In the Santiago face-to-face meeting, relatively high priority was
assigned to ACA based on rather vague feeling. This is potentially
of severe concern to ALMA, because the full scope of the enhancement
may be threatened if we fail to present strong science cases for
ALMA+SD+ACA that cannot be realized with ALMA+SD. Here I summarize
some of my comments/concerns and TBDs on the science cases for ACA
in the following. I hope this helps a bit to finalize the ASAC
report. Any comments and corrections are welcome.

(1) General comments/concerns:
- Mass and motion of the objects are two of the most important
parameters in observational astrophysics. How ACA is related to
these parameters? Accurate imaging does not mean robust extraction
of mass distribution. What is the importance of extremely high
fidelity images probed by molecules other than molecular hydrogen?
A velocity profile accurately measures motions only if all of the flux is represented in that profile.  For instance, it is well known that an e.g. HCO+ 1-0
profile would not be used for kinematics in general because the optical depth is so high that some outer portions of the cloud may totally absorb radiation
from central parts of the cloud.  In a similar way, interferometer maps which miss flux on some spatial scales misrepresent the kinematics of a region.
The C18O images of IRAS16293 from OVRO for instance (ApJ 352, 159) only show extreme velocities as flux near the central velocity is resolved out
by the interferometer.  Without accurate imaging, including flux on all spatial scales, kinematics cannot be reliably determined.

The most important tracer of mass we have is dust.  Dust emits most strongly at the shortest wavelengths ALMA can reach.  It is at those wavelengths--at
which pointing poses the most severe problems, and at which 'short spacings' means anything more extended than 4" or so--at which ALMA will provide
best estimates of mass, but only if the imaging is accurate enough to account for flux on spatial scales of 12m and shorter, where much of the flux lies.
- What fraction of ALMA users really takes advantages of the extremely
high fidelity of ALMA+SD+ACA? Is it of higher priority than to
ensure routine high fidelity imaging with ALMA+SD to all users?
I think that this is wavelength dependent.  I think no user will want ALMA+SD data alone in the submillimeter bands, if he pursues any sort of quantitative measure.

- How easily will we lose the gain achieved with ACA by degrading the
SD observations or by inappropriate deconvolution algorisms?
All deconvolution algorithms tried so far suggest that imaging is robust to the choice of algorithm.

(2) Line ratios:
- In multi-line analyses, ambiguity of chemical abundances, complexity
of the objects, unmatched synthesized beams, inhomogeneity of
physical conditions in the synthesized beam, and robustness of the
model used (e.g., LVG) often limit the robustness of conclusions.
We should evaluate what level of fidelity is necessary/sufficient
for this kind of analyses, and what level of fidelity can be
achieved for realistic cases (noise for realistic integration times
and source strengths).
This is the purpose of the imaging of the two transitions in the M51 and Hogerheidje simulations.  For the latter, Michiel suggests an error in line ratio will compromise conclusions about mass to be drawn; at the ASAC meeting we requested that the conclusions about mass be quantified.
- We should explain why extremely high fidelity is preferred rather
than number of bands (or number of observable lines) that also
enhances ability of multi-line observations.
I am pretty sure that on the last day of ALMA, when the caretaker turns the key to turn the instrument off, the only one of the enhancements we have discussed which will still be present will be the ACA.  All bands will be instrumented, but not with the same receivers which were present on 31 Dec 2010 when construction was finished.

(3) Arm-to-interarm contrast:
- Fidelity of images at extremely high level may not be the limiting
factor that controls conclusions. The arms of spiral galaxies are
not always well-defined, and identification of arms may limit the
robustness. The situation becomes more difficult for flocculent
galaxies. We should examine if different researchers independently
obtain consistent value of the arm-to-interarm contrast with sample
images to the accuracy of current interest.
It seems to me more likely that they will independently obtain consistent values from images of IF=60 than from images of IF=20.
- Uncertainty due to ambiguous luminosity-mass conversion may also
play a major role.
But that is a scientific uncertainty rather than an instrumental one.  ALMA should strive to provide images so good that all the uncertainty lies in the interpretation rather than in the image quality.

(4) Mass spectrum of clouds and cores:
- Ken Tatematsu's analysis at the face-to-face meeting clearly
demonstrated that ALMA+SD already realizes enough fidelity for this
purpose. In realistic cases, complexity of the objects (cloud/clump
overlapping, irregular structure, etc.) and robustness of simple
finding algorisms (e.g., CLUMPFIND) may limit the results.
I think there is general agreement that in this particular image ALMA+SD does a good job of providing good image fidelity.
- Uncertainty due to ambiguous luminosity-mass conversion may also
play a role.
- We should find other cases in which extremely high fidelity is
needed.
I agree.

(5) Polarization:
- Polarization study seems among which ACA might play a role. We need
examples that clearly demonstrate this. We should be sure that ACA
supports polarization capability under severe budgetary constraints
if we are to stress this science case.
I agree that this should be an ACA instrumental priority.

(6) Submillimeter observations:
- Smaller diameter of ACA antenna may be suitable for submillimeter
observations. However, current situation suggests that
submillimeter receivers that enable submillimeter observations may
be competitive with ACA. We should justify why we prefer ACA rather
than Band 8 if we are to stress this science case.
I'm not sure I understand this.  Interferometer images made with elements whose aperture efficiency is less than 40-50% do not provide quantitative science in my experience.  I think an example of a CI image could be included among science examples--a widely distributed, moderately weak line with no sharp contrasts would test the ALMA+SD imaging capacity severely at 492 GHz.
- If we are to stress this science case, optimization of the array
diameter, number of antenna, and configuration may be needed,
because the current optimization is based on imaging capability.
I'm not sure I understand--if we optimize imaging capability it seems to me that we have optimized science capability.  But as Guilloteau pointed out, all we have done is to show that the current ACA concept is sufficient, not optimal.