filters.yaml

- dataset: ATLAS_TTB_DIFF_8TEV_LJ_TPTNORM
  reason: |
   We remove the last bin of the normalised distribution because it is 
   linearly dependent on the others
  rule: "p_TQ<450"

- dataset: ATLAS_TTB_DIFF_8TEV_LJ_TRAPNORM
  reason: |
   We remove the last bin of the normalised distribution because it is 
   linearly dependent on the others
  rule: "yQ<2.05"

- dataset: ATLAS_TTB_DIFF_8TEV_LJ_TTRAPNORM
  reason: |
   We remove the last bin of the normalised distribution because it is 
   linearly dependent on the others
  rule: "yQQ<1.90"

- dataset: ATLAS_TTB_DIFF_8TEV_LJ_TTMNORM
  reason: |
   We remove the last bin of the normalised distribution because it is 
   linearly dependent on the others
  rule: "MQQ<1350"

- dataset: CMSTOPDIFF8TEVTRAPNORM
  reason: |
   We remove the last bin of the normalised distribution because it is
   linearly dependent on the others
  rule: "yQ<2.01"

- dataset: CMSTOPDIFF8TEVTTRAPNORM
  reason: |
   We remove the last bin of the normalised distribution because it is
   linearly dependent on the others
  rule: "yQQ<1.82"

- dataset: ATLAS1JET11
  reason: |
    We keep only the first rapidity bin since the chi2 to the whole set of
    rapidity bins is too large (known problems with correlation matrix).
    This cut will be removed in the new bunch of fits that we are going to do
    for the NNLO jet project.
  rule: eta < 0.3

- dataset: CMS_1JET_8TEV
  reason: |
    We keep only the bins with pTjet>74 GeV because fixed-order theory does
    not provide a good description of the data for smalle values of the	jet
    transverse momentum.
  rule: "p_T2 >= 5476"

- dataset: LHCBWZMU8TEV
  reason: |
    The first two bins in rapidity have unnaturally large K-factors which we
    don't understand, and theory calculations are unstable here, so we remove
    these points at NNLO.
  # Only evaluate rule if PTO matches
  PTO: NNLO
  rule: "etay >= 2.25"

- dataset: LHCBWZMU7TEV
  reason: |
    The first two bins in rapidity have unnaturally large K-factors which we
    don't understand, and theory calculations are unstable here, so we remove
    these points at NNLO.
  PTO: NNLO
  rule: "etay >= 2.25"

- dataset: LHCB_Z_13TEV_DIELECTRON
  reason: |
    The first two bins in rapidity have unnaturally large K-factors which we
    don't understand, and theory calculations are unstable here, so we remove
    these points at NNLO.
  PTO: NNLO
  rule: "etay >= 2.20"

- dataset: LHCB_Z_13TEV_DIMUON
  reason: |
    The first two bins in rapidity have unnaturally large K-factors which we
    don't understand, and theory calculations are unstable here, so we remove
    these points at NNLO.
  PTO: NNLO
  rule: "etay >= 2.20"

- dataset: LHCB_Z_13TEV_DIMUON_5p1fb
  reason: |
    The first two bins in rapidity have unnaturally large K-factors which we
    don't understand, and theory calculations are unstable here, so we remove
    these points at NNLO.
  PTO: NNLO
  rule: "etay >= 2.20"

- dataset: D0WMASY
  reason: |
    Calculation of the NNLO K-factor at the precision required for data points
    for which the asymmetry is small is unfeasible. So we can only fit points
    at NNLO for which the value of the asymmetry is not too small.
  PTO: NNLO
  rule: "fabs(central_value) >= 0.03"

- dataset: D0WEASY
  reason: |
    Calculation of the NNLO K-factor at the precision required for data points
    for which the asymmetry is small is unfeasible. So we can only fit points
    at NNLO for which the value of the asymmetry is not too small.
  PTO: NNLO
  rule: "fabs(central_value) >= 0.03"

- dataset: ATLAS_WP_JET_8TEV_PT
  reason: Avoid the region where small-pT resummation effects become important.
  rule: "p_T >= 30"

- dataset: ATLAS_WM_JET_8TEV_PT
  reason: Avoid the region where small-pT resummation effects become important.
  rule: "p_T >= 30"

- dataset: ATLASZPT7TEV
  reason: Avoid the region where resummation effects become important.
  rule: "p_T2 >= 30**2"

- dataset: ATLASZPT7TEV
  reason: Avoid the region where EWK corrections are important.
  rule: "p_T2 <= 500**2"

