Dear All,
We have collected the comments we received from everybody and have listed them below. Please check that your comments are included and that you are happy with the chosen wording.
Best regards,
Roel & Tjeerd
General comments:
- It seems that the time-dependent equations in Eq. (4) are not relevant. - We think that it would be much more informative to present the integrals of Eq. (5) including non-zero phi_s and the production asymmetry, but excluding detector dependent effects. Those should be easy to derive. Then it seems important to indicate early that the time integrated equations are only valid if you ignore the experimental acceptances and that ignoring those actually leads to non-negligible systematic uncertainties.
Did you in the analysis correct these equations for the decay time acceptance? We think that it would be very informative to show how equation 5 changes if you introduce a lower cut off in the lifetime. In your analysis, it seems that this lower cut off is about 1ps, very large compared to the lifetimes. Could you explain how the described optimization procedure resulted in such a large cut?
- Performing the time integration before folding in lifetime biasing effects, like cuts and the lifetime acceptance, is not correct. The lifetime component corresponding to A_0 and A_|| is tau_l, while for A_perp this is tau_h. Since a time acceptance has a different effect on these two components, it should be folded in before the integration over time is performed. This holds for both the lifetime bias introduced by the tight FD chi^2 cut and the detector time acceptance.
Comments in detail:
Abstract: Why do we quote three amplitudes squared if the sum is equal to unity? Like in the quoted B factory papers we can limit ourself to the first two.
Lines 2-3: - FCNC is not used further and the abbreviation is superfluous. - A suggestion to replace "Studies ... tests" by "Triple product asymmetry studies in this decay mode provide a powerful test".
Line 12: - Remove typing error 'of' in "V-A of structure". Consider "vector minus axial vector (V$-$A) structure", as we mention also vectors and pseudo scalars.
Line 14: - Replace "roughly equal longitudinal and transverse components" by Longitudinal fractions of $f_L = |A_0|^2 \approx 0.5$ ".
Line 17-19: - Replace 'predictions' by 'calculations' and "give ... 70%" by "allow for $f_L$ in the range of 0.4-0.7 [2,3]".
Fig. 1, caption: - Replace "helicity frame" by "center-of-mass frame" and 'mode' to "decay mode".
Lines 22-31: - I suppose that a reference here in stead of Eq.~(4) will be sufficient; we only consider the time integrated results.
Lines 32-34: - Replace "If as in this Letter ... integrate to" by "In an ideal situation where we can integrate over the full time range with CP violating phase $\phi_s=0$ and an equal number of B0s and Bbar0s mesons are produced, the time integrals of the functions $K_i(t)$ are:" (Avoid "in this Letter". This is not the analysis that is integrated.) - It would be much more informative to present the integrals including non-zero phi_s and the production asymmetry, but excluding detector dependent effects. Those should be easy to derive.
Lines 40-47: - The '$\pm$' sign in the definition of $V$ is not correct, because $\Phi$ is already negative or positive by definition in Eq.(6). - A rearrangement of the text arround Eq. (6) and the equation itself is suggested in the following changes: - Replace "due either to ... is violated" by "due either to final state interactions or to T violation, which with conservation of CPT also implies CP violation." - Move sentence "This approach ... is required" to line 47. - Replace "There are two ... triple products" by "There are two observables, triple products of unit vectors," and then start the equation directly with on the left sides in addition: "V =" and "U =". - Replace "These are proportional ..." by "From Fig. 1 it can be deduced that $V$ is positive if $cos\theta_1 \cos\theta_2 \gt 0$ and zero or negative otherwise. The T-odd variables $V$ and $U$ are proportional ..."
It could help to draw $n_{V1}$, $n_{V2}$ and $p_{V1}$ vectors in Fig. 1 on the outer sides of the two decay planes.
Lines 55-56: - If sample size the number should be larger than 801. If signal yiels they should have estimated uncertainties. - Avoid "In this Letter".
Line 70: Remove "consisting of ... GEM)". (It is not clear if 'consisting' applies also to 'Hadron' and if 'distinguishes' applies to only 'Muon'.)
Line 72: Add '(L0)' after 'triggering'. (L0 is used later without definition.)
Line 76: Add "of various selection lines". (Lines are used later without description.)
