Learning arrays.

[David Farber]

I am using the basic Excel help tool within my 2002 version to investigate
how arrays operate. Specifically, I started with the ROW() function. The
Excel help says that if you put, ROW() into a cell, I used C4, then the row
number should appear and that does work. In the second example it says that
if you put ROW(C46) and then use the Ctrl-Shift-Enter, the rows below the
formula should populate with 4, 5, 6. What happens when I try this is that I
only get a 4 displayed in the cell, C4, where the formula is. The other
cells below are left blank. I tried doing the exact same thing in Open
Office Calc and it worked perfectly. Is there some setting or feature that
needs to be changed or added in Excel? By the way, the array curly brackets
do get placed around the formula after I hit Ctrl-Shift-Enter.

Thanks for your reply.

 

[GS]

David Farber pretended :

I am using the basic Excel help tool within my 2002 version to investigate
how arrays operate. Specifically, I started with the ROW() function. The
Excel help says that if you put, ROW() into a cell, I used C4, then the row
number should appear and that does work. In the second example it says that
if you put ROW(C46) and then use the Ctrl-Shift-Enter, the rows below the
formula should populate with 4, 5, 6. What happens when I try this is that I
only get a 4 displayed in the cell, C4, where the formula is. The other cells
below are left blank. I tried doing the exact same thing in Open Office Calc
and it worked perfectly. Is there some setting or feature that needs to be
changed or added in Excel? By the way, the array curly brackets do get placed
around the formula after I hit Ctrl-Shift-Enter.

Thanks for your reply.

Formulas *always* must begin with *=* sign

--
Garry

Free usenet access at http://www.eternal-september.org
Classic VB Users Regroup!
comp.lang.basic.visual.misc
microsoft.public.vb.general.discussion

 

[David Farber]

David Farber pretended :

Formulas *always* must begin with *=* sign

Yes, you are correct. I did put in the equals sign in the spreadsheet. For
some reason, I thought that would have been implied in my question based on
my results but I should have known better. Still looking for the answer.

Thanks for your reply.

 

[GS]

David Farber formulated the question :

Yes, you are correct. I did put in the equals sign in the spreadsheet. For
some reason, I thought that would have been implied in my question based on
my results but I should have known better. Still looking for the answer.

Thanks for your reply.

Array formulas can be entered in a *single* cell then *copied* to any
other cells that you want to use the array formula in, *OR* you can
select a contiguous range of cells to enter the same formula in all the
selected cells. The difference in behavior is as follows...

When I *select A1:A3* and enter "=C46" as an array formula, those 3
cells populate with 4,5,6 respectively. These cells all ref "C46".

If you enter the same array formula in B1 (single cell) and copy it
down to B2/B3, those 3 cells populate with the same results (4,5,6
respectively) *but* the refs change to "C57" in B2 and "C68" in B3.

I suspect that you did not select the 2 cells that you were expecting
"5,6" to populate. Note that formulas can not change other cells, only
the cells that contains the formula. So if C5/C6 are empty then your
formula did not populate them when you did Ctrl+Shift+Enter.

--
Garry

Free usenet access at http://www.eternal-september.org
Classic VB Users Regroup!
comp.lang.basic.visual.misc
microsoft.public.vb.general.discussion

 

[joeu2004]

The Excel help says that if you put, ROW() into a cell,
I used C4, then the row number should appear and that
does work. In the second example it says that if you put
ROW(C46) and then use the Ctrl-Shift-Enter, the rows
below the formula should populate with 4, 5, 6. What
happens when I try this is that I only get a 4 displayed
in the cell, C4, where the formula is. The other cells
below are left blank.

You probably overlooked the subtle instructions. According to the Help
page, after you normally-enter (just press Enter as usual) the formula into
A2, the instructions ask you to select A2:A4, press F2, then press
ctrl+shift+Enter to array-enter the formula.

But please note: that method is nonsensical.

Sometimes the help pages show you how to do things in order to demonstrate
form or method. That does not mean it is the best way to accomplish the
task.

First, it is nonsensical to use ROW(C46) in that context. Since we are
entering the formula into a single column, the ROW parameter should be a
single column, e.g. ROW(C4:C6).

