MC and it was negligible''. + {\it (Which fast MC? What parameters, input? How did you + evaluate? How much is negligible?)}} +\item{``We discarded runs with a bad RP distribution.'' + {\it (Quantify what's ``bad''!)}} +\item{``We extracted the signal the same way as in Run-2''. + {\it (At the very least give a reference where the Run-2 analysis is + described in detail. Second, it's hard to imagine that you did the + exact same thing. I hope you didn't, that you learned in the + meantime, your procedures became more sophisticated. But then they + are different.)}} +\item{``In this step we used NN's code/libraries''. {\it (Using other + people's libraries -- unless, of course, they are recognized part of + the official framework -- is frowned upon. If absolutely + unavoidable, specify the exact location and version, but a better + solution is to copy the source and make it part of your own backup + (HPSS or CVS). This also protects you if later NN decides to change + something in his/her code.)}} +\item{... etc.} + +\end{itemize} + + +\newpage + +%\setcounter{section}{-1} +\label{sec:organization} + +\section{Analysis Organization} +label{sec:organization} +%This is a new section that we (PHENIX PM) want you to fill out to +This is the section where you +describe how your analysis is organized. It serves as a concise +list, a catalogue of the steps of your analysis, the location of all necessary +input and output files as well as all necessary codes, macros to do +your analysis from scratch. +It is meant to enable a third person to look later into your +analysis and have access to everything needed to reproduce your +results, if necessary. Basically it is about preserving the +{\bf data} and {\bf software} necessary. +Detailed description and explanation of the +individual analysis steps -- the {\bf know-how} -- +comes in the later part of the analysis note where +we still suggest some of the information to be repeated. + +This section should include the following information. +\begin{description} +\item [A flowchart of the major analysis steps with code locations]~\\ +The flowchart is a birds-eye view of your analysis flow, with +references to the basic working directories. +``Code location'' here means your working directories on RCF. +Ideally these all should branch off a single {\bf base directory} (which +also makes backing up the ``snapshot'' -- see below -- in HPSS easier). +Important: if you did some of the work locally (e.g. on your laptop), +you have to {\bf migrate} the codes and relevant files {\bf to RCF} +when you are finished, and make sure it works there, too. +%\item [Location of your code and working directories in the RCF]~\\ + +Examples of directories to be specified: +\begin{itemize} +\item Directory for Taxi code +\item Directory to run Taxi +\item Directory of the Taxi output +\item Directory for Simulation code +\item Directory for Simulation output +\item Directory for analysis code and macros that analyze the Taxi output +\item Directory for analysis code and macros that analyze the Simulation output +\item Any other directory used during the analysis +\item Directory for the final data file(s) and macros to produce physics plots. +\end{itemize} +We suggest you write these directories as they are at the time you +finished your analysis, and you reference them in the later part of +the analysis note. After they are backed up in HPSS, part or all of +it can be removed. +%We understand that they can be removed later. But it is not a problem, since +%we require that you should back up all of these directories to be backed up in HPSS. + +\item [Location in HPSS of your analysis snapshot]~\\ +We request that a snapshot of your analysis, namely, all of the +directories and the files in the directories described above, are +backed up in HPSS. +%Ideally those directories should be {\bf clean} in the +%sense that they contain {\bf all those files} that are needed to +%repeat the analysis from scratch, but {\bf only those files} -- +%however, cleaning them up is not a formal requirement. (Obviously +%cleaning up poses the danger that you delete something that's actually +%necessary for the analysis.) + +%Please back up all your working directories +%described above to the HPSS. + +%We request that all of your code and intermediate data files of +%the analysis should be backed up. +Before the backup, make a good faith effort to {\bf clean up} the +directories (such that someone later looking at it isn't unnecessarily +distracted). Log files, for instance, can be deleted. +However, make sure that they contain {\bf all those files} that are +needed to repeat the analysis from scratch. +{\bf Avoid symlinks!} If you +aggregated the taxi output (e.g. with {\it haddPhenix}), you can +make a text file of the run numbers processed (runlist), then delete the +individual run outputs. Same for simulations. However, keep a small +fraction of the original files (and include them in the backup). +%However, if the total amount of the Taxi output or Simulation output +%is too large, it is sufficent to back up a fraction of them. +%For example, if there are 1,000 simulation nDSTs of 1GB each (Total 1TB), +%it would be suffice to back up 100 nDSTs +Then please document the location of your analysis snapshot +in HPSS here. + +{\bf Note:} At some point in the future the Data Preservation Task +Force is planning to centralize the location of those analysis +snapshots into one place. In order to make this easier, in addition +to documenting the HPSS location of your snapshot in the analysis +note, please also send an email indicating this location to Chris +Pinkenburg, Maxim Potekhin, Takahito Todoroki and Gabor David. + +{\bf Dependencies.