Line 83: We don't quite understand why you need to mention "s-plot" here. If you do, you need to mention what was used as separating variable. (In the end, it is just a fancy sideband subtraction. You wouldn't quote that you use sideband subtraction without explicitly mentioning which sideband.)
Note that the optimization using splot explicitly relies on the fact that the signal and background shapes are independent of your cuts. For the background B mass shape that is almost guaranteed not to be the case. (For example, the background B mass slope will depend on pT cuts, and even on vertex chi2 cuts.) We wonder how you took this into account in the optimization.
Line 87: Remove the sentence about the KL distance like in the table. This is just an artefact of our track finding.
Table 1: - Replace "Cut" by "Selection variable". - Are the IP chi2 cut and FD chi2 cut per dof? What is the number of dofs for the FD cut? (if it is 3, that this quantity is different from what most people outside lhcb think that it is.)
Lines 110-112: - Same remark as above about using 's-plot technique': you always should mention the separating variable. - Typing error "Fig. 3". - What function did you use for the two-body phase space component? How did you take the experimental resolution on the KK invariant mass into account?
Line 128: - You say that you use the measurements of Gs and DGs from a different source, including their errors. However, doesn't one expect that the dominating effect is the decay time acceptance?
Fig. 3, caption: - Remove spurious "is applied". - Add the separating variable used for making the s-plot.
Line 142: - We are actually a bit surprised that the effect of ignoring the decay time acceptance is so small. Looking at your analysis note, the lifetime acceptance effectively introduces a cut off of about 1 ps, large compared to tau_l and tau_H. Why don't you see this directly in the amplitudes? Is it because dG/G is small and the sum of the amplitudes is normalized to 1? The analysis note is very brief here: It shows the final numbers but not actually the toy distributions.
Fig. 4: - Are the fit projections corrected for the angular acceptance? This seems impossible to do with just the acceptance weights. If not, then that should be mentioned in the caption.
Table 3: - Replace the '-' with '0.000'. (A dash makes it seem like you didn't compute it.) - We don't understand why you quote separate results and systematic uncertainties for A_par2. You require the sum of the 3 amplitudes to be unity. So, anything that affects two amplitudes, will also affect the 3rd. It seems better to skip A_par2 completely.
For example, given equation 5, it seems hard to understand why A_par2 and A_02 have different sensitivity to the lifetime acceptance: they have roughly the same size, they appear with the same lifetime in your observables, so why do they not have the same sensitivity to the acceptance?
Line 152: - Remove "This has been checked to be valid in simulation studies". Or explain why you would worry about this.
Line 168: Remove sentence "Measurements ... A_0|^2$". The results do not differ so much from ~0.5 with the large spread the other values have.
Lines 169-170: - Replace "in agreement with that" by "almost equal to that". - Replace "also in reasonable agreement" by "in agreement". 'Also' is not correct and 'reasonable' is not subjective.
Line 175: - Replace 'hypothesis of no CP violation" by "hypothesis of CP conservation".
Dear All,
We have collected the comments we received from everybody and have listed them below. Please check that your comments are included and that you are happy with the chosen wording.
With some small modifications, I have submitted our comments to CDS. They can be viewed here: http://cdsweb.cern.ch/record/1428852.
Cheers, Roel
General comments:
- It seems that the time-dependent equations in Eq. (4) are not relevant.
- We think that it would be much more informative to present the integrals
of Eq. (5) including non-zero phi_s and the production asymmetry, but excluding detector dependent effects. Those should be easy to derive. Then it seems important to indicate early that the time integrated equations are only valid if you ignore the experimental acceptances and that ignoring those actually leads to non-negligible systematic uncertainties.
Did you in the analysis correct these equations for the decay time acceptance? We think that it would be very informative to show how equation 5 changes if you introduce a lower cut off in the lifetime. In your analysis, it seems that this lower cut off is about 1ps, very large compared to the lifetimes. Could you explain how the described optimization procedure resulted in such a large cut?
- Performing the time integration before folding in lifetime biasing effects,
like cuts and the lifetime acceptance, is not correct. The lifetime component corresponding to A_0 and A_|| is tau_l, while for A_perp this is tau_h. Since a time acceptance has a different effect on these two components, it should be folded in before the integration over time is performed. This holds for both the lifetime bias introduced by the tight FD chi^2 cut and the detector time acceptance.
