|
||||||||||
PREV CLASS NEXT CLASS | FRAMES NO FRAMES | |||||||||
SUMMARY: NESTED | FIELD | CONSTR | METHOD | DETAIL: FIELD | CONSTR | METHOD |
java.lang.Objectbb.science.Math2.UnitTest
public static class Math2.UnitTest
See the Overview page of the project's javadocs for a general description of this unit test class.
Constructor Summary | |
---|---|
Math2.UnitTest()
|
Method Summary | |
---|---|
void |
benchmark_statisticsFunctions()
Results on 2009-03-16 (2.5 GHz Xeon E5420 desktop, jdk 1.6.0_11 server jvm):
n = 1000
gaussianCdf: first = 144.364 ns, mean = 8.418 ns (CI deltas: -1.923 ps, +2.269 ps), sd = 94.479 ns (CI deltas: -15.358 ns, +22.652 ns) WARNING: execution times have mild outliers, SD VALUES MAY BE INACCURATE
|
void |
test_autocorrelationStatisticsMethods()
|
void |
test_bitTwiddling()
|
void |
test_gaussianAndersonDarling()
When last looked at the output on 2009-09-24, consistently saw:
andersonDarling1 (Gaussian sample) = 0.13459667082165735, andersonDarling2 (uniform distribution sample) = 11170.383267133551
andersonDarling1 (Gaussian sample) = 0.18940867401726738, andersonDarling2 (uniform distribution sample) = 11064.463336125908
andersonDarling1 (Gaussian sample) = 0.3984548680792433, andersonDarling2 (uniform distribution sample) = 11038.706942927185
andersonDarling1 (Gaussian sample) = 0.41313940626760093, andersonDarling2 (uniform distribution sample) = 11217.005208263494
andersonDarling1 (Gaussian sample) = 0.27588003071356976, andersonDarling2 (uniform distribution sample) = 11115.653184588584
The average value above for andersonDarling1 of ~0.25 compares well with the value 0.2576117
that is reported in the "anderson darling normal test y1" section of
this reference. |
void |
test_gaussianFit()
|
void |
test_gaussianKolmogorovSmirnov()
When last looked at the output on 2009-09-24, consistently saw:
kolmogorovSmirnov1 (Gaussian sample) = 0.5506861389029671, kolmogorovSmirnov2 (uniform distribution sample) = 57.55263461482368
kolmogorovSmirnov1 (Gaussian sample) = 0.7466358894082092, kolmogorovSmirnov2 (uniform distribution sample) = 57.34064194461067
kolmogorovSmirnov1 (Gaussian sample) = 0.7624236873765011, kolmogorovSmirnov2 (uniform distribution sample) = 57.16759617069217
kolmogorovSmirnov1 (Gaussian sample) = 0.836100754503244, kolmogorovSmirnov2 (uniform distribution sample) = 57.28774638323175
kolmogorovSmirnov1 (Gaussian sample) = 0.6035095000266333, kolmogorovSmirnov2 (uniform distribution sample) = 57.28436577988172
The average value above for kolmogorovSmirnov1 of ~0.65 is well under the critical value (so this test correctly accepts 1 as a Gaussian). |
void |
test_infinityBehavior()
Confirms that certain corner cases concerning floating point infinity arithmetic behave as my code above assumes. |
void |
test_infoTheory()
|
void |
test_linearLeastSquaresFit()
|
void |
test_logsPowersEtc()
|
void |
test_lowLevelStatisticsArrayMethods()
|
void |
test_modulo()
|
void |
test_normalize()
|
void |
test_statisticsFunctions()
|
void |
test_subtractParallelComponent()
|
Methods inherited from class java.lang.Object |
---|
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait |
Constructor Detail |
---|
public Math2.UnitTest()
Method Detail |
---|
public void test_infinityBehavior()
public void test_modulo()
public void test_logsPowersEtc()
public void test_statisticsFunctions()
public void test_gaussianFit() throws Exception
Exception
public void test_gaussianAndersonDarling()
andersonDarling1 (Gaussian sample) = 0.13459667082165735, andersonDarling2 (uniform distribution sample) = 11170.383267133551
andersonDarling1 (Gaussian sample) = 0.18940867401726738, andersonDarling2 (uniform distribution sample) = 11064.463336125908
andersonDarling1 (Gaussian sample) = 0.3984548680792433, andersonDarling2 (uniform distribution sample) = 11038.706942927185
andersonDarling1 (Gaussian sample) = 0.41313940626760093, andersonDarling2 (uniform distribution sample) = 11217.005208263494
andersonDarling1 (Gaussian sample) = 0.27588003071356976, andersonDarling2 (uniform distribution sample) = 11115.653184588584
The average value above for andersonDarling1 of ~0.25 compares well with the value 0.2576117
that is reported in the "anderson darling normal test y1" section of
this reference.
public void test_gaussianKolmogorovSmirnov()
kolmogorovSmirnov1 (Gaussian sample) = 0.5506861389029671, kolmogorovSmirnov2 (uniform distribution sample) = 57.55263461482368
kolmogorovSmirnov1 (Gaussian sample) = 0.7466358894082092, kolmogorovSmirnov2 (uniform distribution sample) = 57.34064194461067
kolmogorovSmirnov1 (Gaussian sample) = 0.7624236873765011, kolmogorovSmirnov2 (uniform distribution sample) = 57.16759617069217
kolmogorovSmirnov1 (Gaussian sample) = 0.836100754503244, kolmogorovSmirnov2 (uniform distribution sample) = 57.28774638323175
kolmogorovSmirnov1 (Gaussian sample) = 0.6035095000266333, kolmogorovSmirnov2 (uniform distribution sample) = 57.28436577988172
The average value above for kolmogorovSmirnov1 of ~0.65 is well under the critical value (so this test correctly accepts 1 as a Gaussian).
The average value above for kolmogorovSmirnov2 of ~57 is way over the critical value (so this test correctly rejects 2 as a Gaussian).
public void benchmark_statisticsFunctions()
n = 1000
gaussianCdf: first = 144.364 ns, mean = 8.418 ns (CI deltas: -1.923 ps, +2.269 ps), sd = 94.479 ns (CI deltas: -15.358 ns, +22.652 ns) WARNING: execution times have mild outliers, SD VALUES MAY BE INACCURATE
public void test_infoTheory()
public void test_bitTwiddling()
public void test_normalize()
public void test_subtractParallelComponent()
public void test_linearLeastSquaresFit()
public void test_autocorrelationStatisticsMethods()
public void test_lowLevelStatisticsArrayMethods()
|
||||||||||
PREV CLASS NEXT CLASS | FRAMES NO FRAMES | |||||||||
SUMMARY: NESTED | FIELD | CONSTR | METHOD | DETAIL: FIELD | CONSTR | METHOD |