} +If you are aware that your code uses libraries other than the ones +created from your own code or are part of the core PHENIX software, +please point out this fact here. Ideally the source code for those +other libraries should also be included in your backup. + +{\bf Remember}, working directories are handy, easy to + access, but they can easily come and go, be deleted, altered, so a + copy of the snapshot of your analysis - files, codes, macros - at + the time of preliminary request or the final publication + should exist in HPSS/CVS. + + If you don't know how to back up a directory tree to HPSS, you'll find + instructions at the end of this section. + +\item [Brief summary of your analysis]~\\ +State the physics observable analyzed, the basic method and the +subdetectors involved. Then please write a brief description +(overview) of the steps of your analysis and in +which your working directory each step is done. +These are essentially bullets explaining the flowchart and adding some +basic information, with pointers to working directories. +Detailed description comes in later sections of the analysis notes. +\item [Data Set] ~\\ +\begin{itemize} +\item Run, beam species, energy, trigger(s) +\item Good run list and its location. (Don't put the good run list +itself here. If you want to include it in the analysis note, too, put +it in an Appendix) +\item Number of events analyzed and the corresponding number of MB +events or integrated luminosity +\end{itemize} +\item [Taxi] ~\\ +\begin{itemize} +\item Taxi code and its CVS location, including version number +\item List of recalibrators used for the analysis. Are all +recalibrators public? If there is any private recalibrator used, +please list them and their location. +\item List of files like efficieny map or data for recalibrator +\item Taxi output +\end{itemize} +\item [Simulation] ~\\ +Simulation code and any related files +\begin{itemize} +\item List of simulation code used for the analysis +\item Event generator(s), including tuning/parameter files +\item PISA setup file +\item Tuning of PISA +\item Output of simulation (PISA, PYTHIA, etc) +\item List and location of code to analyze the simulation data and its output +\item Any other pertinent information +\end{itemize} +\item [List of major intermediate analysis files] ~\\ +There are several steps in the analysis from Taxi run to the final +data file. You will produce intermediate analysis file in each +step. Please list major intermediate files produced by the +analysis. Remember: reproducibility means that {\bf each} step of your +analysis process can be unambiguously traced. For example + +\begin{itemize} +\item Code to produce the raw data histogram of pi0 from Taxi output +\item ROOT file of the raw data histogram of pi0 pT distribution +\item ROOT file of PISA simulation for pi0 efficiency calculation +\item Embedding code/macro +\item Code to analyze the PISA simulation file to produce pi0 efficiency +\item ROOT file of the efficiency vs pT of pi0 +\item Macro to calculate the cross section from the raw histogram and efficiency +\item Any other pertinent information +\end{itemize} +\item [Final data file and plotting macro]~\\ +We now request that final physics data for figures for a journal paper and preliminary plot should be +put in ROOT files, and the figures/plots are produced by plotting macros from these final data files. +We request that all of the ROOT files of plotted data and the plotting macros are in a single +directory. +Please list the final data files, plotting macros, and their location here. +Please indicate which macro will produce with plot from which data file.\\ + +In addition you have to provide your {\bf final data points} with +uncertainties in {\bf text files} (ASCII), and specify their +location. These files are needed because 1/ we have to post the data +on our public website 2/ we want to upload them in HEP databases. +(While strictly speaking this is not a requirement for preliminary +data, it is good practice to provide ASCII files from those, too.) + +Example:\\ +The data file and plotting macros of the preliminary request is located at +\begin{verbatim} +/phenix/u/analyzer/analysis/prelim2019.10 +\end{verbatim} +\verb*|DataFile.root| The data file of all plots requested for preliminary\\ +\verb*|plot_xs_pi0.C| Macro to produce Plot1 (cross section of pi0).\\ +\verb*|plot_RAA_pi0.C| Macro to produce Plot 2 (RAA of pi0)\\ + +The ROOT files of the plotted data and the plotting macros should be backed up in HPSS as a part of +the snapshot of your analysis. + +We request that the directory of the ROOT files of the plotted data and the plotting macros +should be under your user directory (\verb*|/phenix/u/analyzer| ) so that it is in RAID part +of RCF filesystem. We request that this final data directory should be "frozen" +at the time of preliminary request or final paper publication, except for editing of the plotting macros +for cosmetic changes of the plots. +Important: this directory should be duplicated on the rcas disk, and +be part of your analysis snapshot, too (backed up in HPSS). + +\item[How to put a directory tree in HPSS?]