Comments in detail:
Abstract: Why do we quote three amplitudes squared if the sum is equal to unity? Like in the quoted B factory papers we can limit ourself to the first two.
Lines 2-3:
- FCNC is not used further and the abbreviation is superfluous.
- A suggestion to replace "Studies ... tests" by "Triple product asymmetry studies in this decay mode provide a powerful test".
Line 12:
- Remove typing error 'of' in "V-A of structure". Consider "vector minus axial vector (V$-$A) structure", as we mention also vectors and pseudo scalars.
Line 14:
- Replace "roughly equal longitudinal and transverse components" by Longitudinal fractions of $f_L = |A_0|^2 \approx 0.5$ ".
Line 17-19:
- Replace 'predictions' by 'calculations' and "give ... 70%" by "allow for $f_L$ in the range of 0.4-0.7 [2,3]".
Fig. 1, caption:
- Replace "helicity frame" by "center-of-mass frame" and 'mode' to "decay mode".
Lines 22-31:
- I suppose that a reference here in stead of Eq.~(4) will be sufficient; we only consider the time integrated results.
Lines 32-34:
- Replace "If as in this Letter ... integrate to" by "In an ideal situation where we can integrate over the full time range with CP violating phase $\phi_s=0$ and an equal number of B0s and Bbar0s mesons are produced, the time integrals of the functions $K_i(t)$ are:" (Avoid "in this Letter". This is not the analysis that is integrated.)
- It would be much more informative to present the integrals including non-zero phi_s and the production asymmetry, but excluding detector dependent effects. Those should be easy to derive.
Lines 40-47:
The '$\pm$' sign in the definition of $V$ is not correct, because $\Phi$ is already negative or positive by definition in Eq.(6).
A rearrangement of the text arround Eq. (6) and the equation itself is suggested in the following changes:
Replace "due either to ... is violated" by "due either to final state interactions or to T violation, which with conservation of CPT also implies CP violation."
Move sentence "This approach ... is required" to line 47.
Replace "There are two ... triple products" by "There are two observables, triple products of unit vectors," and then start the equation directly with on the left sides in addition: "V =" and "U =".
Replace "These are proportional ..." by "From Fig. 1 it can be deduced that $V$ is positive if $cos\theta_1 \cos\theta_2 \gt 0$ and zero or negative otherwise. The T-odd variables $V$ and $U$ are proportional ..."
It could help to draw $n_{V1}$, $n_{V2}$ and $p_{V1}$ vectors in Fig. 1 on the outer sides of the two decay planes.
Lines 55-56:
- If sample size the number should be larger than 801. If signal yiels they should have estimated uncertainties.
- Avoid "In this Letter".
Line 70: Remove "consisting of ... GEM)". (It is not clear if 'consisting' applies also to 'Hadron' and if 'distinguishes' applies to only 'Muon'.)
Line 72: Add '(L0)' after 'triggering'. (L0 is used later without definition.)
Line 76: Add "of various selection lines". (Lines are used later without description.)
Line 83: We don't quite understand why you need to mention "s-plot" here. If you do, you need to mention what was used as separating variable. (In the end, it is just a fancy sideband subtraction. You wouldn't quote that you use sideband subtraction without explicitly mentioning which sideband.)
Note that the optimization using splot explicitly relies on the fact that the signal and background shapes are independent of your cuts. For the background B mass shape that is almost guaranteed not to be the case. (For example, the background B mass slope will depend on pT cuts, and even on vertex chi2 cuts.) We wonder how you took this into account in the optimization.
Line 87: Remove the sentence about the KL distance like in the table. This is just an artefact of our track finding.
Table 1:
- Replace "Cut" by "Selection variable".
- Are the IP chi2 cut and FD chi2 cut per dof? What is the number of dofs for the FD cut? (if it is 3, that this quantity is different from what most people outside lhcb think that it is.)
Lines 110-112:
- Same remark as above about using 's-plot technique': you always should mention the separating variable.
- Typing error "Fig. 3".
- What function did you use for the two-body phase space component? How did you take the experimental resolution on the KK invariant mass into account?
