[b1_ana] Axx

Fri Apr 26 19:33:39 EDT 2013

Hi Narbe,

As I said in my draft note, we aren't limited to taking the difference
between the upstream, unpolarized cup data and the downstream, polarized
one, which would just be the configuration for the first polarization
period.

In the second period we would polarize the upstream cup, and turn off
the microwaves for the downstream cup. Then, we can take four
differences, please see equations (15) and (16) of my draft:

1 left eq.  (15) - left eq.  (16) = upstream   unpol - downstream pol
2 right eq. (15) - right eq. (16) = upstream   pol   - downstream unpol
3 left eq.  (15) - right eq. (16) = upstream   unpol - upstream   pol
4 right eq. (15) - left eq.  (16) = downstream unpol - downstream pol

differences 1 and 2 are the method discussed up to eq. (14);
differences 3 and 4 are the same-cup ones, like that shown on eq. (19),
which is a difference type 3.

As you can see, the difference types 3 and 4 are exactly what Dustin is
talking about, except that he talks in terms of dividing through by the
unpolarized charge normalized counts N_U/Q_U.

We can alternate the pol and unpol periods many times, throughout the
experiment. As I say after eq. (19), we can combine the statistics of
all the sets of differences types 3 and 4, which are independent
measurements, to get the best statistics. And you can see that the
systematics are very favorable.

But since there would be a question of the time dependence of the
detector efficiency from one period to the next, the differences types 1
and 2, which have efficiency as a common factor, would let us control
this dependence.

Moreover, as Dustin also points out, there is no need for Pzz < 0. And
on top of that, there is no need to put the field along q, requiring a
chicane, etc. Both these items are required for measuring Azz directly.

I realize that developing Pzz < 0 for solid polarized targets is an
interesting technical advance. I just don't think we need it to get b1.

For everyone's convenience, here is my draft
https://twist.phys.virginia.edu/~or/b1/b1_method-v2.pdf

