<div>Hi Graeme,</div><br><div>Sorry to be coming late to this.</div><br><div>I absolutely agree that we should get rid of the fake "unit bin width" for unbinned quantities -- that was an ad hoc solution from deep in prehistory to HepData's fundamental data type being binned, and now that's not the case we should move away from it. Writing out HepData 1D data types to YODA as a Scatter1D would be best (and then, as Jon as noted before, we just need to make sure that those get plotted in a reasonable way).</div><br><div>The level of disagreement between HepData and Rivet records is staggering, and I'm sorry that we've not had the manpower to deal with that. Holger was previously working on it, and I think found similar numbers, but then moved institute (to FNAL) before he could act on it; maybe there is still some possibility to follow it up, Holger?</div><br><div>Since you have diagnostics scripts operational at the moment, Graeme, could you produce a list/summary of how many records only differ by the <em>binning</em> of observables, and maybe another with a <em>loose</em> consistency check on bin contents? That would give us some insight into how many differ because the Rivet files "just" need to be synced with HepData, and how many have completely inconsistent dataset naming. The latter are the ones that will need some analysis code updates.</div><br><div>I appreciate that this task maybe falls into a gap between purely clerical and something requiring physics insight. However, if we can manage to isolate a purely clerical component then help from Joanna would be really valuable: as you're aware we have no dedicated manpower whatsoever on the Rivet side, and tasks of this sort are the hardest to motivate postdocs to work on. I am also pessimistic that we can get help from the LHC experiments (who supplied us with the inconsistent Rivet/HD records in the first place...), but we can try: breakdown of the numbers by experiment would also be very useful.</div><br><div>Thanks,</div><div>Andy</div><br><div><signature id="initial"><div><table cellpadding="0" cellspacing="0"><tbody><tr><td colspan="2"><div style="padding-bottom:15px;"><div><strong>Dr Andy Buckley, Lecturer / Royal Society University Research Fellow</strong></div><div>Particle Physics Experiment Group, University of Glasgow</div></div></td></tr><tr><td style="vertical-align:top;"></td><td><div style="font-size:0.9em;white-space:nowrap;border-left:2px solid gray;margin-left:20px;padding-left:20px;"><div><div></div><div></div></div></div></td></tr></tbody></table></div></signature></div><div class="gmail_quote_attribution">On Jun 15 2018, at 2:20 pm, Graeme Watt <Graeme.Watt@durham.ac.uk> wrote:</div><blockquote><br><div><div> Dear Peter (and Rivet developers),</div><br><div>No, I don't think this is something that Joanne (or Keith) could help with.</div><br><div><a href="https://github.com/HEPData/miscellaneous/blob/master/scripts/rivet-diffhepdata-all" title="https://github.com/HEPData/miscellaneous/blob/master/scripts/rivet-diffhepdata-all">https://github.com/HEPData/miscellaneous/blob/master/scripts/rivet-diffhepdata-all</a></div><br><div>I wrote another script "rivet-diffhepdata-all" that loops over all Rivet analyses listed in <a href="http://rivet.hepforge.org/analyses.json" title="http://rivet.hepforge.org/analyses.json">http://rivet.hepforge.org/analyses.json</a> and compares each Rivet .yoda file with the HEPData download. It calls functions from the previous "rivet-diffhepdata" script which in turn calls "yodadiff".</div><br><div>I added the URL option to pass the Rivet analysis name to HEPData when requesting a YODA conversion, e.g.