[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#768936: ITP: nufft -- Library implementing the Non-Uniform Fast Fourier Transform



Chris Bannister wrote:
> Ghislain Antony Vaillant wrote:
>> Package: wnpp
>> Severity: wishlist
>> Owner: Ghislain Antony Vaillant <ghisvail@gmail.com>
>> 
>> * Package name    : nufft
>>   Version         : 1.3.3
>>   Upstream Author : Leslie Greengard <greengard@cims.nyu.edu>
>> * URL             : http://www.cims.nyu.edu/cmcl/nufft/nufft.html
>> * License         : BSD
>>   Programming Lang: FORTRAN
>>   Description     : Library implementing the Non-Uniform Fast Fourier Transform
>> 
>> Implementation of the Non-Uniform Fast Fourier Transform (NUFFT) coded in 
>> FORTRAN using a fast, procedural, algorithm. 
                       ^           ^
>> Compared to existingly packaged solutions, like the NFFT library, the NUFFT 
>> provides Octave (and MATLAB) compatible bindings. Because of the procedural 
>> implementation, usage of the NUFFT is more straightforward. However, it lacks 
>> support for high dimensionality than 3 and support for precomputation in case
>              ^^^^^^^^^^^^^^^^^^^^^^^^^^
> I think that would be better worded as "dimensions greater than three"
> or similar.

Agreed.

Also, surplus commas in the first paragraph; and I'd recommend
rearranging it to avoid the misreading that it was coded using this
algorithm.

   This is a FORTRAN implementation of the Non-Uniform Fast Fourier Transform
   (NUFFT) using a fast procedural algorithm.
 
Then in the second paragraph, "existingly packaged" has got to go;
and that first sentence never does explicitly compare anything - it's
only implied that NFFT doesn't provide these bindings.  Say:

   Unlike solutions such as the NFFT library, the NUFFT provides Octave (and
   MATLAB) compatible bindings. [...]

Then at the end:

>> support for high dimensionality than 3 and support for precomputation in case
>> a transform needs to be repeated. As a result both the NFFT and NUFFT package
>> may coexist as they fit different needs.

Nitpicking, "it lacks support for A and support for B" is needless
repetition, the "in case" should be "in the case that" or simply "if",
and "both [...] may coexist" is redundant and sounds as if you don't
know whether in fact they do exist.  So:

   This is a FORTRAN implementation of the Non-Uniform Fast Fourier Transform
   (NUFFT) using a fast procedural algorithm.
   .
   Unlike solutions such as the NFFT library, the NUFFT provides Octave (and
   MATLAB) compatible bindings. Because of the procedural implementation, usage
   of the NUFFT is more straightforward. However, it lacks support for
   dimensions greater than three or for precomputation if a transform needs to
   be repeated. As a result the NFFT and NUFFT packages fit different needs and
   can coexist.

-- 
JBR	with qualifications in linguistics, experience as a Debian
	sysadmin, and probably no clue about this particular package


Reply to: