Another issue - perhaps one of you has encountered this before. It's sort of a neat problem. I'm attempting to build array subscripts on the fly based on user input. IE the number of static/variable elements is changing, which allows the user to pick different axes to plot. Nevermind all that.
Anyway, let's say a user wants a particular axis to be variable. In this case, the dataset array where I'm attempting to pull values from would contain a *, to get all these elements. Unfortunately, I do not know in advance which dimension of the array I will be using, so I am attempting to build the subscript based on a string.
This was my original thought: a = dindgen(5,5,5) b = ['3','3','3'] print, a[b] but this just returns a[3], a[3], a[3]
So, I figured I'd do it this way: c = '3' print, a[c,c,c] -- This works!
Now for the gold, d = '*' print, a[c,c,d] -- error - can't convert string-> long so I get an idea-- maybe I'll just use the ascii value for the asterisk. d = String(42b) print, a[c,d,d] -- error - can't convert string-> long
> Another issue - perhaps one of you has encountered this before. It's > sort of a neat problem. I'm attempting to build array subscripts on > the fly based on user input. IE the number of static/variable elements > is changing, which allows the user to pick different axes to plot. > Nevermind all that.
> Anyway, let's say a user wants a particular axis to be variable. In > this case, the dataset array where I'm attempting to pull values from > would contain a *, to get all these elements. Unfortunately, I do not > know in advance which dimension of the array I will be using, so I am > attempting to build the subscript based on a string.
> This was my original thought: > a = dindgen(5,5,5) > b = ['3','3','3'] > print, a[b] > but this just returns a[3], a[3], a[3]
> So, I figured I'd do it this way: > c = '3' > print, a[c,c,c] -- This works!
> Now for the gold, > d = '*' > print, a[c,c,d] -- error - can't convert string-> long > so I get an idea-- maybe I'll just use the ascii value for the > asterisk. > d = String(42b) > print, a[c,d,d] -- error - can't convert string-> long
> Any thoughts ? > Thanks in advance > --Justin
Why try to force the '*' - might not SIZE be more useful? e.g. s=SIZE(a) print, a[c,c,s[3]] ; for a[c,c,d] print, a[c,s[2],s[3]]; for a[c,d,d]
> On Jul 14, 11:16 am, humanumbre...@gmail.com wrote:
> > Hello all,
> > Another issue - perhaps one of you has encountered this before. It's > > sort of a neat problem. I'm attempting to build array subscripts on > > the fly based on user input. IE the number of static/variable elements > > is changing, which allows the user to pick different axes to plot. > > Nevermind all that.
> > Anyway, let's say a user wants a particular axis to be variable. In > > this case, the dataset array where I'm attempting to pull values from > > would contain a *, to get all these elements. Unfortunately, I do not > > know in advance which dimension of the array I will be using, so I am > > attempting to build the subscript based on a string.
> > This was my original thought: > > a = dindgen(5,5,5) > > b = ['3','3','3'] > > print, a[b] > > but this just returns a[3], a[3], a[3]
> > So, I figured I'd do it this way: > > c = '3' > > print, a[c,c,c] -- This works!
> > Now for the gold, > > d = '*' > > print, a[c,c,d] -- error - can't convert string-> long > > so I get an idea-- maybe I'll just use the ascii value for the > > asterisk. > > d = String(42b) > > print, a[c,d,d] -- error - can't convert string-> long
> > Any thoughts ? > > Thanks in advance > > --Justin
> Why try to force the '*' - might not SIZE be more useful? > e.g. > s=SIZE(a) > print, a[c,c,s[3]] ; for a[c,c,d] > print, a[c,s[2],s[3]]; for a[c,d,d]
Hey Bob,
Thanks for the post! I think I may need to elaborate a bit more -- I need the entire row of the multi-dimensional array. So, for example, let's say I have an array that is 30 x 20 x 50 I will need *,0,0 to plot the first 30 values but I could just as easily need 0,*,0 or 0,0,* Depending on user input, so I can't anticipate that in advance.
> On Jul 14, 11:41 am, Bob Crawford <Snowma...@gmail.com> wrote:
> > On Jul 14, 11:16 am, humanumbre...@gmail.com wrote:
> > > Hello all,
> > > Another issue - perhaps one of you has encountered this before. It's > > > sort of a neat problem. I'm attempting to build array subscripts on > > > the fly based on user input. IE the number of static/variable elements > > > is changing, which allows the user to pick different axes to plot. > > > Nevermind all that.
> > > Anyway, let's say a user wants a particular axis to be variable. In > > > this case, the dataset array where I'm attempting to pull values from > > > would contain a *, to get all these elements. Unfortunately, I do not > > > know in advance which dimension of the array I will be using, so I am > > > attempting to build the subscript based on a string.