Second, it is nonsensical to refer to another column unnecessarily, C4:C6 in
this case. And that goes double for original range C46.

The problem that causes is: it creates a dependency on those cells. So
whenever any of C4:C6 is modified, A2:A4 will be recalculated.

Sometimes that is our intent. And even if it isn't, it is not a big deal in
this very simple case.

But generally, it could be a big deal when ROW(C4:C6) is part of a more
complicated and time-consuming formula, e.g. a lookup operation.

So ROW(A4:A6) would be a better choice in this context.

Finally, it is nonsensical to use an array-entered formula at all in this
particular example. Simply normally-enter =ROW(A4) into A2, then copy the
formula down through A6.

I avoid array-entered formulas, especially multi-cell array-entered
formulas. They are difficult to modify.

For example, if you had array-entered =ROW(A4:A6) into A2:A4, then tried to
replace the formula in A2 with =ROW(A4) as I suggested, you probably
encountered an error to the effect "cannot change part of an array".

To avoid the error, we must first select A2:A4 and delete the formula.

 

[David Farber]

You probably overlooked the subtle instructions. According to the Help
page, after you normally-enter (just press Enter as usual) the formula
into A2, the instructions ask you to select A2:A4, press F2, then press
ctrl+shift+Enter to array-enter the formula.

But please note: that method is nonsensical.

Sometimes the help pages show you how to do things in order to demonstrate
form or method. That does not mean it is the best way to accomplish the
task.

First, it is nonsensical to use ROW(C46) in that context. Since we are
entering the formula into a single column, the ROW parameter should be a
single column, e.g. ROW(C4:C6).

Second, it is nonsensical to refer to another column unnecessarily, C4:C6
in this case. And that goes double for original range C46.

The problem that causes is: it creates a dependency on those cells. So
whenever any of C4:C6 is modified, A2:A4 will be recalculated.

Sometimes that is our intent. And even if it isn't, it is not a big deal
in this very simple case.

But generally, it could be a big deal when ROW(C4:C6) is part of a more
complicated and time-consuming formula, e.g. a lookup operation.

So ROW(A4:A6) would be a better choice in this context.

Finally, it is nonsensical to use an array-entered formula at all in this
particular example. Simply normally-enter =ROW(A4) into A2, then copy the
formula down through A6.

I avoid array-entered formulas, especially multi-cell array-entered
formulas. They are difficult to modify.

For example, if you had array-entered =ROW(A4:A6) into A2:A4, then tried
to replace the formula in A2 with =ROW(A4) as I suggested, you probably
encountered an error to the effect "cannot change part of an array".

To avoid the error, we must first select A2:A4 and delete the formula.

I think the help section intended the example to be more instructional than
practically useful. Let me just finish this part of my question by saying
that the function works perfectly well in Open Office's Calc. See image
here: http://webpages.charter.net/mrfixiter/Excel/ArrayExample.jpg Maybe we
can sidestep this whole discussion and let me explain why I am even
researching this topic. I was interested in converting a Word document which
contained a contact list into Excel format. This list was to be used in a
mail merge to print labels. Long story short, I imported the list into Excel
and that worked fine except the City State and Zip were all included in one
field. I found many ways to break this into separate fields/columns but the
method I liked most required commas between the city and state and the data
document didn't have commas inserted in those positions. So I found this
link with this formula at :
http://www.excelforum.com/showthread.php?t=342639&p=876198&viewfull=1#post876198=MID(A5,SEARCH(",",A1)+2,MIN(SEARCH({0,1,2,3,4,5,6,7,8,9},A1&"0123456789"))-2-(SEARCH(",",A1)+1))....confirmed with CONTROL+SHIFT+ENTER.I was curious to know how that expression was evaluated and how I couldmodify it to search for any digit, 0-9 within the CityStateZip string. Ithought I would be able to locate where the zip code began in the string andsubsequently I could count spaces backward to find the state and then thecity. This could all be accomplished because there are no cities or stateswhich contain numbers in their names.Thanks for your reply.--David FarberLos Osos, CA

 

[David Farber]

David Farber formulated the question :

Array formulas can be entered in a *single* cell then *copied* to any
other cells that you want to use the array formula in, *OR* you can
select a contiguous range of cells to enter the same formula in all
the selected cells. The difference in behavior is as follows...

