- Aug 19, 2011
-
-
Liam M. Healy authored
-
Liam M. Healy authored
-
- Aug 17, 2011
-
-
Liam M. Healy authored
-
- Jul 29, 2011
-
-
Liam M. Healy authored
For the Dirichlet PDF functions, the vector length should be K, the first argument, not K-1. Reported by Nils Bertschinger. Conflicts: random/dirichlet.lisp
-
Liam M. Healy authored
For the Dirichlet PDF functions, the vector length should be K, the first argument, not K-1. Reported by Nils Bertschinger.
-
- May 26, 2011
-
-
Liam M. Healy authored
-
- Apr 23, 2011
-
-
Liam M. Healy authored
-
- Jan 12, 2011
-
-
Liam M. Healy authored
-
Liam M. Healy authored
Don't import grid:element-type, grid:foreign-array, and grid:matrix, but rather refer to them with package prefix explicitly. Superficially checked but requires more careful check.
-
- Jan 10, 2011
-
-
Liam M. Healy authored
-
- Jan 07, 2011
-
-
Liam M. Healy authored
Change literal floats in tests from an unspecified precision to explicit single floats, so that if *read-default-float-format* is something other than 'single-float the tests will still pass. About half the failures now fixed. Function #'cdot is returning a pointer instead of a numerical value.
-
- Jan 01, 2011
-
-
Liam M. Healy authored
-
- Nov 27, 2010
-
-
Liam Healy authored
In cl-convert-form, use fsbv:object on foreign structs if FSBV has been loaded. Previously, the conversion of (complex single-float) was wrong because it assumed numbers were double-float; the test results for dot and cdot have been fixed.
-
- Oct 03, 2010
-
-
Liam Healy authored
-
- Oct 01, 2010
-
-
Sumant Oemrawsingh authored
Effectively, non-radix-2 algorithm tests on power-of-2 vector sizes are added. The result is, that some tests fail, and the cause at this point is not yet clear. The tests that fail, are all related to non-radix-2 algorithms working on power-of-2 vector sizes. Only tests that are missing, are the so-called bitreverse tests.
-
- Sep 15, 2010
-
-
Sumant Oemrawsingh authored
Previously, each element that was checked off-stride would count as an assertion.
-
Sumant Oemrawsingh authored
-
Sumant Oemrawsingh authored
For stride > 1 elements, the results are compared only on-stride, while (for complex only) it is verified that off-stride elements are left alone. Tests are still disabled due to time it takes to complete tests. Tests are not yet complete wrt GSL's FFT tests.
-
- Sep 04, 2010
-
-
Liam Healy authored
-
- Aug 23, 2010
-
-
Liam Healy authored
The FFT tests were testing nothing of consequence, just that NIL was equal to NIL, because it was looking at the result of (reset-urand) and then three non-existent multiple values. This is now fixed, so that fft-*-result-check actually binds the results of the calculations. Also, single-float tests are referenced to +sgl-epsilon+ and not +dbl-epsilon+. However, for (all-fft-test-forms 9 3 (64 99)), many tests fail: FAST-FOURIER-TRANSFORM: 235 assertions passed, 55 failed. so the tests are commented out.
-
- Aug 14, 2010
-
-
Liam Healy authored
Add FFT tests; these are a subset of the GSL tests. For some reason the tests take a lot longer in GSLL, so instead of testing all sizes 1...99, we test only 1...9 and then 64 and 99 for stride 1 only.
-
- Aug 08, 2010
-
-
Liam Healy authored
The standard declares that 'pi is an implementation-dependent long-float, but GSL requires double float, so the constant 'dpi has been defined as (coerce pi 'double-float). This should make CLISP at the least happier.
-
- Jul 16, 2010
-
-
Liam Healy authored
Copy with #'grid:copy methods and eliminate copy-to-destination and copy-making-destination. For non-grid objects, the presence or absence of destination selects different foreign functions.
-
- Jul 14, 2010
-
-
Liam M. Healy authored
1. Fix argument order in permutation-inverse test. 2. Test results for cholesky need to nest list for array contents. 3. Fix incorrect initial contents for matrix-transpose. Now the tests results on SBCL 64 bit, GSL 1.14 are the same as the master branch: TOTAL: 4014 assertions passed, 2 failed, 0 execution errors.
-
Liam Healy authored
Define get-value and (setf get-value) methods for matrices, and compute two indices when used, by making use of affi::delinearize-index. Add &rest to *defmfun-llk* so that it won't be included in the ignore list. Minor fixes to make more tests work.
-
- Jul 08, 2010
-
-
Liam Healy authored
Fix polynomial and numerical-integration tests by initializing with double-floats.
-
- Jul 07, 2010
-
-
Liam M. Healy authored
Do not use cl-array in tests because it isn't always available. The safer way is by doing a copy out to a new CL array.
-
Liam Healy authored
The generic function #'copy now needs a keyword argument for the destination :grid-type. Eliminate unnecessary copy tests; these are better tested in grid as they didn't use any GSL functions.
-
Liam Healy authored
Since the copy generic function has been established in grid, eliminate copy-to-destination for arrays, mvectors, matrices.
-
- Jul 01, 2010
-
-
Liam Healy authored
Use #'gsl:size to call #'grid:total-size for foreign arrays, and use the former function exclusively in GSLL. Fix some more tests.
-
Liam Healy authored
In the tests, complex arrays were initialized with reals (giving real, imaginary in sequence), but that is no longer permitted, so these were converted into complex.
-
- Jun 30, 2010
-
-
Liam Healy authored
-
- Jun 02, 2010
-
-
Liam M. Healy authored
-
Liam M. Healy authored
The gamma tests that were failing under Debian Lenny/SBCL 1.0.18/GSL 1.11 pass under Debian squeeze/SBCL 1.0.36.25/GSL 1.14, so they are now uncommented. The exponential-function test that failed in the first case passes in the second. TOTAL: 4062 assertions passed, 2 failed, 0 execution errors.
-
Liam Healy authored
The macro #'assert-to-tolerance was not correct but now maps to sf-check-results. The order of arguments in the sf-frac-diff call in sf-check-single has reversed to match test_sf_check_result in the GSL code. The remaining exponential-function tests have been ported except for the one involving gsl_sf_exprel_n_CF_e which hasn't been ported. All the tests except one pass.
-
- Jun 01, 2010
-
-
Liam M. Healy authored
Fixed the argument order to an exponential-functions test. Error estimates used in special function calculations are optional in sf-check-results. There is an exponential-functions test which fails even though it is identical to a GSL test; commented out.
-
Liam Healy authored
The return value from #'exp-err-scaled should be sf-result-e10, now fixed. Test macros renamed, made to match their corresponding GSL versions better, and moved to augment.lisp. Some of the GSL exponential function tests are now ported, and two fail.
-
- May 30, 2010
-
-
Liam Healy authored
-
Liam Healy authored
The special function tests now match much better the GSL (C) tests, and the gamma function tests ported from C all pass except for a few. These few also fail in small C program tests, but for some reason pass in the GSL "make check" test.
-
- May 28, 2010
-
-
Liam Healy authored
The special function test for gamma from GSL are in place, and 32 fail, some by a substantial amount. Some are commented out.
-