Line 128:
- You say that you use the measurements of Gs and DGs from a different source, including their errors. However, doesn't one expect that the dominating effect is the decay time acceptance?
Fig. 3, caption:
- Remove spurious "is applied".
- Add the separating variable used for making the s-plot.
Line 142:
- We are actually a bit surprised that the effect of ignoring the decay time acceptance is so small. Looking at your analysis note, the lifetime acceptance effectively introduces a cut off of about 1 ps, large compared to tau_l and tau_H. Why don't you see this directly in the amplitudes? Is it because dG/G is small and the sum of the amplitudes is normalized to 1? The analysis note is very brief here: It shows the final numbers but not actually the toy distributions.
Fig. 4:
- Are the fit projections corrected for the angular acceptance? This seems impossible to do with just the acceptance weights. If not, then that should be mentioned in the caption.
Table 3:
Replace the '-' with '0.000'. (A dash makes it seem like you didn't compute it.)
We don't understand why you quote separate results and systematic uncertainties for A_par2. You require the sum of the 3 amplitudes to be unity. So, anything that affects two amplitudes, will also affect the 3rd. It seems better to skip A_par2 completely.
For example, given equation 5, it seems hard to understand why A_par2 and A_02 have different sensitivity to the lifetime acceptance: they have roughly the same size, they appear with the same lifetime in your observables, so why do they not have the same sensitivity to the acceptance?
Line 152:
- Remove "This has been checked to be valid in simulation studies". Or explain why you would worry about this.
Line 168: Remove sentence "Measurements ... A_0|^2$". The results do not differ so much from ~0.5 with the large spread the other values have.
Lines 169-170:
- Replace "in agreement with that" by "almost equal to that".
- Replace "also in reasonable agreement" by "in agreement". 'Also' is not correct and 'reasonable' is not subjective.
Line 175:
- Replace 'hypothesis of no CP violation" by "hypothesis of CP conservation".
Bfys-physics mailing list Bfys-physics@nikhef.nl https://mailman.nikhef.nl/mailman/listinfo/bfys-physics
Dear All,
We have collected the comments we received from everybody and have listed them below. Please check that your comments are included and that you are happy with the chosen wording.
With some small modifications, I have submitted our comments to CDS. They can be viewed here: http://cdsweb.cern.ch/record/1428852.
The authors have responded to our comments and their answers can be viewed at the link above. If a satisfactory answer has not been provided to any or our questions, please let me know so an appropriate response can be formulated.
Cheers, Roel
General comments:
- It seems that the time-dependent equations in Eq. (4) are not relevant.
- We think that it would be much more informative to present the integrals
of Eq. (5) including non-zero phi_s and the production asymmetry, but excluding detector dependent effects. Those should be easy to derive. Then it seems important to indicate early that the time integrated equations are only valid if you ignore the experimental acceptances and that ignoring those actually leads to non-negligible systematic uncertainties.
Did you in the analysis correct these equations for the decay time acceptance? We think that it would be very informative to show how equation 5 changes if you introduce a lower cut off in the lifetime. In your analysis, it seems that this lower cut off is about 1ps, very large compared to the lifetimes. Could you explain how the described optimization procedure resulted in such a large cut?
- Performing the time integration before folding in lifetime biasing effects,
like cuts and the lifetime acceptance, is not correct. The lifetime component corresponding to A_0 and A_|| is tau_l, while for A_perp this is tau_h. Since a time acceptance has a different effect on these two components, it should be folded in before the integration over time is performed. This holds for both the lifetime bias introduced by the tight FD chi^2 cut and the detector time acceptance.
Comments in detail:
Abstract: Why do we quote three amplitudes squared if the sum is equal to unity? Like in the quoted B factory papers we can limit ourself to the first two.
Lines 2-3:
- FCNC is not used further and the abbreviation is superfluous.
- A suggestion to replace "Studies ... tests" by "Triple product asymmetry studies in this decay mode provide a powerful test".
Line 12:
- Remove typing error 'of' in "V-A of structure". Consider "vector minus axial vector (V$-$A) structure", as we mention also vectors and pseudo scalars.
Line 14:
- Replace "roughly equal longitudinal and transverse components" by Longitudinal fractions of $f_L = |A_0|^2 \approx 0.5$ ".