When I *select A1:A3* and enter "=C46" as an array formula, those 3
cells populate with 4,5,6 respectively. These cells all ref "C46".

If you enter the same array formula in B1 (single cell) and copy it
down to B2/B3, those 3 cells populate with the same results (4,5,6
respectively) *but* the refs change to "C57" in B2 and "C68" in
B3.
I suspect that you did not select the 2 cells that you were expecting
"5,6" to populate. Note that formulas can not change other cells, only
the cells that contains the formula. So if C5/C6 are empty then your
formula did not populate them when you did Ctrl+Shift+Enter.

Starting with a blank worksheet, when I selected cells A1:A3 and entered
"=C46" as an array formula, those 3 cells populate with zeroes. Was I
supposed to use the ROW() function there? If I do, I at least get a 4 to
appear in A1. The others are still zeroes.

When I entered the same array formula in B1 (single cell) and copied it down
to B2/B3, those 3 cells populate with zeroes. The refs did change to "C57"
in B2 and "C68" in B3. Same question as above regarding the ROW()
function.

Note that formulas can not change other cells, only

the cells that contains the formula. So if C5/C6 are empty then your
formula did not populate them when you did Ctrl+Shift+Enter.

I agree with you about the part that C5/C6 are empty but I'm not sure what
conclusion you have come to about that. Could something be amiss with the
way Excel is interpreting my "Ctrl-Shift-Enter?"

Thanks for your reply.

 

[David Farber]

Starting with a blank worksheet, when I selected cells A1:A3 and
entered "=C46" as an array formula, those 3 cells populate with
zeroes. Was I supposed to use the ROW() function there? If I do, I at
least get a 4 to appear in A1. The others are still zeroes.

When I entered the same array formula in B1 (single cell) and copied
it down to B2/B3, those 3 cells populate with zeroes. The refs did
change to "C57" in B2 and "C68" in B3. Same question as above
regarding the ROW() function.

Note that formulas can not change other cells, only

I agree with you about the part that C5/C6 are empty but I'm not sure
what conclusion you have come to about that. Could something be amiss
with the way Excel is interpreting my "Ctrl-Shift-Enter?"

Thanks for your reply.

Ok, I re-read your instructions again. When I preselect the range (and use
the ROW() function) before I enter the formula, then it works. And that
goes to your comment that it cannot change the contents of cells that aren't
selected. I hope I interpreted that correctly.

I guess the Open Office version works under a different set of rules.

Thanks for your reply.

 

[joeu2004]

[....]
I guess the Open Office version works under a different set of rules.

Open Office?!

You started by claiming you are asking about MICROSOFT Excel.

Now you say you are asking about APACHE Open Office.

That is a different animal altogether.

Any similarity in behavior is completely accidental. (Albeit perhaps by
design.)

 

[David Farber]

I am using the basic Excel help tool within my 2002
version to investigate how arrays operate.

[....]
I guess the Open Office version works under a different set of rules.

Open Office?!

You started by claiming you are asking about MICROSOFT Excel.

Now you say you are asking about APACHE Open Office.

That is a different animal altogether.

Any similarity in behavior is completely accidental. (Albeit perhaps
by design.)

In my original message I said, "...I tried doing the exact same thing in
Open Office Calc and it worked perfectly. " My thinking was at least some
part of my logic had to be correct. (-:

Thanks for your reply.

 

[joeu2004]

In my original message I said, "...I tried doing the exact
same thing in Open Office Calc and it worked perfectly. "

Ah, yes. I went looking for such a reference before posting, but I missed
it.

To be clear: you are indeed asking about Microsoft Excel.

I'm not sure what the mystery still is.

I pointed to the subtle instructions in the ROW help page. You can enter
=ROW(C46) into C4 (although =ROW(C4:C6) makes more sense for your
purposes), then select C4:C6, press F2, then press ctrl+shift+Enter.

Or with forethought, you can select C4:C46, type =ROW(C46), then press
ctrl+shift+Enter.