~\\ +First make a .tar file of the directory tree, +preferably on the PHENIX scratch area (if you don't already have a +directory there, you can create one) \\ +\begin{verbatim} +cd /gpfs/mnt/gpfs02/phenix/scratch/david +\end{verbatim} +then do the equivalent of this +\begin{verbatim} +tar -cvM -f plhf1_david_taxi.tar /gpfs/mnt/gpfs02/phenix/plhf/plhf1/david/taxi +\end{verbatim} +While sitting in this (scratch) directory, log in to HPSS (hsi +command, the password is the one you use for rcf3, rcas) +\begin{verbatim} +[david@rcas2064 david]$ hsi +Warning: "Network Options" section is empty in HPSS.conf file +nd_krb_preexist_auth: No credentials cache found retrieving principal +name from cred.cache +nd_kerberos_auth: No credentials cache found on krb5_mk_req call +Kerberos Principal: david +Password for david@SDCC.BNL.GOV: +Username: david UID: 1113 Acct: 1113(1113) Copies: 1 Firewall: off +[hsi.5.0.2.p3 Mon Jun 29 16:29:32 EDT 2015] +? pwd +pwd0: /home/david +\end{verbatim} +You end up in your HPSS directory. Create a subdirectory +for the new full backup (in this case it is my old taxi directory), +then ``mput'' the tar file in HPSS (per default ``mput'' will copy files +from the directory where you have been when you logged in to HPSS with +the hsi command): +\begin{verbatim} +? mkdir taxi_20190828 +mkdir: /home/david/taxi_20190828 +? cd taxi_20190828 +? mput * +mput 'plhf1_david_taxi.tar'? ([Y]es,[N]o,[A]ll,[Q]uit) y +Active: 1, Queued 0 [In-flight: 193.25GB Xferred: 1.10GB 0.57%] [Done: 0 F +\end{verbatim} +and it will keep showing you the progress how the file is uploaded in HPSS. +Since the file first goes to a huge buffer-disk, the transfer is +pretty fast. It ends with this message, when you log out: +\begin{verbatim} +mput 'plhf1_david_taxi.tar' : +'/home/david/taxi_20190828/plhf1_david_taxi.tar' +( 193246924800 bytes, 220612.0 KBS (cos=13)) +? ls +/home/david/taxi_20190828: +plhf1_david_taxi.tar +? q +[david@rcas2064 david]$ +\end{verbatim} +That's all, you are back on rcas, in the scratch directory. Be nice +and delete the .tar file right away (don't wait for the automatic +deletion). +\end{description} + +\newpage + +\section{Introduction} + +{\it + Define the purpose and scope of your analysis. Describe (briefly!) + the physics you are after and how/why is the signal you extract + relevant to it. If this is not the first time such signal is + analyzed, refer (briefly!) to earlier PHENIX work and point out + what's different (method? dataset? statistics? better simulations? + new insight in some bias? bugs found and fixed? ...) + Describe (briefly!) the improvements you made and the major + unresolved issues. +} + +{\it In what follows, keep in mind that your analysis has to be + {\bf reproducible}. If you fulfilled requirements described in + Sec.~\ref{sec:organization} it means that the {\bf software} environment is + preserved. But that's obviously not sufficient. + Now you have to make sure that the {\bf know-how} is + preserved too, the ``recipees'' how to do the individual steps, + proper sequence, dependencies are spelled out. In some cases this + is trivial, but it can also get very tricky, interconnected, + iterative; you have to make sure, that a third person can + understand, what to do with the stored software. +} + +\section{Data set, QA, general cuts} + +{\it + Decribe precisely your input dataset (a table of all runs analyzed + should appear in the Appendix~\footnote{Yes, this still leaves the + question of segments open, but that would be overkill}): + data taking period, master DST/pDST production tag, train number. + Describe your run selection (QA) criteria, preferably with + justification. Plot the relevant global quantities (centrality? + reaction plane? multiplicity? ...) from the runs passing your QA. + + If you have event selection criteria, list and justify them. Show + the relevant global quantities before and after selection. If you + use triggered data, show the turn-on curve. + + Summarizing in a table what fraction of runs/events were eliminated + by different cuts is good practice. + + Plot live/dead areas (maps) in the relevant detectors. Indicate if + they influence acceptance in a $p_T$- or charge-dependent way. + Indicate if they are fixed over the dataset or run-dependent. + +} + +\section{Data Analysis} + + +%\subsection{CVS location of the codes and macros} +%\label{CVS} + +%{\it +% Give exact location(s) - not just ``it's in CVS'' - of all your codes and +% macros used in the analysis. Include version number and time when +% your final scan was done (DB contents!). It should be ready to use: +% if there are some special requirements to run it, please add a README. +% Don't forget simulation codes (event generator, PISA, response, +% reconstruction, evaluation codes, fast MC, whatever you used). +% Include special (parameter) files used in your simulations, if any. +% If you used embedding code, provide its location, too. +%} + +\subsection{Analysis chain} + +{\it + List the steps of your analysis (typically this will be a much + longer list than the one in Sec.