How does Math::GSL accept data?

[hda]

Oh wise monks!

I am trying to use Math::GSL::histogram to construct an histogram with logarithmic bins. What I want to do is to read the data from an array or a PDL piddle. However, Math::GSL seems to require a "stream", i.e. a file to read data from. I tried to save my data to a file in several formats but I could not manage to get Math::GSL to read it. I am obviously doing something wrong and I will appreciate any hint from you!

Here is my code, it terminates with segment fault:

my $fo = gsl_fopen('myfile', 'w');  # sets the output file
my $fi = gsl_fopen( $input_stream ); # here is where I can't understan
+d how it works
my $h = gsl_histogram_alloc (5);
gsl_histogram_fscanf($fi, $h);
my $range = [ 1.0, 10.0, 100.0, 1000.0, 10000.0, 100000.0 ];
gsl_histogram_set_ranges($h, $range, 4);
gsl_histogram_fprintf($fo, $h, "%g", "%g");
gsl_fclose($fi);
gsl_fclose($fo);

Thanks in advance!

[Anonymous Monk]

http://cpansearch.perl.org/src/LETO/Math-GSL-0.18/t/Histogram.t

    $stream = gsl_fopen("histogram", 'r');
    my $h = gsl_histogram_alloc(100);  
    ok_status(gsl_histogram_fread($stream, $h));

so

my $fi = gsl_fopen( 'infile','r');

[hda]

Hi! thanks for your answer, but unfortunately I cannot make it to work. Here is the beggining of my data:

1 3 25 1 5 1 3 1 2 2 3 3 3 25 1 1 5 2 4500 1 1 1 1 5 5 8000 1 1 5 500 1 1 1 5 2 1 1 1000 1000 1000 1000 5000 1 25 6000 500 25 25 9000 1 1 5 1000 6000 25 10 5 5 5 5 1000 9540 1000 5 1 5 500 2500 2 5 5 5 ...

Is just a sequence of integers, stored in a plain text file called "data".

Now, this is my code:



my $fi = gsl_fopen('data', 'r');
my $fo = gsl_fopen('hist', 'w');
my $h = gsl_histogram_alloc(6);
my $range = [ 1.0, 10.0, 100.0, 1000.0, 10000.0, 100000.0 ];
gsl_histogram_fread($fi, $h);
gsl_histogram_set_ranges($h,$range,7);
gsl_histogram_fprintf($fo, $h, "%3.1f", "%3.1f");
gsl_fclose($fo);
gsl_fclose($fi);


What I need is to produce a histogram with 6 logarithmic bins (see the $range variable in the code). When I run this code I get no errors, and the result is:

1.0 10.0 0.0
10.0 100.0 0.0
100.0 1000.0 0.0
1000.0 10000.0 0.0
10000.0 100000.0 0.0
100000.0 0.0 0.0


So, either no data is read or something of what I have no idea is going on. In the documentation one finds:

gsl_histogram_fread($stream, $h) - This function reads into the histogram $h from the open stream $stream, which has been opened by the gsl_fopen function from the Math::GSL module, in binary format. The histogram $h must be preallocated with the correct size since the function uses the number of bins in $h to determine how many bytes to read. The return value is 0 for success and $GSL_EFAILED if there was a problem reading from the file. The data is assumed to have been written in the native binary format on the same architecture.

But I find it a bit complicated to understand if a text file can be read or not. Oh wise monks! any hint will be appreciated!

[Anonymous Monk]

Well, I don't know Perl, but I do know GSL. Statemens in your code are in a wrong order. 1) Allocate 6 bins gsl_histogram* h=gsl_histogram_alloc(6); 2) Set the ranges. 6 bins require 7 points. (you have 6) double ranges7={1.,2.,3.,4.,5.,6.,7.}; gsl_histogram_set_ranges(h,ranges,7); 3) And only after initialization one can use the histogram 4) In C one should free the histogram after using it. No idea about Perl.

[etj]

gsl_histogram is currently not part of the PDL GSL bindings. I would have expected either PDL::Basic's hist or PDL::NDBin to accommodate user-provided ranges, in line with your requirement, but it seems neither of them do. Pull requests to rectify any or all of those would be most welcome!

[Anonymous Monk]

I didn't investigate what's wrong with OP's code and how to use Math::GSL for the task, but

I would have expected either PDL::Basic's hist or PDL::NDBin to accommodate user-provided ranges, in line with your requirement, but it seems neither of them do.

I think you are wrong here, see Variable-width bins

pdl> p$data = pdl qw/ 1 3 25 1 5 1 3 1 2 2 3 3 3 25 1 1 5 2 4500 1 1 1
+ 1 5 5 8 000 1 1 5 500 1 1 1 5 2 1 1 1000 1000 1000 1000 5000 1 25 60
+00 500 25 25 9000 1 1 5 1000 6000 25 10 5 5 5 5 1000 9540 1000 5 1 5 
[phone]/
[1 3 25 1 5 1 3 1 2 2 3 3 3 25 1 1 5 2 4500 1 1 1 1 5 5 8000 1 1 5 500
+ 1 1 1 5 2 1 1 1000 1000 1000 1000 5000 1 25 6000 500 25 25 9000 1 1 
[phone] 5 5 1000 9540 1000 5 1 5 500 2500 2 5 5 5]
pdl> $binner = PDL::NDBin->new(axes=>[[x => grid => 10 ** sequence 6]]
+)
pdl> $binner->feed(x => $data)
pdl> $binner->process
pdl> p $binner->output->{histogram}
[47 7 3 15 0]
pdl>

[Anonymous Monk]

sorry, Variable-width bins :)