> > > This was my original thought: > > > a = dindgen(5,5,5) > > > b = ['3','3','3'] > > > print, a[b] > > > but this just returns a[3], a[3], a[3]
> > > So, I figured I'd do it this way: > > > c = '3' > > > print, a[c,c,c] -- This works!
> > > Now for the gold, > > > d = '*' > > > print, a[c,c,d] -- error - can't convert string-> long > > > so I get an idea-- maybe I'll just use the ascii value for the > > > asterisk. > > > d = String(42b) > > > print, a[c,d,d] -- error - can't convert string-> long
> > > Any thoughts ? > > > Thanks in advance > > > --Justin
> > Why try to force the '*' - might not SIZE be more useful? > > e.g. > > s=SIZE(a) > > print, a[c,c,s[3]] ; for a[c,c,d] > > print, a[c,s[2],s[3]]; for a[c,d,d]
> Hey Bob,
> Thanks for the post! > I think I may need to elaborate a bit more -- > I need the entire row of the multi-dimensional array. > So, for example, let's say I have an array that is 30 x 20 x 50 > I will need *,0,0 to plot the first 30 values > but I could just as easily need 0,*,0 or 0,0,* Depending on user > input, so I can't anticipate that in advance.
> Cheers, > --Justin- Hide quoted text -
> - Show quoted text -
Oops. I posted too soon (thank you for the clarification Justin - that is what I was trying to do) Here is what I should have posted:
print, a[c,c,0:(s[3]-1)] ; for a[c,c,d] print, a[c,0:(s[2]-1),0:(s[3]-1)]; for a[c,d,d]
Isn't '*' just short form notation for 0:(s[n]-1), anyway?
> On Jul 14, 11:49 am, humanumbre...@gmail.com wrote:
> > On Jul 14, 11:41 am, Bob Crawford <Snowma...@gmail.com> wrote:
> > > On Jul 14, 11:16 am, humanumbre...@gmail.com wrote:
> > > > Hello all,
> > > > Another issue - perhaps one of you has encountered this before. It's > > > > sort of a neat problem. I'm attempting to build array subscripts on > > > > the fly based on user input. IE the number of static/variable elements > > > > is changing, which allows the user to pick different axes to plot. > > > > Nevermind all that.
> > > > Anyway, let's say a user wants a particular axis to be variable. In > > > > this case, the dataset array where I'm attempting to pull values from > > > > would contain a *, to get all these elements. Unfortunately, I do not > > > > know in advance which dimension of the array I will be using, so I am > > > > attempting to build the subscript based on a string.
> > > > This was my original thought: > > > > a = dindgen(5,5,5) > > > > b = ['3','3','3'] > > > > print, a[b] > > > > but this just returns a[3], a[3], a[3]
> > > > So, I figured I'd do it this way: > > > > c = '3' > > > > print, a[c,c,c] -- This works!
> > > > Now for the gold, > > > > d = '*' > > > > print, a[c,c,d] -- error - can't convert string-> long > > > > so I get an idea-- maybe I'll just use the ascii value for the > > > > asterisk. > > > > d = String(42b) > > > > print, a[c,d,d] -- error - can't convert string-> long
> > > > Any thoughts ? > > > > Thanks in advance > > > > --Justin
> > > Why try to force the '*' - might not SIZE be more useful? > > > e.g. > > > s=SIZE(a) > > > print, a[c,c,s[3]] ; for a[c,c,d] > > > print, a[c,s[2],s[3]]; for a[c,d,d]
> > Hey Bob,
> > Thanks for the post! > > I think I may need to elaborate a bit more -- > > I need the entire row of the multi-dimensional array. > > So, for example, let's say I have an array that is 30 x 20 x 50 > > I will need *,0,0 to plot the first 30 values > > but I could just as easily need 0,*,0 or 0,0,* Depending on user > > input, so I can't anticipate that in advance.
> > Cheers, > > --Justin- Hide quoted text -
> > - Show quoted text -
> Oops. > I posted too soon (thank you for the clarification Justin - that is > what I was trying to do) > Here is what I should have posted:
> print, a[c,c,0:(s[3]-1)] ; for a[c,c,d] > print, a[c,0:(s[2]-1),0:(s[3]-1)]; for a[c,d,d]
> Isn't '*' just short form notation for 0:(s[n]-1), anyway?
Hey Bob,
Yes, I think '*' is short for 0:(s[n]-1) but I read somewhere that you shouldn't use the range because of performance issues... Can anyone shed light on that issue?