- dataset: ATLASZPT8TEVMDIST
  reason: Avoid the region where resummation effects become important.
  rule: "p_T >= 30"

- dataset: ATLASZPT8TEVYDIST
  reason: Avoid the region where resummation effects become important.
  rule: "p_T2 >= 30**2"

- dataset: ATLASZPT8TEVYDIST
  reason: Avoid the region where EWK corrections are important.
  rule: "p_T2 <= 150**2"

- dataset: CMSZDIFF12
  reason: Avoid the region where resummation effects become important.
  rule: "p_T2 >= 30**2"

- dataset: CMSZDIFF12
  reason: Avoid the region where EWK corrections are important.
  rule: "p_T2 <= 170**2"

- dataset: CMSZDIFF12
  reason: |
    removes the last rapidity bin, where we cannot provide a decent
    description due to unknown reasons.
  rule: "etay <= 1.6"

- dataset: ATLASWPT31PB
  reason: Avoid the region where EWK corrections are important.
  rule: "p_T > 30"

- dataset: ATLAS_DY_2D_8TEV_LOWMASS
  reason: Avoid overlap with the high mass ATLAS2DDY8TEV dataset.
  local_variables:
    M: sqrt(M2)
    max_M: 116.0
  rule: M <= max_M
  
#- dataset: CMSDY2D11
#  reason: |
#    Removes data points for which the NNLO K-factors are very large and
#    thus the NLO calculation is unreliable.
#  PTO: NNLO-
#  local_variables:
#    M: sqrt(M2)
#    max_M: 200.0
#  rule: M <= max_M

#- dataset: CMSDY2D11
#  reason: Remove data points for which electroweak corrections are large.
#  PTO: NNLO-
#  local_variables:
#    M: sqrt(M2)
#    min_M: 30.0
#    max_rapidity: 2.2
#  rule: M >= min_M and etay <= max_rapidity

#- dataset: CMSDY2D11
#  reason: Remove data points for which electroweak corrections are large.
#  PTO: NNLO
#  local_variables:
#    M: sqrt(M2)
#    max_rapidity: 2.2
#    max_M: 200.0
#  rule: M <= max_M and etay <= max_rapidity

#- dataset: CMSDY2D12
#  reason: |
#    Removes data points for which the NNLO K-factors are very large and
#    thus the NLO calculation is unreliable.
#  PTO: NNLO-
#  local_variables:
#    M: sqrt(M2)
#    max_M: 200.0
#  rule: M <= max_M

#- dataset: CMSDY2D12
#  reason: Remove data points for which electroweak corrections are large.
#  PTO: NNLO-
#  local_variables:
#    M: sqrt(M2)
#    min_M: 30.0
#    max_rapidity: 2.2
#  rule: M >= min_M and etay <= max_rapidity

#- dataset: CMSDY2D12
#  reason: Remove data points for which electroweak corrections are large.
#  PTO: NNLO
#  local_variables:
#    M: sqrt(M2)
#    max_rapidity: 2.2
#    max_M: 200.0
#  rule: M <= max_M and etay <= max_rapidity

#- dataset: ATLASZHIGHMASS49FB
#  reason: Avoid the region where resummation effects become important.
#  local_variables:
#    max_M: 200.0
#  rule: M_ll <= max_M

# - dataset: LHCBLOWMASS37PB
#   rule: "pT <= maxCMSDY2Dminv"

# - dataset: ATLASLOMASSDY11
#   rule: "False if (pto in (0, 1) and idat < 6) else True"

- dataset: ATLASLOMASSDY11EXT
  reason: Removes data points for which NNLO K-factors are large.
  PTO: NNLO-
  rule: "idat >= 2"

- dataset: DYE886P
  reason: |
    Remove data points for which the fixed-order perturbative expansion is not
    reliable since resummation effects are large. A justification of these cuts
    can be found in arXiv:1507.01006.
  process_type: DYP
  local_variables:
    tau: M2 / sqrts**2
    y0: -0.5 * log(tau)
    maxTau: 0.080
    maxY: 0.663
  rule: "tau <= maxTau and fabs(y/y0) <= maxY"

- dataset: DYE605
  reason: |
    Remove data points for which the fixed-order perturbative expansion is not
    reliable since resummation effects are large. A justification of these cuts
    can be found in arXiv:1507.01006.
  process_type: DYP
  local_variables:
    tau: M2 / sqrts**2
    ymax: -0.5 * log(tau)
    maxTau: 0.080
    maxY: 0.663
  rule: "tau <= maxTau and fabs(y/ymax) <= maxY"