</div><br><div><a href="https://hepdata.net/record/ins319520?format=yoda&rivet=ALEPH_1991_S2435284" title="https://hepdata.net/record/ins319520?format=yoda&rivet=ALEPH_1991_S2435284">https://hepdata.net/record/ins319520?format=yoda&rivet=ALEPH_1991_S2435284</a></div><br><div>This is necessary, for example, for Rivet analysis names containing the SPIRES ID rather than the INSPIRE ID. The "rivet-mkanalysis" script could be modified accordingly:</div><br><div><a href="https://rivet.hepforge.org/trac/browser/bin/rivet-mkanalysis#L139" title="https://rivet.hepforge.org/trac/browser/bin/rivet-mkanalysis#L139">https://rivet.hepforge.org/trac/browser/bin/rivet-mkanalysis#L139</a></div><div>hdurl = <a href="http://www.hepdata.net/record/ins%s?format=yoda&rivet=%s" title="http://www.hepdata.net/record/ins%s?format=yoda&rivet=%s"><span style="color:red"><strong>MailScanner has detected a possible fraud attempt from "www.hepdata.net" claiming to be</strong></span></a><a href="http://www.hepdata.net/record/ins%s?format=yoda&rivet=%s" title="http://www.hepdata.net/record/ins%s?format=yoda&rivet=%s"><font color="red"><b>MailScanner has detected a possible fraud attempt from "www.hepdata.net" claiming to be</b></font> "http://www.hepdata.net/record/ins%s?format=yoda&rivet=%s"</a> % (ANAINSPIREID, ANANAME)</div><br><div>The web directory <a href="http://ippp.dur.ac.uk/%7Ewatt/RivetDiffHEPData/Rivet-2.6.0/" title="http://ippp.dur.ac.uk/%7Ewatt/RivetDiffHEPData/Rivet-2.6.0/">http://ippp.dur.ac.uk/~watt/RivetDiffHEPData/Rivet-2.6.0/</a> contains the output of running:</div><br><div>rivet-diffhepdata-all -r ../Rivet-2.6.0/analyses -d HEPDataYoda -o YodaDiffOutput > rivet-diffhepdata-all.txt</div><br><div>The summary line doesn't look too promising:</div><br><div>"Of 359 Rivet analyses in ../Rivet-2.6.0/analyses, 66 (18.4%) were compatible and 293 (81.6%) were incompatible."</div><br><div>Here, compatibility is defined as a zero exit status returned by "yodadiff". Only 31 of these 293 incompatible analyses are missing a HEPData record.</div><br><div>Note that the "yodadiff" script gives a ZeroDivisionError in the function "eq(a, b)" when "a" and "b" have opposite sign due to the return value:</div><br><div>return abs(float(a) - float(b))/(float(a) + float(b)) < opts.TOL</div><br><div>For example, d01-x01-y01 of ATLAS_2013_I1190187.yoda distributed with Rivet has yerr- = 6.8 and yerr+ = -6.8. The HEPData table ( <a href="http://www.hepdata.net/record/ins1190187?version=1&table=Table1" title="http://www.hepdata.net/record/ins1190187?version=1&table=Table1">http://www.hepdata.net/record/ins1190187?version=1&table=Table1</a> ) gives both yerr- and yerr+ as 1.212930e+01, so it seems that the Rivet .yoda file contains only the statistical error (with a wrong sign for yerr+). The Rivet .yoda file also assigns an artificial bin width of 1 for sqrt(s), whereas the HEPData table does not assign a bin width for sqrt(s). Looking at the YODA export from the old HepData site:</div><br><div><a href="http://hepdata.cedar.ac.uk/view/ins1190187/d1/yoda" title="http://hepdata.cedar.ac.uk/view/ins1190187/d1/yoda">http://hepdata.cedar.ac.uk/view/ins1190187/d1/yoda</a></div><br><div>again there are zero xerr- and xerr+ values, but the AIDA export:</div><br><div><a href="http://hepdata.cedar.ac.uk/view/ins1190187/d1/aida" title="http://hepdata.cedar.ac.uk/view/ins1190187/d1/aida">http://hepdata.cedar.ac.uk/view/ins1190187/d1/aida</a></div><br><div>has a unit bin width written by "AidaFormatter.java" with a comment (by Andy?):</div><br><div>// If there's only one bin and it has no width, give</div><div>// it unit width so that it can be filled and the height</div><div>// doesn't go mad when Rivet tries to use it.