> > > > > Another issue - perhaps one of you has encountered this before. It's > > > > > sort of a neat problem. I'm attempting to build array subscripts on > > > > > the fly based on user input. IE the number of static/variable elements > > > > > is changing, which allows the user to pick different axes to plot. > > > > > Nevermind all that.
> > > > > Anyway, let's say a user wants a particular axis to be variable. In > > > > > this case, the dataset array where I'm attempting to pull values from > > > > > would contain a *, to get all these elements. Unfortunately, I do not > > > > > know in advance which dimension of the array I will be using, so I am > > > > > attempting to build the subscript based on a string.
> > > > > This was my original thought: > > > > > a = dindgen(5,5,5) > > > > > b = ['3','3','3'] > > > > > print, a[b] > > > > > but this just returns a[3], a[3], a[3]
> > > > > So, I figured I'd do it this way: > > > > > c = '3' > > > > > print, a[c,c,c] -- This works!
> > > > > Now for the gold, > > > > > d = '*' > > > > > print, a[c,c,d] -- error - can't convert string-> long > > > > > so I get an idea-- maybe I'll just use the ascii value for the > > > > > asterisk. > > > > > d = String(42b) > > > > > print, a[c,d,d] -- error - can't convert string-> long
> > > > Why try to force the '*' - might not SIZE be more useful? > > > > e.g. > > > > s=SIZE(a) > > > > print, a[c,c,s[3]] ; for a[c,c,d] > > > > print, a[c,s[2],s[3]]; for a[c,d,d]
> > > Hey Bob,
> > > Thanks for the post! > > > I think I may need to elaborate a bit more -- > > > I need the entire row of the multi-dimensional array. > > > So, for example, let's say I have an array that is 30 x 20 x 50 > > > I will need *,0,0 to plot the first 30 values > > > but I could just as easily need 0,*,0 or 0,0,* Depending on user > > > input, so I can't anticipate that in advance.
> > > Cheers, > > > --Justin- Hide quoted text -
> > > - Show quoted text -
> > Oops. > > I posted too soon (thank you for the clarification Justin - that is > > what I was trying to do) > > Here is what I should have posted:
> > print, a[c,c,0:(s[3]-1)] ; for a[c,c,d] > > print, a[c,0:(s[2]-1),0:(s[3]-1)]; for a[c,d,d]
> > Isn't '*' just short form notation for 0:(s[n]-1), anyway?
> Hey Bob,
> Yes, I think '*' is short for 0:(s[n]-1) but I read somewhere that you > shouldn't use the range because of performance issues... > Can anyone shed light on that issue?
IE: From the "Help" pages on "Arrays" "Processing subscript ranges is inefficient. When possible, use an array or scalar subscript instead of specifying a subscript range where the beginning and ending subscripts are separated by the colon character."
humanumbre...@gmail.com writes: > "Processing subscript ranges is inefficient. When possible, use an > array or scalar subscript instead of specifying a subscript range > where the beginning and ending subscripts are separated by the colon > character."
-- David Fanning, Ph.D. Fanning Software Consulting, Inc. Coyote's Guide to IDL Programming: http://www.dfanning.com/ Sepore ma de ni thui. ("Perhaps thou speakest truth.")
> humanumbre...@gmail.com writes: > > "Processing subscript ranges is inefficient. When possible, use an > > array or scalar subscript instead of specifying a subscript range > > where the beginning and ending subscripts are separated by the colon > > character."
> -- > David Fanning, Ph.D. > Fanning Software Consulting, Inc. > Coyote's Guide to IDL Programming:http://www.dfanning.com/ > Sepore ma de ni thui. ("Perhaps thou speakest truth.")
It looks like the asterisk character is interpreted as something different inside the brackets of an array. This interpretation is not a string '*', so that would explain the error of trying to use the string in this context.
Is anyone familiar with how to reformat the * in such a way so as the interpreter would recognize it ?
> > > > > > Another issue - perhaps one of you has encountered this before. It's > > > > > > sort of a neat problem. I'm attempting to build array subscripts on > > > > > > the fly based on user input. IE the number of static/variable elements > > > > > > is changing, which allows the user to pick different axes to plot. > > > > > > Nevermind all that.
> > > > > > Anyway, let's say a user wants a particular axis to be variable. In > > > > > > this case, the dataset array where I'm attempting to pull values from > > > > > > would contain a *, to get all these elements. Unfortunately, I do not > > > > > > know in advance which dimension of the array I will be using, so I am > > > > > > attempting to build the subscript based on a string.