Line 17-19:
- Replace 'predictions' by 'calculations' and "give ... 70%" by "allow for $f_L$ in the range of 0.4-0.7 [2,3]".
Fig. 1, caption:
- Replace "helicity frame" by "center-of-mass frame" and 'mode' to "decay mode".
Lines 22-31:
- I suppose that a reference here in stead of Eq.~(4) will be sufficient; we only consider the time integrated results.
Lines 32-34:
- Replace "If as in this Letter ... integrate to" by "In an ideal situation where we can integrate over the full time range with CP violating phase $\phi_s=0$ and an equal number of B0s and Bbar0s mesons are produced, the time integrals of the functions $K_i(t)$ are:" (Avoid "in this Letter". This is not the analysis that is integrated.)
- It would be much more informative to present the integrals including non-zero phi_s and the production asymmetry, but excluding detector dependent effects. Those should be easy to derive.
Lines 40-47:
The '$\pm$' sign in the definition of $V$ is not correct, because $\Phi$ is already negative or positive by definition in Eq.(6).
A rearrangement of the text arround Eq. (6) and the equation itself is suggested in the following changes:
Replace "due either to ... is violated" by "due either to final state interactions or to T violation, which with conservation of CPT also implies CP violation."
Move sentence "This approach ... is required" to line 47.
Replace "There are two ... triple products" by "There are two observables, triple products of unit vectors," and then start the equation directly with on the left sides in addition: "V =" and "U =".
Replace "These are proportional ..." by "From Fig. 1 it can be deduced that $V$ is positive if $cos\theta_1 \cos\theta_2 \gt 0$ and zero or negative otherwise. The T-odd variables $V$ and $U$ are proportional ..."
It could help to draw $n_{V1}$, $n_{V2}$ and $p_{V1}$ vectors in Fig. 1 on the outer sides of the two decay planes.
Lines 55-56:
- If sample size the number should be larger than 801. If signal yiels they should have estimated uncertainties.
- Avoid "In this Letter".
Line 70: Remove "consisting of ... GEM)". (It is not clear if 'consisting' applies also to 'Hadron' and if 'distinguishes' applies to only 'Muon'.)
Line 72: Add '(L0)' after 'triggering'. (L0 is used later without definition.)
Line 76: Add "of various selection lines". (Lines are used later without description.)
Line 83: We don't quite understand why you need to mention "s-plot" here. If you do, you need to mention what was used as separating variable. (In the end, it is just a fancy sideband subtraction. You wouldn't quote that you use sideband subtraction without explicitly mentioning which sideband.)
Note that the optimization using splot explicitly relies on the fact that the signal and background shapes are independent of your cuts. For the background B mass shape that is almost guaranteed not to be the case. (For example, the background B mass slope will depend on pT cuts, and even on vertex chi2 cuts.) We wonder how you took this into account in the optimization.
Line 87: Remove the sentence about the KL distance like in the table. This is just an artefact of our track finding.
Table 1:
- Replace "Cut" by "Selection variable".
- Are the IP chi2 cut and FD chi2 cut per dof? What is the number of dofs for the FD cut? (if it is 3, that this quantity is different from what most people outside lhcb think that it is.)
Lines 110-112:
- Same remark as above about using 's-plot technique': you always should mention the separating variable.
- Typing error "Fig. 3".
- What function did you use for the two-body phase space component? How did you take the experimental resolution on the KK invariant mass into account?
Line 128:
- You say that you use the measurements of Gs and DGs from a different source, including their errors. However, doesn't one expect that the dominating effect is the decay time acceptance?
Fig. 3, caption:
- Remove spurious "is applied".
- Add the separating variable used for making the s-plot.
Line 142:
- We are actually a bit surprised that the effect of ignoring the decay time acceptance is so small. Looking at your analysis note, the lifetime acceptance effectively introduces a cut off of about 1 ps, large compared to tau_l and tau_H. Why don't you see this directly in the amplitudes? Is it because dG/G is small and the sum of the amplitudes is normalized to 1? The analysis note is very brief here: It shows the final numbers but not actually the toy distributions.
Fig. 4:
- Are the fit projections corrected for the angular acceptance? This seems impossible to do with just the acceptance weights. If not, then that should be mentioned in the caption.