As I noted before, I suspect you simply typed =ROW(C46) in C4, then
pressed ctrl+shift+Enter. Of course, at that point, Excel does not know how
many rows to propagate the array formula into. That is why you see only 4
in C4.

I would be surprised the Apache Open Office spreadsheet (whatever it is
called) knows any better either, when you make the same mistake. But
anything is possible in software.

Be that as it may, I don't see how this exercise (and unnecessary way to
accomplish the same result with ROW per se) helps with your original problem
or curiosity. To wit....

I found this link with this formula at :
=MID(A5,SEARCH(",",A1)+2,MIN(SEARCH({0,1,2,3,4,5,6,7,8,9},A1&"0123456789"))
-2-(SEARCH(",",A1)+1))....confirmed with CONTROL+SHIFT+ENTER.
I was curious to know how that expression was evaluated

Since SEARCH takes only a single-valued expression for the first parameter,
array-entering the formula cause the evaluation of the 2nd SEARCH expression
to create an array. In effect, it is as you had written (if we could, which
we cannot):

MIN({SEARCH(0,A1&"0123456789"),SEARCH(1,A1&"0123456789"),...,SEARCH(9,A1&"0123456789")})

By appending "0123456789", the SEARCH will always succeed. That is needed
so that MIN does not propagate the Excel error that SEARCH would return if
it failed.

Thus, MIN sees an array of indexes (character positions) in A1&"0123456789".
By returning the smallest such index, MIN returns the position of the first
number in A1, if any.

Caveat: Microsoft Excel is inconsistent with where it permits the use such
"array expressions". For example, SUMPRODUCT(VLOOKUP(A1:A10,Table,2))
should be evaluated as if we had written
SUMPRODUCT({VLOOKUP(A1,Table,2),VLOOKUP(A2,Table,2),...,VLOOKUP(A10,Table,2)),
returning the sum of the 10 lookup results. But it does not :-(.

The point is: Always test any array-entered formula to be sure it is doing
what you intended. And avoid them when you can.

 

[David Farber]

Ah, yes. I went looking for such a reference before posting, but I
missed it.

To be clear: you are indeed asking about Microsoft Excel.

I'm not sure what the mystery still is.

I pointed to the subtle instructions in the ROW help page. You can
enter =ROW(C46) into C4 (although =ROW(C4:C6) makes more sense for
your purposes), then select C4:C6, press F2, then press
ctrl+shift+Enter.
Or with forethought, you can select C4:C46, type =ROW(C46), then
press ctrl+shift+Enter.

As I noted before, I suspect you simply typed =ROW(C46) in C4, then
pressed ctrl+shift+Enter. Of course, at that point, Excel does not
know how many rows to propagate the array formula into. That is why
you see only 4 in C4.

I would be surprised the Apache Open Office spreadsheet (whatever it
is called) knows any better either, when you make the same mistake. But
anything is possible in software.

Be that as it may, I don't see how this exercise (and unnecessary way
to accomplish the same result with ROW per se) helps with your
original problem or curiosity. To wit....




Since SEARCH takes only a single-valued expression for the first
parameter, array-entering the formula cause the evaluation of the 2nd
SEARCH expression to create an array. In effect, it is as you had
written (if we could, which we cannot):

MIN({SEARCH(0,A1&"0123456789"),SEARCH(1,A1&"0123456789"),...,SEARCH(9,A1&"0123456789")})

By appending "0123456789", the SEARCH will always succeed. That is
needed so that MIN does not propagate the Excel error that SEARCH
would return if it failed.

Thus, MIN sees an array of indexes (character positions) in
A1&"0123456789". By returning the smallest such index, MIN returns
the position of the first number in A1, if any.

Caveat: Microsoft Excel is inconsistent with where it permits the
use such "array expressions". For example,
SUMPRODUCT(VLOOKUP(A1:A10,Table,2)) should be evaluated as if we had
written
SUMPRODUCT({VLOOKUP(A1,Table,2),VLOOKUP(A2,Table,2),...,VLOOKUP(A10,Table,2)),
returning the sum of the 10 lookup results. But it does not :-(.
The point is: Always test any array-entered formula to be sure it is
doing what you intended. And avoid them when you can.