</div><br><div>I would argue that any construction of artificial bin widths would be better handled on the Rivet side rather than in the HEPData export of the data. Removing artificial bin widths from the Scatter objects in the Rivet .yoda files would resolve some of the incompatibilities between HEPData and Rivet, but many more would remain. I don't propose a universal solution for now, but just want to start by identifying where the differences lie. It will be interesting to monitor whether the degree of incompatibility, now easily quantified by my new "rivet-diffhepdata-all" script, improves for subsequent Rivet releases.</div><br><div>Best regards,</div><div>Graeme Watt (HEPData)</div><br><br><div><div>On 25/05/18 23:17, Peter Skands wrote:</div></div><blockquote><br><div><font style="font-size:13px"><font style="font-family:Helvetica, Arial">Hi All,</font></font></div><div><br></div><div><font style="font-size:13px"><font style="font-family:Helvetica, Arial">Agree it sounds like it could make sense as part of Rivet, and could make a real difference by making the step to check consistency almost trivial; easy to run when developing / releasing analyses. In the latter case, feedback could go back to the people submitting a new analysis if there is a discrepancy, which would then put the burden of ensuring consistency mainly on the people who contribute the analyses, without adding significantly to the Rivet / HepData authors. Regarding the existing / old analyses, Graeme, any chance you think of the person in Durham having a go at the backlog of old analyses, or would that be too challenging for her? It might be worth preparing an example, show it to Keith, and see what he thinks?</font></font></div><div><br></div><div><font style="font-size:13px"><font style="font-family:Helvetica, Arial">Cheers,</font></font></div><div><font style="font-size:13px"><font style="font-family:Helvetica, Arial">Peter</font></font></div><div><br></div><div><div><font style="font-family:Helvetica"><font style="font-size:11px">—</font></font></div><div><div><font style="font-family:Helvetica, Arial"><strong>PETER SKANDS</strong></font></div><div><font style="font-family:Helvetica, Arial"><font style="font-size:11px"><font style="font-family:Helvetica">Associate Professor</font></font></font></div><br><div><font style="font-family:Helvetica, Arial"><strong>School of Physics and Astronomy</strong></font></div><div><font style="font-family:Helvetica, Arial"><font style="font-size:11px"><font style="font-family:Helvetica">Monash University </font></font></font></div><div><font style="font-family:Helvetica, Arial"><font style="font-size:11px"><font style="font-family:Helvetica">10 College Walk, Clayton Campus</font></font></font></div><div><font style="font-family:Helvetica, Arial"><font style="font-size:11px"><font style="font-family:Helvetica">Melbourne, VIC 3800</font></font></font></div><div><font style="font-family:Helvetica, Arial"><font style="font-size:11px"><font style="font-family:Helvetica">Australia</font></font></font></div><br><div><font style="font-family:Helvetica, Arial"><font style="font-size:11px"><font style="font-family:Helvetica">T: </font></font></font><font style="font-family:Helvetica, Arial"><font style="font-size:11px"><font style="font-family:Helvetica"><a href="tel://+61%203%20990%2053692" title="tel://+61%203%20990%2053692">+61 3 990 53692</a></font></font></font></div><div><font style="font-family:Helvetica, Arial"><font style="font-size:11px"><font style="font-family:Helvetica">E: </font></font></font><font style="font-family:Helvetica, Arial"><font style="font-size:11px"><font style="font-family:Helvetica"><a href="mailto:peter.skands@monash.edu" title="mailto:peter.skands@monash.edu">peter.skands@monash.edu</a></font></font></font></div><div><font style="font-family:Helvetica, Arial"><font