Table 3:
Replace the '-' with '0.000'. (A dash makes it seem like you didn't compute it.)
We don't understand why you quote separate results and systematic uncertainties for A_par2. You require the sum of the 3 amplitudes to be unity. So, anything that affects two amplitudes, will also affect the 3rd. It seems better to skip A_par2 completely.
For example, given equation 5, it seems hard to understand why A_par2 and A_02 have different sensitivity to the lifetime acceptance: they have roughly the same size, they appear with the same lifetime in your observables, so why do they not have the same sensitivity to the acceptance?
Line 152:
- Remove "This has been checked to be valid in simulation studies". Or explain why you would worry about this.
Line 168: Remove sentence "Measurements ... A_0|^2$". The results do not differ so much from ~0.5 with the large spread the other values have.
Lines 169-170:
- Replace "in agreement with that" by "almost equal to that".
- Replace "also in reasonable agreement" by "in agreement". 'Also' is not correct and 'reasonable' is not subjective.
Line 175:
- Replace 'hypothesis of no CP violation" by "hypothesis of CP conservation".
Bfys-physics mailing list Bfys-physics@nikhef.nl https://mailman.nikhef.nl/mailman/listinfo/bfys-physics
Bfys-physics mailing list Bfys-physics@nikhef.nl https://mailman.nikhef.nl/mailman/listinfo/bfys-physics
Hi Roel,
Thanks a lot for collecting all of this.
I still don't understand why they don't explicitly put the effect of the production asymmetry and non-zero phis in the integrals in equation 5. There argument that the reader can easily derive this doesn't make much sense to me: equation 5 is even easier to derive. I don't know how you could politely formulate that.
If you like I could also post the questions below directly on the indico.
Table 1:
- Are the IP chi2 cut and FD chi2 cut per dof?
What is the number of dofs for the FD cut? (If it is 3, then this quantity is different from what most people outside LHCb think it is.)
Both of them are not per dof. The FD cut we believe has 1 dof as this is a significance and not the result of a fit.
Unfortunately, it really depends on which functor you call. I suspect that the FD chi2 cut is actually the 3D chi2 difference between PV and B vertex, e.g. not projected on the direction. If so, it has 3dofs, and it should probably not be called flight distance chi2.
Our studies have shown that the larger effect is that of the central value of the lifetimes rather than acceptance. I
From equations 5 this is not easy hard to understand so I propose that you elaborate. Your lifetime acceptance introduces a cutoff that is essentially 2/3 of the B lifetime. Equation 5 shows that you measure something that is the product of the lifetime and the amplitude. The only way one can understand that a lifetime acceptance doesn't play a role, is if a constant cutoff (say integrating form t>1ps) would not affect the _relative_ integrals in 5. That's why we proposed to derive (at least in the analysis note) how a constant cutoff affects the integrals. Probably, for zero phi_s ad zero dGs, all amplitudes just scale (normalization). In that case, your constraint that the amplitudes add up to 1, just removes an overall scale factor.
- We don't understand why you quote separate results and systematic
uncertainties for A_par2. You require the sum of the three squared amplitudes to be unity. So, anything that affects two amplitudes, will also affect the third. It seems better to skip A_par2 completely.
If we did not quote the Apara_sq errors independently (as CDF do for their fixed parameter), then we would not have been able to show the plot in conferences of the f_L vs. f_parallel comparison (or only with great difficulty). We have the number, so we do not see a reason not to use it.
From the current abstract and conclusion it seems that you are measuring 3 numbers while you measure really only two independent numbers. You should emphasize this at least in the conclusion. In the caption of table 2 you mention the correlation coefficient for 1 pair. But the correlations for the other pairs are big as well.
For example, given equation 5, it seems hard to understand why A_par2 and A_02 have different sensitivity to the lifetime acceptance: they have roughly the same size, they appear with the same lifetime in your observables, so why do they not have the same sensitivity to the acceptance?
A_par2 will have sensitivity to both lifetime acceptances due to its correlation with Aperp2.
I am not sure what you mean with 'both lifetime acceptances', but in any case: given that A02 and Aperp2 are about 50% correlated, the correlations between Apar2 and Aperp2, and between Apar2 and A02 cannot be very different: they are all large, because somehow they have to "add up" to 100%. So, I still don't understand it: how can 1 parameter have a 3x smaller sensitivity to lifetime acceptance than the other 2?
