1. 02 Jul, 2010 1 commit
  2. 24 Jun, 2010 2 commits
  3. 10 Jun, 2010 1 commit
    • Liam M. Healy's avatar
      Fix fsbv:defcfun call · 451bceda
      Liam M. Healy authored
      The expansion of a foreign structure-by-value function did not
      substitute the foreign function name.  It turns out this didn't matter
      because cffi::parse-name-and-options used by fsbv:defcfun simply
      ignored the symbol 'GSL-NAME for the foreign function name and pulled
      the actual name out of the symbol-name for the Lisp symbol of the
      function name.  This fix makes it clearer in the macroexpansion
      however.
      451bceda
  4. 08 Jun, 2010 1 commit
  5. 06 Jun, 2010 1 commit
    • Liam Healy's avatar
      Eliminate :cl-array argument to make-marray · 3f62c7f6
      Liam Healy authored
      The :cl-array argument to make-marray is obsolete and has been
      eliminated; the :data argument has replaced it (the argument must have
      been make by c-array:make-ffa).  Added initial-element and
      initial-contents arguments to grid:make-grid-data method.
      3f62c7f6
  6. 02 Jun, 2010 3 commits
    • Liam M. Healy's avatar
      a944009d
    • Liam M. Healy's avatar
      All gamma tests uncommented · 9f12e41e
      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.
      9f12e41e
    • Liam Healy's avatar
      Fix assert-to-tolerance; remaining exponential-function tests ported · c26f7f8e
      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.
      c26f7f8e
  7. 01 Jun, 2010 3 commits
  8. 30 May, 2010 2 commits
  9. 28 May, 2010 1 commit
    • Liam Healy's avatar
      Tests for gamma special function · 5ebd8df9
      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.
      5ebd8df9
  10. 27 May, 2010 2 commits
  11. 26 May, 2010 3 commits
  12. 25 May, 2010 1 commit
    • Liam Healy's avatar
      New random sample tests; rewrite testpdf · ff9126d9
      Liam Healy authored
      Rewrote testpdf so that it takes a function of the random variable,
      and the list of distribution name and keyword-argument pair.  Added
      new tests levy and exponential-power; several more (beta, chi-squared,
      exponential) written but commented out because they take an
      unreasonably long time to complete; these require further
      investigation.
      ff9126d9
  13. 24 May, 2010 1 commit
  14. 23 May, 2010 2 commits
  15. 22 May, 2010 4 commits
    • Liam Healy's avatar
      Added tests from GSL for more random distributions · a5476401
      Liam Healy authored
      Added chi-squared (which was not loaded at all), exponential,
      exponential-power, fdist, gamma-randist, tdist.  Introduced new macros
      assert-posinf and assert-neginf to check for expected infinite
      results.  Improved the automatic test converter and noted its known
      failings.
      a5476401
    • Liam Healy's avatar
      Replace random/gaussian tests with GSL tests · e9017ff3
      Liam Healy authored
      The tests for #'ugaussian-p, #'ugaussian-q, #'ugaussian-pinv,
      #'ugaussian-qinv are automatically converted from the GSL tests
      instead of converted from the examples.
      e9017ff3
    • Liam Healy's avatar
      Convert GSL test file to a list of CL forms · 14968147
      Liam Healy authored
      Convert the cdf/test.c GSL test file to a list of CL forms.  This
      requires cl-ppcre and iterate, and is not normally loaded with either
      gsll or gsll-tests.
      14968147
    • Liam Healy's avatar
      New macro assert-to-tolerance and constants defining tolerances · 039d1089
      Liam Healy authored
      Introduced a new macro #'assert-to-tolerance to be used in
      lisp-unit:define-test forms that takes three arguments: form
      expected-value tolerance.  This replaces test-tol6 that was defined in
      tdist.lisp.  Defined the tolerance constants +test-tol6+ the exact
      same way as GSL, but grovelling machine.h where GSL_DBL_EPSILON is
      defined.
      039d1089
  16. 26 Apr, 2010 4 commits
  17. 20 Apr, 2010 1 commit
  18. 18 Apr, 2010 1 commit
  19. 17 Apr, 2010 1 commit
  20. 15 Apr, 2010 1 commit
  21. 14 Apr, 2010 1 commit
    • Liam M. Healy's avatar
      Correct return values from jacobian-elliptic-functions; add tests · aa68fe67
      Liam M. Healy authored
      The GSL function gsl_sf_elljac_e writes sn, cn, dn values to doubles
      for which the pointer has been passed in, as shown in
      /usr/include/gsl/gsl_sf_elljac.h, and not to sf-result as one might be
      led to conclude from the "_e" at the end of the name.  This is now
      fixed in GSLL, so that the tests, which sometimes passed and sometimes
      didn't depending on what was run before it, now should always work.
      The single non-error test been replaced by the 35 GSL tests in
      gsl/specfunc/test_sf.c, and they all pass in SBCL Linux AMD64.
      aa68fe67
  22. 10 Apr, 2010 1 commit
    • Liam Healy's avatar
      Use CL RNG object, and not the foreign pointer in simulated annealing · 49cb0c32
      Liam Healy authored
      Jame Wright pointed out that the function trivial-example-step tried
      to apply the function #'sample to what turns out to be a foreign
      pointer, but it should take a CL object.  This is because we are
      making a random number generator, passing the foreign pointer to GSL,
      and then it is passing that pointer back to our routines so that we
      can use it in the step function.  This has been changed so that there
      is a special variable cl-generator which is bound to the CL object in
      #'simulated-annealing, and then used in the step function, while the
      foreign pointer being returned to us is ignored.
      49cb0c32
  23. 05 Apr, 2010 1 commit
  24. 28 Mar, 2010 1 commit