> > > > > > This was my original thought: > > > > > > a = dindgen(5,5,5) > > > > > > b = ['3','3','3'] > > > > > > print, a[b] > > > > > > but this just returns a[3], a[3], a[3]
> > > > > > So, I figured I'd do it this way: > > > > > > c = '3' > > > > > > print, a[c,c,c] -- This works!
> > > > > > Now for the gold, > > > > > > d = '*' > > > > > > print, a[c,c,d] -- error - can't convert string-> long > > > > > > so I get an idea-- maybe I'll just use the ascii value for the > > > > > > asterisk. > > > > > > d = String(42b) > > > > > > print, a[c,d,d] -- error - can't convert string-> long
> > > > > Why try to force the '*' - might not SIZE be more useful? > > > > > e.g. > > > > > s=SIZE(a) > > > > > print, a[c,c,s[3]] ; for a[c,c,d] > > > > > print, a[c,s[2],s[3]]; for a[c,d,d]
> > > > Hey Bob,
> > > > Thanks for the post! > > > > I think I may need to elaborate a bit more -- > > > > I need the entire row of the multi-dimensional array. > > > > So, for example, let's say I have an array that is 30 x 20 x 50 > > > > I will need *,0,0 to plot the first 30 values > > > > but I could just as easily need 0,*,0 or 0,0,* Depending on user > > > > input, so I can't anticipate that in advance.
> > > Oops. > > > I posted too soon (thank you for the clarification Justin - that is > > > what I was trying to do) > > > Here is what I should have posted:
> > > print, a[c,c,0:(s[3]-1)] ; for a[c,c,d] > > > print, a[c,0:(s[2]-1),0:(s[3]-1)]; for a[c,d,d]
> > > Isn't '*' just short form notation for 0:(s[n]-1), anyway?
> > Hey Bob,
> > Yes, I think '*' is short for 0:(s[n]-1) but I read somewhere that you > > shouldn't use the range because of performance issues... > > Can anyone shed light on that issue?
> IE: > From the "Help" pages on "Arrays" > "Processing subscript ranges is inefficient. When possible, use an > array or scalar subscript instead of specifying a subscript range > where the beginning and ending subscripts are separated by the colon > character."- Hide quoted text -
> - Show quoted text -
It appears from my reading of David's page that quoting the subscripts by way of using a colon is not any different (memory wise) than using a '*'.
Given the passage you've quoted above doesn't state that using a '*' is any more efficient than ':' subsetting, but perhaps doing (as suggested):
humanumbre...@gmail.com wrote: > On Jul 14, 11:41 am, Bob Crawford <Snowma...@gmail.com> wrote: >> On Jul 14, 11:16 am, humanumbre...@gmail.com wrote:
>>> Hello all, >>> Another issue - perhaps one of you has encountered this before. It's >>> sort of a neat problem. I'm attempting to build array subscripts on >>> the fly based on user input. IE the number of static/variable elements >>> is changing, which allows the user to pick different axes to plot. >>> Nevermind all that. >>> Anyway, let's say a user wants a particular axis to be variable. In >>> this case, the dataset array where I'm attempting to pull values from >>> would contain a *, to get all these elements. Unfortunately, I do not >>> know in advance which dimension of the array I will be using, so I am >>> attempting to build the subscript based on a string. >>> This was my original thought: >>> a = dindgen(5,5,5) >>> b = ['3','3','3'] >>> print, a[b] >>> but this just returns a[3], a[3], a[3] >>> So, I figured I'd do it this way: >>> c = '3' >>> print, a[c,c,c] -- This works! >>> Now for the gold, >>> d = '*' >>> print, a[c,c,d] -- error - can't convert string-> long >>> so I get an idea-- maybe I'll just use the ascii value for the >>> asterisk. >>> d = String(42b) >>> print, a[c,d,d] -- error - can't convert string-> long >>> Any thoughts ? >>> Thanks in advance >>> --Justin >> Why try to force the '*' - might not SIZE be more useful? >> e.g. >> s=SIZE(a) >> print, a[c,c,s[3]] ; for a[c,c,d] >> print, a[c,s[2],s[3]]; for a[c,d,d]
> Hey Bob,
> Thanks for the post! > I think I may need to elaborate a bit more -- > I need the entire row of the multi-dimensional array. > So, for example, let's say I have an array that is 30 x 20 x 50 > I will need *,0,0 to plot the first 30 values > but I could just as easily need 0,*,0 or 0,0,* Depending on user > input, so I can't anticipate that in advance.