For the S-wave I cna sort of understand what is going on: The S wave affects CP-odd versus even. So, that's why the changes in 'odd' exactly add up to the changes in 'even'. However, for the lifetime I don't see that happening: the lifetime doesn't really separate one amplitude from another.
Cheers, Wouter
On 03/08/2012 04:27 PM, Roel Aaij wrote:
Dear All,
We have collected the comments we received from everybody and have listed them below. Please check that your comments are included and that you are happy with the chosen wording.
With some small modifications, I have submitted our comments to CDS. They can be viewed here: http://cdsweb.cern.ch/record/1428852.
The authors have responded to our comments and their answers can be viewed at the link above. If a satisfactory answer has not been provided to any or our questions, please let me know so an appropriate response can be formulated.
Cheers, Roel
General comments:
- It seems that the time-dependent equations in Eq. (4) are not relevant.
- We think that it would be much more informative to present the integrals
of Eq. (5) including non-zero phi_s and the production asymmetry, but excluding detector dependent effects. Those should be easy to derive. Then it seems important to indicate early that the time integrated equations are only valid if you ignore the experimental acceptances and that ignoring those actually leads to non-negligible systematic uncertainties.
Did you in the analysis correct these equations for the decay time acceptance? We think that it would be very informative to show how equation 5 changes if you introduce a lower cut off in the lifetime. In your analysis, it seems that this lower cut off is about 1ps, very large compared to the lifetimes. Could you explain how the described optimization procedure resulted in such a large cut?
- Performing the time integration before folding in lifetime biasing effects,
like cuts and the lifetime acceptance, is not correct. The lifetime component corresponding to A_0 and A_|| is tau_l, while for A_perp this is tau_h. Since a time acceptance has a different effect on these two components, it should be folded in before the integration over time is performed. This holds for both the lifetime bias introduced by the tight FD chi^2 cut and the detector time acceptance.
Comments in detail:
Abstract: Why do we quote three amplitudes squared if the sum is equal to unity? Like in the quoted B factory papers we can limit ourself to the first two.
Lines 2-3:
- FCNC is not used further and the abbreviation is superfluous.
- A suggestion to replace "Studies ... tests" by "Triple product asymmetry studies in this decay mode provide a powerful test".
Line 12:
- Remove typing error 'of' in "V-A of structure". Consider "vector minus axial vector (V$-$A) structure", as we mention also vectors and pseudo scalars.
Line 14:
- Replace "roughly equal longitudinal and transverse components" by Longitudinal fractions of $f_L = |A_0|^2 \approx 0.5$ ".
Line 17-19:
- Replace 'predictions' by 'calculations' and "give ... 70%" by "allow for $f_L$ in the range of 0.4-0.7 [2,3]".
Fig. 1, caption:
- Replace "helicity frame" by "center-of-mass frame" and 'mode' to "decay mode".
Lines 22-31:
- I suppose that a reference here in stead of Eq.~(4) will be sufficient; we only consider the time integrated results.
Lines 32-34:
- Replace "If as in this Letter ... integrate to" by "In an ideal situation where we can integrate over the full time range with CP violating phase $\phi_s=0$ and an equal number of B0s and Bbar0s mesons are produced, the time integrals of the functions $K_i(t)$ are:" (Avoid "in this Letter". This is not the analysis that is integrated.)
- It would be much more informative to present the integrals including non-zero phi_s and the production asymmetry, but excluding detector dependent effects. Those should be easy to derive.
Lines 40-47:
The '$\pm$' sign in the definition of $V$ is not correct, because $\Phi$ is already negative or positive by definition in Eq.(6).
A rearrangement of the text arround Eq. (6) and the equation itself is suggested in the following changes:
Replace "due either to ... is violated" by "due either to final state interactions or to T violation, which with conservation of CPT also implies CP violation."
Move sentence "This approach ... is required" to line 47.
Replace "There are two ... triple products" by "There are two observables, triple products of unit vectors," and then start the equation directly with on the left sides in addition: "V =" and "U =".