Cheers,

Oscar

narbe at jlab.org wrote:
> Hi Oscar,
> 
> I don't think anyone is really putting method ahead of physics quantity. The thing is that the issue of needing negative tensor polarization hasn't conclusively gone away (yet) and doesn't really seem to. It is more of the idea that this would spur the much needed r&d, as well as also generating a new program if physics with such a target.
> 
> My question is if there is anything precluding us from mentioning both methods, like for instance, having Dustin's suggested route in an appendix for an alternative. That way we could cover our bases. 
> 
> The issue I worry about with the 2 cup design is the possible contribution to the downstream cup from the upstream cup. I don't know if switching back and forth can alleviate this. I'm sure COMPASS had  some way to correct for this, and I'm willing to look into it if that's what we choose over the single cup.
> 
> Narbe
> 
> 
> Sent from my Verizon Wireless Phone
> 
> ----- Reply message -----
> From: "O. A. Rondon" <or at virginia.edu>
> To: "Dustin Keller" <dustin at jlab.org>
> Cc: "b1_ana at jlab.org" <b1_ana at jlab.org>
> Subject: [b1_ana] Axx
> Date: Fri, Apr 26, 2013 2:57 pm
> 
> 
> Hi,
> 
> As I pointed out in my message about Axx, the Np and Nu have to be
> coincidence e'p events for them to be associated to Axx or Azz.  And the
> coincidence events have to be correlated quasielastic e-p. In such case,
> Np, Nu are related to rates from the m=+/-1, 0 states, because only then
> Axx is a well defined analyzing power or asymmetry. But it does not work
> for DIS, and may not even work for high Q^2 q.e.
> 
> If Np, Nu are inclusive events, and we are dealing with ND3,or even LiD,
> then we are back, at best, to my method's eq. (19). Feel free to plug in
> the detailed expressions and see how things propagate. Keep in mind that
> a factor related to the time dependent detector efficiency needs to be
> included in all the equations. Such factor is common for the upstream -
> downstream difference (data taken simultaneously) but not for the
> same-cup one (different periods).
> 
> I think one thing that may be ambiguous is what is the emphasis of the
> proposal, namely to study a physics quantity by the most accessible
> means, or to use the measurement to advance existing techniques, which
> is a perfectly valid and worthy goal, but I must say I'm more interested
> in the quantity than in the technique.
> 
> Cheers,
> 
> Oscar
> 
> Dustin Keller wrote:
>> Hi,
>> Getting 20% uncertainty using hole burning for negative tensor 
>> polarization down to 5% seems a bit of a stretch to me especially given 
>> the small single people are interested in for b1.  But I'm glad to help in 
>> anyway I can to try to get it through.
>>
>> We really should consider the form of
>> Azz=(2/Pzz)(N^p-N^u)/N^u
>> it get the uncertainty from 20% to 10%, a much better starting point.
>>
>> dustin
>>
>> On Fri, 26 Apr 2013, Karl Slifer wrote:
>>
>>> Hi,
>>>
>>> While it would be easier to run with only positive Pzz, there is no
>>> technical or theoretical reason that I know of that prevents us from using
>>> negative pol.  This will require target development to achieve large
>>> negative as well as positive tensor polarizations, along with careful study
>>> of the systematics in extracting these values.  I thought we all agreed on
>>> this yesterday....And I also see no technical or theoretical reason (other
>>> than it is difficult and will require R&D) which limits the enhanced tensor
>>> polarization to 10%.  I believe Don, Chris, Josh would all agree with this,
>>> atleast they all did when I talked to them within the last 6 months.
>>>
>>> So to be clear, I believe we can propose an experiment where we enhance the
>>> m=0 population (via rf saturation or by using two independent microwave
>>> sources) and measure N_0 unpolarized electrons inclusively scattered while
>>> in this state, and then we deplete the m=0 state to obtain a positive
>>> polarization and measure N_1 unpolarized electrons scattered while in this
>>> state.  Then we form the asymmetry (with appropriate numerical factors).
>>> The Pzz will not be the maximal positive or negative value in either state,
>>> but we can correct for this by the relative Pzz in each state.    One
>>> significant concern is that this introduces time dependent systematics
>>> since it will likely require some time to switch between the two states.
>>> This has to be studied, but I do not see it as a fundamental limitation.
>>>
>>> After careful study of the systematics, its possible that the cross section
>>> difference method may well turn out to be the best way to do the
>>> experiment, but I suspect we will struggle mightily to convince a very
>>> skeptical PAC in 30 mins that we really can control all the systematic
>>> effects to the level needed for a cross section measurement.
>>>
>>> With this in mind, I think it is reasonable to aim for conditional approval
>>> based on demonstration of the target performance to the level needed (+-20%
>>> tensor pol with about 5% relative uncertainty).  I believe we can defend
>>> these as reasonable goals, although we should get something in the way of a
>>> support statement from Don or Chris.  And I believe conditional approval is
>>> a highly desirable state, since the target groups will not be able to
>>> commit serious R&D to this without an approved experiment for motivation.
>>> In addition, it opens the door to attract more theory support and start
>>> consideration of several other possible experiments.
>>>
>>> my few further cents,
>>>
>>> Karl
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>> ---
>>> Karl J. Slifer
>>> Assistant Professor
>>> University of New Hampshire
>>> Telephone : 603-722-0695
>>>
>>>
>>> On Fri, Apr 26, 2013 at 11:04 AM, Dustin Keller <dustin at jlab.org> wrote:
>>>
>>>> As I mentioned in the meeting using the notation Axx can be
>>>> mis-leading especially in the case of DIS where azimuthal
>>>> control is not obvious.  However the relationship for
>>>> sigma^{+/-} for m=+1,-1 is sigma^{+/-}=sigma^u(1+(1/2)AzzPzz).
>>>> If you believe that then Axx=Azz, and the conclusion is the same.
>>>> If you call it Axx or Azz in either case you just measure the
>>>> ratio of polarized and unpolarized cross sections.  This will lead
>>>> to a target tensor polarization of about 10%.  There are certainly
>>>> other systematic concerns but this is the best we can do target wise.
>>>>
>>>> dustin
>>>> _______________________________________________
>>>> b1_ana mailing list
>>>> b1_ana at jlab.org
>>>> https://mailman.jlab.org/mailman/listinfo/b1_ana
>>>>
>> _______________________________________________
>> b1_ana mailing list
>> b1_ana at jlab.org
>> https://mailman.jlab.org/mailman/listinfo/b1_ana
>>
> 
> 
> 
> 
> 
> _______________________________________________
> b1_ana mailing list
> b1_ana at jlab.org
> https://mailman.jlab.org/mailman/listinfo/b1_ana