What about using execute? I didn't have any problems constructing a string to execute that included the '*' character:
pro testit, n a=indgen(30,20,50) help, a info=size(a,/structure)
index = make_array(info.n_dimensions,value='0') index[n] = '*'
exestring = 'x = reform(a['+strjoin(index,',')+'])' result = execute(exestring) help, x end
IDL> testit,0 A INT = Array[30, 20, 50] X INT = Array[30] IDL> testit,1 A INT = Array[30, 20, 50] X INT = Array[20] IDL> testit,2 A INT = Array[30, 20, 50] X INT = Array[50]
> humanumbre...@gmail.com wrote: > > On Jul 14, 11:41 am, Bob Crawford <Snowma...@gmail.com> wrote: > >> On Jul 14, 11:16 am, humanumbre...@gmail.com wrote:
> >>> Hello all, > >>> Another issue - perhaps one of you has encountered this before. It's > >>> sort of a neat problem. I'm attempting to build array subscripts on > >>> the fly based on user input. IE the number of static/variable elements > >>> is changing, which allows the user to pick different axes to plot. > >>> Nevermind all that. > >>> Anyway, let's say a user wants a particular axis to be variable. In > >>> this case, the dataset array where I'm attempting to pull values from > >>> would contain a *, to get all these elements. Unfortunately, I do not > >>> know in advance which dimension of the array I will be using, so I am > >>> attempting to build the subscript based on a string. > >>> This was my original thought: > >>> a = dindgen(5,5,5) > >>> b = ['3','3','3'] > >>> print, a[b] > >>> but this just returns a[3], a[3], a[3] > >>> So, I figured I'd do it this way: > >>> c = '3' > >>> print, a[c,c,c] -- This works! > >>> Now for the gold, > >>> d = '*' > >>> print, a[c,c,d] -- error - can't convert string-> long > >>> so I get an idea-- maybe I'll just use the ascii value for the > >>> asterisk. > >>> d = String(42b) > >>> print, a[c,d,d] -- error - can't convert string-> long > >>> Any thoughts ? > >>> Thanks in advance > >>> --Justin > >> Why try to force the '*' - might not SIZE be more useful? > >> e.g. > >> s=SIZE(a) > >> print, a[c,c,s[3]] ; for a[c,c,d] > >> print, a[c,s[2],s[3]]; for a[c,d,d]
> > Hey Bob,
> > Thanks for the post! > > I think I may need to elaborate a bit more -- > > I need the entire row of the multi-dimensional array. > > So, for example, let's say I have an array that is 30 x 20 x 50 > > I will need *,0,0 to plot the first 30 values > > but I could just as easily need 0,*,0 or 0,0,* Depending on user > > input, so I can't anticipate that in advance.
> What about using execute? I didn't have any problems constructing a string to execute that > included the '*' character:
> pro testit, n > a=indgen(30,20,50) > help, a > info=size(a,/structure)
> index = make_array(info.n_dimensions,value='0') > index[n] = '*'
> exestring = 'x = reform(a['+strjoin(index,',')+'])' > result = execute(exestring) > help, x > end
> IDL> testit,0 > A INT = Array[30, 20, 50] > X INT = Array[30] > IDL> testit,1 > A INT = Array[30, 20, 50] > X INT = Array[20] > IDL> testit,2 > A INT = Array[30, 20, 50] > X INT = Array[50]
> ??
> cheers,
> paulv
Paulv,
You're the champion of the day! Thanks kindly for your suggestion!
> It looks like the asterisk character is interpreted as something > different inside the brackets of an array. > This interpretation is not a string '*', so that would explain the > error of trying to use the string in this context.
> Is anyone familiar with how to reformat the * in such a way so as the > interpreter would recognize it ?
> Thanks in advance, > --Justin
well... Bob gave you two solutions already 0:s[n] and indgen(s[n])
Now if you really want to use the *, you might want to have a look at the command "execute" IDL> a = '*' IDL> b = indgen(3,3) IDL> c = '1' IDL> tmp= execute('e = b[' + c + ',' + a +']') IDL> print,e 1 4 7
but be aware of the limitations of this command (read the help) Jean
On Jul 14, 3:44 pm, Jean H <jghas...@DELTHIS.ucalgary.ANDTHIS.ca> wrote:
> > b=INDGEN(s(3)) > > print, a[c,c,b]
> > might be more efficient - dunno, haven't tested.
> You would have to reformat (i.e. repeat the value) c so it has the same > size as b.
> Jean
Thanks Bob,
In my particular application, it was difficult to use the solution involving the range, because I had to put ':' in the dataset access function. I do not know how many dimensions are in advance, so I would need to switch on the number of dimensions in order to put the right number of ':' in. I could be missing something,
Thanks Jean for the caution on the execute command, and I will check out the help menu.
|