Replace "These are proportional ..." by "From Fig. 1 it can be deduced that $V$ is positive if $cos\theta_1 \cos\theta_2 \gt 0$ and zero or negative otherwise. The T-odd variables $V$ and $U$ are proportional ..."
It could help to draw $n_{V1}$, $n_{V2}$ and $p_{V1}$ vectors in Fig. 1 on the outer sides of the two decay planes.
Lines 55-56:
- If sample size the number should be larger than 801. If signal yiels they should have estimated uncertainties.
- Avoid "In this Letter".
Line 70: Remove "consisting of ... GEM)". (It is not clear if 'consisting' applies also to 'Hadron' and if 'distinguishes' applies to only 'Muon'.)
Line 72: Add '(L0)' after 'triggering'. (L0 is used later without definition.)
Line 76: Add "of various selection lines". (Lines are used later without description.)
Line 83: We don't quite understand why you need to mention "s-plot" here. If you do, you need to mention what was used as separating variable. (In the end, it is just a fancy sideband subtraction. You wouldn't quote that you use sideband subtraction without explicitly mentioning which sideband.)
Note that the optimization using splot explicitly relies on the fact that the signal and background shapes are independent of your cuts. For the background B mass shape that is almost guaranteed not to be the case. (For example, the background B mass slope will depend on pT cuts, and even on vertex chi2 cuts.) We wonder how you took this into account in the optimization.
Line 87: Remove the sentence about the KL distance like in the table. This is just an artefact of our track finding.
Table 1:
- Replace "Cut" by "Selection variable".
- Are the IP chi2 cut and FD chi2 cut per dof? What is the number of dofs for the FD cut? (if it is 3, that this quantity is different from what most people outside lhcb think that it is.)
Lines 110-112:
- Same remark as above about using 's-plot technique': you always should mention the separating variable.
- Typing error "Fig. 3".
- What function did you use for the two-body phase space component? How did you take the experimental resolution on the KK invariant mass into account?
Line 128:
- You say that you use the measurements of Gs and DGs from a different source, including their errors. However, doesn't one expect that the dominating effect is the decay time acceptance?
Fig. 3, caption:
- Remove spurious "is applied".
- Add the separating variable used for making the s-plot.
Line 142:
- We are actually a bit surprised that the effect of ignoring the decay time acceptance is so small. Looking at your analysis note, the lifetime acceptance effectively introduces a cut off of about 1 ps, large compared to tau_l and tau_H. Why don't you see this directly in the amplitudes? Is it because dG/G is small and the sum of the amplitudes is normalized to 1? The analysis note is very brief here: It shows the final numbers but not actually the toy distributions.
Fig. 4:
- Are the fit projections corrected for the angular acceptance? This seems impossible to do with just the acceptance weights. If not, then that should be mentioned in the caption.
Table 3:
Replace the '-' with '0.000'. (A dash makes it seem like you didn't compute it.)
We don't understand why you quote separate results and systematic uncertainties for A_par2. You require the sum of the 3 amplitudes to be unity. So, anything that affects two amplitudes, will also affect the 3rd. It seems better to skip A_par2 completely.
For example, given equation 5, it seems hard to understand why A_par2 and A_02 have different sensitivity to the lifetime acceptance: they have roughly the same size, they appear with the same lifetime in your observables, so why do they not have the same sensitivity to the acceptance?
Line 152:
- Remove "This has been checked to be valid in simulation studies". Or explain why you would worry about this.
Line 168: Remove sentence "Measurements ... A_0|^2$". The results do not differ so much from ~0.5 with the large spread the other values have.
Lines 169-170:
- Replace "in agreement with that" by "almost equal to that".
- Replace "also in reasonable agreement" by "in agreement". 'Also' is not correct and 'reasonable' is not subjective.
Line 175:
- Replace 'hypothesis of no CP violation" by "hypothesis of CP conservation".
Bfys-physics mailing list Bfys-physics@nikhef.nl https://mailman.nikhef.nl/mailman/listinfo/bfys-physics
Bfys-physics mailing list Bfys-physics@nikhef.nl https://mailman.nikhef.nl/mailman/listinfo/bfys-physics
Bfys-physics mailing list Bfys-physics@nikhef.nl https://mailman.nikhef.nl/mailman/listinfo/bfys-physics
-
Roel Aaij -
Wouter Hulsbergen