Finally, I think we're on the same wavelength now. The initial mystery is
mostly solved. That was that I needed to highlight the cells where the data
was going to be displayed before entering the formula. The other part of the
mystery was how the other software, Apache's Calc, was able to do this
without prompting. For now, I'm just going to concentrate on Excel.

The reason I chose to use this as a starting point to figure out arrays was
it seemed like a very simple example of how arrays are created. I didn't
expect that I would be writing several lengthy messages trying to figure it
out.

Now back to the big picture. I like how you explained this:

Since SEARCH takes only a single-valued expression for the first
parameter, array-entering the formula cause the evaluation of the 2nd
SEARCH expression to create an array. In effect, it is as you had
written (if we could, which we cannot):

MIN({SEARCH(0,A1&"0123456789"),SEARCH(1,A1&"0123456789"),...,SEARCH(9,A1&"0123456789")})

Now my question is, which I have been unable to figure out by
experimentation, what is the correct result of:
SEARCH({0,1,2,3,4,5,6,7,8,9},A1&"0123456789") assuming for example that A1
contains a city, state, and zip? You said it would produce an array. That
makes perfect sense to me but do I have to select a group of cells as before
and then enter the formula? And how should it be entered into the formula
box? Do I use Ctrl-Shift-Enter to add braces even though there are already
braces around the search text? Every way I try it, I can only get a single
number returned. I was hoping a nice array would have been produced
somewhere.

Just to be perfectly clear, I chose as my city, state, zip field, "anytown,
ca, 91234" and the search result returns 19 which is where the first 0
appears in the concatenated string, "anytown, ca 91234012345679"

Thanks for your reply.

 

[joeu2004]

Now my question is, which I have been unable to figure
out by experimentation, what is the correct result of:
SEARCH({0,1,2,3,4,5,6,7,8,9},A1&"0123456789") assuming
for example that A1 contains a city, state, and zip?
You said it would produce an array.

And it does. But if you simply array-enter (press ctrl+shift+Enter):

=SEARCH({0,1,2,3,4,5,6,7,8,9},A1&"0123456789")

into a single cell, Excel assumes you want to select a single element of the
array, the first element in this case.

You should have array-entered (press ctrl+shift+Enter):

=MIN(SEARCH({0,1,2,3,4,5,6,7,8,9},A1&"0123456789"))

It is MIN that knows what to do with the array.

If you want to see the array returned by array-entering that SEARCH
expression, use the Evaluate Formula operation. (It is available in Excel
2003. I don't know about Excel 2002.)

Alternatively, select B1:K1 (a row of 10 cells, not a column), type the
following formula:

=SEARCH({0,1,2,3,4,5,6,7,8,9},A1&"0123456789")

then press ctrl+shift+Enter.

 

[David Farber]

And it does. But if you simply array-enter (press ctrl+shift+Enter):

=SEARCH({0,1,2,3,4,5,6,7,8,9},A1&"0123456789")

into a single cell, Excel assumes you want to select a single element
of the array, the first element in this case.

You should have array-entered (press ctrl+shift+Enter):

=MIN(SEARCH({0,1,2,3,4,5,6,7,8,9},A1&"0123456789"))

It is MIN that knows what to do with the array.

If you want to see the array returned by array-entering that SEARCH
expression, use the Evaluate Formula operation. (It is available in
Excel 2003. I don't know about Excel 2002.)

Alternatively, select B1:K1 (a row of 10 cells, not a column), type
the following formula:

=SEARCH({0,1,2,3,4,5,6,7,8,9},A1&"0123456789")

then press ctrl+shift+Enter.

That was easy. (-; Worked like a charm. From there, I was able to save the
search value, I'll call it n, in a hidden column to make the subsequent
formulas much shorter. Then I used that value to construct another string
which started at the n-4th character (there are two spaces between state and
zipcode) and was two characters long for state. For the city, I started at
the beginning of the string and ended it at the n-5th character.

In case anyone is interested, this does work in Apache's Open Office Calc
except function arguments and array values are separated by semicolons, not
commas.

Thanks for your great help.