- dataset: DYE605_dw
  reason: |
    Remove data points for which the fixed-order perturbative expansion is not
    reliable since resummation effects are large. A justification of these cuts
    can be found in arXiv:1507.01006.
  process_type: DYP
  local_variables:
    tau: M2 / sqrts**2
    ymax: -0.5 * log(tau)
    maxTau: 0.080
    maxY: 0.663
  rule: "tau <= maxTau and fabs(y/ymax) <= maxY"

- dataset: DYE605_sh
  reason: |
    Remove data points for which the fixed-order perturbative expansion is not
    reliable since resummation effects are large. A justification of these cuts
    can be found in arXiv:1507.01006.
  process_type: DYP
  local_variables:
    tau: M2 / sqrts**2
    ymax: -0.5 * log(tau)
    maxTau: 0.080
    maxY: 0.663
  rule: "tau <= maxTau and fabs(y/ymax) <= maxY"

- dataset: DYE605_dw_ite
  reason: |
    Remove data points for which the fixed-order perturbative expansion is not
    reliable since resummation effects are large. A justification of these cuts
    can be found in arXiv:1507.01006.
  process_type: DYP
  local_variables:
    tau: M2 / sqrts**2
    ymax: -0.5 * log(tau)
    maxTau: 0.080
    maxY: 0.663
  rule: "tau <= maxTau and fabs(y/ymax) <= maxY"

- dataset: DYE605_sh_ite
  reason: |
    Remove data points for which the fixed-order perturbative expansion is not
    reliable since resummation effects are large. A justification of these cuts
    can be found in arXiv:1507.01006.
  process_type: DYP
  local_variables:
    tau: M2 / sqrts**2
    ymax: -0.5 * log(tau)
    maxTau: 0.080
    maxY: 0.663
  rule: "tau <= maxTau and fabs(y/ymax) <= maxY"

- process_type: DIS_ALL
  reason: Generalised cut for all DIS processes
  local_variables:
    w2: Q2 * (1 - x) / x
  rule: "Q2 > q2min and w2 > w2min"

- dataset: EMCF2P
  reason: |
    The inclusive EMC structure function measurements where known to be
    affected with a problem in the experimental measurement which rendered them
    unreliable for x < 0.1.
  rule: "x > 0.1"

- dataset: EMCF2D
  reason: |
    The inclusive EMC structure function measurements where known to be
    affected with a problem in the experimental measurement which rendered them
    unreliable for x < 0.1.
  rule: "x > 0.1"

- process_type: DIS_NCP_CH
  reason: |
    Heavy quark corrections are large in the FONLL-A calculation, and
    points at small-x and small Q2 need to be removed.
  FNS: FONLL-A
  local_variables:
    Q2cut1_f2c: 4
    Q2cut2_f2c: 10
    xcut_f2c: 1e-3
  rule: >-
    Q2 > Q2cut1_f2c and
    (Q2 > Q2cut2_f2c or x > xcut_f2c)

- process_type: DIS_NCP_CH
  reason: |
    Missing higher order corrections to Delta F_IC, the piece that needs
    to be added to the FONLL-C calculation in the case of fitted charm.
  FNS: FONLL-C
  IC: True
  rule: "Q2 > 8"


- dataset: CMSDY2D11
  reason: |
    Remove points at large values of the rapidity. The theoretical description
    of these points is poor, even at NNLO QCD + NLO EW. Removing these points
    results in a significant improvement in this dataset's chi2. Note that this
    cut is part of the standard NNPDF cuts for this dataset.
  rule: etay <= 2.2

- dataset: CMSDY2D11
  reason: |
    Specific to PDF4EFT fits. Cut points that go beyond LO kinematic limit in
    rapidity for highest invariant mass bin.
  rule: etay < 1.5 if M2 == 7.225e+05 else True

- dataset: ATLASDY2D8TEV
  reason: |
    Specific to PDF4EFT fits. Cut points that go beyond LO kinematic limit in
    rapidity for highest invariant mass bin.
  rule: etay < 1.68 if M2 == 1.0e+06 else True


- dataset: ATLAS_TTBAR_13TEV_HADRONIC_2D_TTM_ABSYTTNORM
  reason: |
   We remove the last bin of the normalised distribution because it is
   linearly dependent on the others
  rule: "(yQQ<1.4) or (mu2<3610000)"


- dataset: ATLAS_TTBAR_13TEV_TTMNORM
  reason: |
   We remove the last bin of the normalised distribution because it is
   linearly dependent on the others
  rule: "MQQ<1600"