style="font-size:11px"><font style="font-family:Helvetica">W: </font></font></font><font style="font-family:Helvetica, Arial"><font style="font-size:11px"><font style="font-family:Helvetica"><a href="http://skands.physics.monash.edu/" title="http://skands.physics.monash.edu/">skands.physics.monash.edu</a></font></font></font></div></div></div><br><div>On 26 May 2018 at 2:52:23 am, Graeme Watt (<a href="mailto:graeme.watt@durham.ac.uk" title="mailto:graeme.watt@durham.ac.uk">graeme.watt@durham.ac.uk</a>) wrote:</div><br><blockquote><div><div><div>Dear David,</div><br><div>OK, here's another much simpler Python script that downloads the YODA</div><div>file from HEPData and then calls yodadiff:</div><br><div><a href="https://github.com/HEPData/miscellaneous/blob/master/scripts/rivet-diffhepdata" title="https://github.com/HEPData/miscellaneous/blob/master/scripts/rivet-diffhepdata">https://github.com/HEPData/miscellaneous/blob/master/scripts/rivet-diffhepdata</a></div><br><div>I'm not sure if this even warrants a script, given that the</div><div>functionality of:</div><br><div> rivet-diffhepdata ATLAS_2017_I1614149.yoda -i 1614149</div><br><div>could be obtained with two commands, e.g.</div><br><div> curl -L <a href="https://hepdata.net/record/ins1614149?format=yoda" title="https://hepdata.net/record/ins1614149?format=yoda">https://hepdata.net/record/ins1614149?format=yoda</a> | tar zx</div><div> yodadiff ATLAS_2017_I1614149.yoda HEPData-ins1614149-v2-yoda.yoda</div><br><div>The yodadiff script gives more detailed output of differences than my</div><div>previous script, but it does not compare annotations (maybe this</div><div>functionality could be added to yodadiff as an option?). Also, the</div><div>yodadiff script flags additional analysis objects (like covariance</div><div>matrices) that are present in the HEPData YODA file but not in the Rivet</div><div>YODA file (as is the case for ATLAS_2017_I1614149), whereas my previous</div><div>script was specifically written to ignore these differences. I suppose</div><div>this is OK if the user just ignores the warnings from yodadiff about</div><div>additional analysis objects in the HEPData YODA file.</div><br><div>Best regards,</div><div>Graeme</div><br><br><div>On 25/05/18 11:01, David Grellscheid wrote:</div><div>> Hi Graeme,</div><div>></div><div>> Yes, I did mean yodadiff, sorry! There's no way it will start to include</div><div>> a Hepdata download option, it is meant to do one job of comparing two</div><div>> yoda files, which may have nothing at all to do with HEP or Rivet.</div><div>></div><div>> The download option could be provided (in rivet/bin, not yoda/bin!) by a</div><div>> thin layer over the top of yoadiff, though. If you'd like to adapt your</div><div>> script in that way, we'd be happy to include it in the rivet distribution.</div><div>></div><div>> See you,</div><div>></div><div>> David</div><div>></div><div>></div><div>> On 22/05/2018 12:20, Graeme Watt wrote:</div><div>>> Dear David,</div><div>>></div><div>>> Thanks, I think you mean yodadiff (not yodacmp). You're right, it looks</div><div>>> like yodadiff does much the same as my script, other than the HEPData</div><div>>> download and optional comparison of annotations. Maybe these features</div><div>>> could be added to yodadiff? In any case, it would be good to make such</div><div>>> comparisons part of the validation and release procedure of each new</div><div>>> Rivet analysis, which is apparently not happening at the moment (other</div><div>>> than possibly for ATLAS analyses).