- Nominal scale choice: try to find saddle point (xsection stationary wrt scales). If not, stick to conventions (pT(gamma) or M(gamma gamma) for single/double photons). Vary scale by factor 1/2, 2.

- Last slide : work in progress, enhanced direct-fragmentation component at NLO in DIPHOX (enhance low delta phi xsection by ~1.6)? (very preliminary)

- New photon + b/c results CDF / D0 with ~full Tevatron dataset (arXiv:1203.5865). More results on full dataset by the summer.

- Pythia with increased gluon splitting to heavy quarks agrees better with data

- Update on diphoton production by CDF with 9.5fb-1. Comparison with many generators : Pythia, Resbos, Diphox, MCFM, Sherpa

- Comparison of previous diphoton CDF analysis (5.4 fb-1) and 2gNNLO => very good (impressive) agreement in almost the whole kinematical range

- New result on gamma+jet with 37b-1 was presented. Data is still lower than Jetphox at low ptgamma. Comparison with Jetphox, Pythia born, Pythia born+frag, Sherpa. Some discrepancies (data/Jetphox) for very forward jets, although not significant. Best agreement with full sim is provided by Sherpa.

- Some studies on the differential cross section for different isolation criteria by Philippe : up to 15% difference in some kinematical regions when changing parton-level isolation cut by +- 2GeV .

- isolated photon measurements from colliders: 35 measurements (~400 data points) at LHC/TeV/SppS/RHIC

- Reduction of the gluon pdf uncertainty by 20% by inclusion of photon measurements (and thus the Higgs gluon fusion cross-section uncertainty). Inclusion of photon data in global fit would be useful.

- Jetphox/Diphox presentation. Both can now output ROOT ntuples and provide PDF error band in one run. Future version of Diphox to contain improved calculation of direct-fragmentation component mentioned by Michel

- Resbos : more flexible than in the past (different pdf available, different isolation criteria). Should be easier now to get grids for PDF uncertainties. Xsection matched to NLO at high pT => ~10% uncertainty?

- Sherpa : advocating for measurements with the democratic approach (photons inside jets)

- Grappa : working on a NLO+PS with consistent photon fragmentation. Box is not included yet. Needs further development

- 2gNNLO : NNLO parton-level calculation. Uses MSTW2008 NNLO PDFs. Remarkable agreement with Tevatron and LHC results. Still using smooth Frixione isolation. Showing that isolation criterion does not matter much for the total cross-section if the cut is tight enough. Not yet available for user (work in progress). Plans also to include the fixed cones and fragmentation function in (later) future.

- Issues in matching theoretical predictions with data in photon measurements :

- discussion about PU subtraction with Fastjet : OK as long as correcting data with data and observable is clearly defined (also at truth-particle level)

- discussion about the difficulty of matching isolation prescription at parton-level and detector level => UE correction : still keep the same methodology (ratio between PS MC with/without UE+hadronization. Use different UE tunes). ATLAS: apply fast-jet technique to remove/reduce UE both at experimental level (reconstructed soft jets) and truth-particle level (truth-particle jets)

- Testing Frixione isolation at NLO and in Pythia to kill fragmentation contribution

- Able to kill fragmentation contribution at NLO, but don't know the direct photon efficiency (too tight cuts => total cross-section not reliable)

- Difficult to kill fragmentation in parton-shower with high signal efficiency. Is it due to UE/hadronization? Definition of fragmentation in Pythia?

- non-isolated xsections, differential cross section versus isolation cut => sensitive to fragmentation function

- Ratios 8/7 TeV ? would be not interesting for gluon pdf but would reduce the scale uncertainties