~\ref{sec:organization}) and + point to the subsections where they are described in detail. If you + are doing something new/unusual, point out this fact here. +} + +{\it As said before, there's no way to write up a general, mechanical + ``checklist''. Use your judgement and always try to think along the + lines: ``If I were the outsider, would I fully understand what has + been done?'' If not, keep working on the analysis note. +} + +\subsubsection{Analysis step 1} + +{\it + Give a meaningful title to the subsection. Provide a reference to + the directory where this step has been done and what macros, scripts + were run, with what parameters. + Define your cuts precisely, and plot its + effects (before/after). For log distributions consider ratios. If + you make a fit, always give the precise functional form, its parameters + and plot the data/fit ratio. +} + + + +\subsubsection{Analysis step 2, ... n} + +{\it + In general, plot intermediate results. It is much easier to believe + not only your ideas, but the actual work you have done if one sees + how your raw data evolved in the individual steps into the final, + fully corrected results. +} + +\subsubsection{Simulations} + +{\it + Your simulations should also be reproducible: provide all pertinent + information (generator, acceptance cuts, ranges, ...). If + simulation output has to be modified (like an additional smearing), + explain. Provide plots that show how well simulations describe real + data (like simulated and real peaks, widths). + Provide location and tag of the embedding code (if any). +} + + +\subsection{Systematic errors} + +{\it +{\bf Make your own estimate} for each error related to your own analysis, +except for ``external'' quantities you take from someone else (centrality, +reference spectrum...) - for those give exact references. If you +claim some error cancels in a composite quantity computed from your +and an ``external'' result (like a p+p denominator in a heavy ion +$R_{AA}$) make sure and prove with a reference that the relevant parts +of the analyses were done the exact same way (often they were not!). +} + +\subsubsection{Error source 1} + +{\it + Describe the error source, explain, what type it is (A,B,C,...) and + whether it is centrality-dependent, absolute or relative. Explain + how you estimated its contribution, and whether your estimate is + full extent, $\sigma$ or something else. Include a plot from which + the reader can pass his/her own judgement on your estimate. Always + give a specific number or upper limit (not just ``it's small''). + Your errors should be clearly defined here (if necessary, listed in + a table) for all data points. + + Confessing that a particular estimate is ``soft'' is not a shame + (even if the PWG grilles you for it). + +} + +\subsubsection{Error source 2, ...n} + +{\it + If possible, for $p_T$-correlated errors provide an estimate how is + it (anti)correlated; a function is even better. Justify it. +} + +\subsubsection{Summary of systematic errors} + +{\it + Provide a table that summarizes your systematic errors. If such + table would become too big, you may want to select only a few + points, representative of your entire range ($p_T$, centrality,...). + + +} + + +\subsubsection{(If relevant: changes in error estimates since earlier results)} + +{\it + If this is not the first analysis of its kind (preliminary to final, + re-analysis of a dataset, etc.) and if your systematic errors differ + significantly from the ones found before, discuss in detail what + changed (method? assumptions? new insight? etc.) and why do you + think the new estimate is better than the previous one. +} + +\subsubsection{(If relevant: how to improve systematic errors in the future?)} + +{\it + This could be one of the most useful parts of your note. + During your analysis you often got frustrated, because you couldn't + improve on something due to lack of specific data, software + capabilities, unresolved detector problems, etc. Discuss them, + particularly if they turned out to be your dominant errors, and + suggest improvements. + Try to be specific, beyond generalities like ``we should do more + simulations'' or ``we need better hadron rejection''. +} + + +\section{Results} + +{\it + Plot all your results with clear captions. If there are too many + (say, more than 10-12 pages) break them + up into subsections for easier reading/navigation. If you use + ``external'' data (like reference spectra), plot them separately, + too. If previous results of the same quantity exist, plot old/new, + and if there are big discrepancies, describe briefly the reason (even + if you discussed it already in detail in the analysis section). +} + +\subsection{Discussion, comparison to theory} +{\it + If possible, relevant and available, compare your results to theories, + plot and discuss the results. This is not a paper to be published, + so feel free to wander off in less-than-mainstream directions as well. + They might actually give someone an idea... +} + + +\section{Appendix} + +\subsection{Data tables} +{\it + No results without data tables! List errors separately + (statistical, type A, B, C) and the total. If you have many tables, + breaking them up into subsections is strongly recommended. + Any analysis note without data tables will be immediately rejected. +} + + + +\subsection{(Runlist)} + + + +\begin{thebibliography}{9} + +\bibitem{ref1} Your reference +{\rm http://www.phenix.bnl.gov/yourreference/ref1.pdf} + + +\end{thebibliography} + +\end{document} + + +