</div><div>>></div><div>>> Best regards,</div><div>>> Graeme</div><div>>></div><div>>></div><div>>> On 22/05/18 09:54, David Grellscheid wrote:</div><div>>>> Hi Graeme,</div><div>>>></div><div>>>> thanks for your email, we have started discussing it. Just one technical</div><div>>>> point, YODA comes with a yodacmp script already, which addresses many of</div><div>>>> your technical issues with comparisons. Maybe you can use that</div><div>>>> internally in your script?</div><div>>>></div><div>>>> See you,</div><div>>>></div><div>>>> David</div><div>>>></div><div>>>></div><div>>>> On 21/05/2018 20:16, Graeme Watt wrote:</div><div>>>>> Dear Rivet developers,</div><div>>>>></div><div>>>>> I wrote a Python script to compare a YODA reference data file, intended</div><div>>>>> for inclusion in Rivet, with the corresponding YODA file downloaded from</div><div>>>>> HEPData:</div><div>>>>></div><div>>>>> <a href="https://github.com/HEPData/miscellaneous/blob/master/scripts/yoda_compare_hepdata.py" title="https://github.com/HEPData/miscellaneous/blob/master/scripts/yoda_compare_hepdata.py">https://github.com/HEPData/miscellaneous/blob/master/scripts/yoda_compare_hepdata.py</a></div><div>>>>></div><div>>>>></div><div>>>>> Example usage: ./yoda_compare_hepdata.py ATLAS_2017_I1614149.yoda -i</div><div>>>>> 1614149 -a</div><div>>>>></div><div>>>>> This means: compare a local YODA file "ATLAS_2017_I1614149.yoda" with a</div><div>>>>> YODA file downloaded from the HEPData record with INSPIRE ID "1614149"</div><div>>>>> and also compare YODA annotations "-a". Since the HEPData YODA file</div><div>>>>> might contain additional analysis objects compared to the Rivet YODA</div><div>>>>> file, and since there might be inconsequential rounding errors or</div><div>>>>> differences in number formats, comparison using a simple "diff" of .yoda</div><div>>>>> files is not always adequate.</div><div>>>>></div><div>>>>> I had a few problems with the YODA 1.7.0 software when writing the</div><div>>>>> Python script, which could perhaps be improved in future:</div><div>>>>></div><div>>>>> * Calling dump() on Scatter objects does not output the same format as</div><div>>>>> in the input .yoda files, e.g. dumping a Scatter2D gives "HISTO1D" in</div><div>>>>> the output and the central value of the x bin is not output. This might</div><div>>>>> be due to deficiencies in</div><div>>>>> <a href="https://yoda.hepforge.org/trac/browser/src/WriterFLAT.cc" title="https://yoda.hepforge.org/trac/browser/src/WriterFLAT.cc">https://yoda.hepforge.org/trac/browser/src/WriterFLAT.cc</a> .</div><div>>>>> * I expected to be able to check (fuzzy) equality of two Scatter objects</div><div>>>>> using "s == s1", which seems to be implemented in C++ but not in Python.</div><div>>>>> * Checking (fuzzy) equality of two Point2D objects using "p == p1" is</div><div>>>>> implemented in Python, but it only compares the x axes and not the y</div><div>>>>> axes. Similarly, for Point3D objects, the (fuzzy) equality operator</div><div>>>>> "==" only compares the x and y axes, but not the z axes.</div><div>>>>> * In the end, I just copied the definition of "fuzzyEquals" from the C++</div><div>>>>> code into my script and did my own comparisons, without relying on the</div><div>>>>> "==" operator for Point or Scatter objects.</div><div>>>>></div><div>>>>> Recall that Holger Schulz made some similar comparisons in 2016 between</div><div>>>>> YODA reference data files from Rivet and from the old HepData, where he</div><div>>>>> found significant inconsistencies:</div><div>>>>></div><div>>>>> <a href="https://www.hepforge.org/lists-archive/rivet/2016-October/007318.html" title="https://www.hepforge.org/lists-archive/rivet/2016-October/007318.html">https://www.hepforge.org/lists-archive/rivet/2016-October/007318.html</a></div><div>>>>> <a