Indexing to table subset from unique

Have a fairly complex table that contains, amongst other things, a Formula field (altho what the field is is immaterial to the question, just for specifics to talk about) that, in part, looks like...

>> tS62671=tShow(tShow.Acct=='S62671',:)
tS62671 =
  75×12 table
    Date      Acct      Corpus      Realized    Unrealized      Value               Formula                 Gift           GiftDate          Donor            Note          Comments  
    _____    ______    _________    ________    __________    _________    _________________________    ____________    ______________    ____________    ____________    ____________
    Mar13    S62671    200000.00        0.00         0.00     200000.00    {'=185943.12+14056.88'  }    {2×1 double}    {2×1 datetime}    {2×1 string}    {2×1 string}    {2×4 string}
    Apr13    S62671    200000.00        0.33         0.00     200000.33    {'200000'               }    {0×0 double}    {0×0 double  }    {0×0 double}    {0×0 double}    {0×0 double}
    May13    S62671    200000.00     -871.33     -1625.31     197503.36    {'200000'               }    {0×0 double}    {0×0 double  }    {0×0 double}    {0×0 double}    {0×0 double}
    Jun13    S62671    200000.00     -435.81     -3994.75     195569.44    {'200000'               }    {0×0 double}    {0×0 double  }    {0×0 double}    {0×0 double}    {0×0 double}
    Jun14    S62671    200000.00     8609.48     -9999.63     198609.85    {'200000'               }    {0×0 double}    {0×0 double  }    {0×0 double}    {0×0 double}    {0×0 double}
    Jul14    S62671    200200.00    10080.07    -12645.56     197634.51    {'=200000+200'          }    {0×0 double}    {0×0 double  }    {0×0 double}    {0×0 double}    {0×0 double}
    ...
    Aug14    S62671    200200.00    11374.44    -12116.10     199458.34    {'=200000+200'          }    {0×0 double}    {0×0 double  }    {0×0 double}    {0×0 double}    {0×0 double}
    Sep14    S62671    211645.04    13268.32    -14577.51     210335.85    {'=200200+11445.04'     }    {0×0 double}    {0×0 double  }    {0×0 double}    {0×0 double}    {0×0 double}
    Oct14    S62671    219145.04    14879.44    -15271.01     218753.47    {'=200200+11445.04+7500'}    {0×0 double}    {0×0 double  }    {0×0 double}    {0×0 double}    {0×0 double}
    ...
    Mar15    S62671    219145.04    22978.06    -19844.04     222279.06    {'=200200+11445.04+7500'}    {0×0 double}    {0×0 double  }    {0×0 double}    {0×0 double}    {0×0 double}
    Apr15    S62671    249145.04    24888.67    -22428.68     251605.03    {'=219145.04+30000'     }    {0×0 double}    {0×0 double  }    {0×0 double}    {0×0 double}    {0×0 double}
    May15    S62671    249145.04    26977.14    -24702.08     251420.10    {'249145.04'            }    {0×0 double}    {0×0 double  }    {0×0 double}    {0×0 double}    {0×0 double}
    Jun15    S62671    200000.00    29059.97    -28400.44     200659.53    {'=249145.04-49145.04'  }    {4×1 double}    {4×1 datetime}    {4×1 string}    {4×1 string}    {4×4 string}
    Jul15    S62671    200000.00    30926.96    -29425.22     201501.74    {'=249145.04-49145.04'  }    {2×1 double}    {2×1 datetime}    {2×1 string}    {2×1 string}    {2×4 string}
    Aug15    S62671    200100.00    32806.59    -31009.28     201897.31    {'=200000+100'          }    {2×1 double}    {2×1 datetime}    {2×1 string}    {2×1 string}    {2×4 string}
    ...

Underneath this if have

fnAddn=(@(s) find(contains(s,{'='}) & contains(s,{'+'})));      % only formula rows with additions 
fnSubn=(@(s) find(contains(s,{'='}) & contains(s,{'-'})));      % only formula rows with withdrawals

What I need is then the first month for each unique addition/withdrawal entry --

[uf,ia]=unique(tS62671(fnAddn(tS62671.Formula),'Formula'));

which leads to the question -- is there a way to dereference the original table to return the unique rows in tS62671 other than by creating a temporary table first before the unique operation?

[dpb]

The not-too-bad workaround came to me just after posting but I'll leave the Q? anyway as it is pretty generic overall with just a specific example to illustrate...

It's not terrible to save the indexing variable first instead of wrapping the subset operation into the reference...

ixAdd=fnAddn(tS62671.Formula);
tS62671Addns=tS62671(ixAdd(ia),'Formula');

[Asvin Kumar]

Just out of curiosity,

"It's not terrible to save the indexing variable first instead of wrapping the subset operation into the reference..."

You found the indexing variable 'ia' in the first place by wrapping the subset operations within the unique function call. Would your workaround still hold if you didn't have the 'ia'? You won't be able to meaningfully call unique on 'ixAdd'. Looks like we are still dependent on indexing a part of the original table to pass as input to unique.

[dpb]

Yeah, that's the problem...you lose the reference to the original otherwise...being unable to dereference a function expression syntactically. I think Octave has some facilities along that line but not sure can do this and don't have install to test.