href="https://rivet.hepforge.org/trac/browser/contrib/devscripts/HepDataConsistency" title="https://rivet.hepforge.org/trac/browser/contrib/devscripts/HepDataConsistency">https://rivet.hepforge.org/trac/browser/contrib/devscripts/HepDataConsistency</a></div><div>>>>></div><div>>>>></div><div>>>>></div><div>>>>> While fixing all Rivet/HEPData inconsistencies is probably unrealistic</div><div>>>>> for now, we could at least start by ensuring that analyses added to new</div><div>>>>> Rivet releases include a YODA file that's compatible with the YODA</div><div>>>>> download given by HEPData. My new script should be useful for these</div><div>>>>> purposes. (This work was prompted by a conversation with Peter Skands</div><div>>>>> and Jon Butterworth last month in Durham.) You're welcome to modify my</div><div>>>>> script as you need and include it in a future Rivet release. The script</div><div>>>>> could be run by the experiment contact persons before they upload new</div><div>>>>> analyses to the Rivet "contrib" area. It could also be run (perhaps in</div><div>>>>> a more automated way) by the Rivet developers when moving analyses from</div><div>>>>> the "contrib" area to a new Rivet release. I just ran the script for</div><div>>>>> all the new analyses in the current Rivet "contrib" area (</div><div>>>>> <a href="https://www.hepforge.org/archive/rivet/contrib/" title="https://www.hepforge.org/archive/rivet/contrib/">https://www.hepforge.org/archive/rivet/contrib/</a> ) and it already turned</div><div>>>>> up some useful information:</div><div>>>>></div><div>>>>> * ATLAS_2014_I1310835, ATLAS_2017_I1614149, and ATLAS_2017_I1624693, are</div><div>>>>> all compatible with HEPData.</div><div>>>>> * ATLAS_2016_I1502620 has multiple .yoda files which can't be handled by</div><div>>>>> my script.</div><div>>>>> * ATLAS_2017_I1625109 showed some apparent inconsistencies but the</div><div>>>>> problem looks to be on the HEPData side. The dataset index starts at 0</div><div>>>>> (instead of 1) due to a bug in the hepdata-converter package (which I'll</div><div>>>>> fix) and the year in the analysis name is 2018 (instead of 2017) taken</div><div>>>>> from the journal publication.</div><div>>>>> * ALICE_2017_I1620477 is incompatible with HEPData.</div><div>>>>> * CMS_2016_I1487277 is compatible with HEPData.</div><div>>>>> * CMS_2012_I1111014 and CMS_2016_I1491950 are incompatible with HEPData.</div><div>>>>> * CMS_2017_I1499471, CMS_2017_I1635889, CMS_2018_I1662081 and</div><div>>>>> CMS_2018_I1663958 are all missing from HEPData.</div><div>>>>> * CMS_2017_I1605749 has a HEPData record, but the YODA conversion fails</div><div>>>>> due to a problem with the original HEPData submission.</div><div>>>>> * LHCF_2015_I1351909 is compatible with HEPData, but the annotations</div><div>>>>> differ.</div><div>>>>> * LHCF_2016_I1385877 is incompatible with HEPData.</div><div>>>>></div><div>>>>> I hope this gives you something to discuss at this week's Rivet</div><div>>>>> workshop! :-)</div><div>>>>></div><div>>>>> Best regards,</div><div>>>>> Graeme Watt (HEPData)</div><div>>>>></div><div>>>>></div><div>>>>></div><div>>>>></div><div>>>>> _______________________________________________</div><div>>>>> Rivet mailing list</div><div>>>>> <a href="mailto:Rivet@projects.hepforge.org" title="mailto:Rivet@projects.hepforge.org">Rivet@projects.hepforge.org</a></div><div>>>>> <a href="https://www.hepforge.org/lists/listinfo/rivet" title="https://www.hepforge.org/lists/listinfo/rivet">https://www.hepforge.org/lists/listinfo/rivet</a></div><div>>>>></div><br></div></div></blockquote></blockquote><br><div>_______________________________________________</div><div>Rivet mailing list</div><div>Rivet@projects.hepforge.org</div><div>https://www.hepforge.org/lists/listinfo/rivet</div></div></blockquote>