Python cmath.e方法代码示例

# 需要导入模块: import cmath [as 别名]
# 或者: from cmath import e [as 别名]
def test_constants(self):
        e_expected = 2.71828182845904523536
        pi_expected = 3.14159265358979323846
        self.assertAlmostEqual(cmath.pi, pi_expected, places=9,
            msg="cmath.pi is {}; should be {}".format(cmath.pi, pi_expected))
        self.assertAlmostEqual(cmath.e, e_expected, places=9,
            msg="cmath.e is {}; should be {}".format(cmath.e, e_expected)) 

# 需要导入模块: import cmath [as 别名]
# 或者: from cmath import e [as 别名]
def check_polar(self, func):
        def check(arg, expected):
            got = func(arg)
            for e, g in zip(expected, got):
                self.rAssertAlmostEqual(e, g)
        check(0, (0., 0.))
        check(1, (1., 0.))
        check(-1, (1., pi))
        check(1j, (1., pi / 2))
        check(-3j, (3., -pi / 2))
        inf = float('inf')
        check(complex(inf, 0), (inf, 0.))
        check(complex(-inf, 0), (inf, pi))
        check(complex(3, inf), (inf, pi / 2))
        check(complex(5, -inf), (inf, -pi / 2))
        check(complex(inf, inf), (inf, pi / 4))
        check(complex(inf, -inf), (inf, -pi / 4))
        check(complex(-inf, inf), (inf, 3 * pi / 4))
        check(complex(-inf, -inf), (inf, -3 * pi / 4))
        nan = float('nan')
        check(complex(nan, 0), (nan, nan))
        check(complex(0, nan), (nan, nan))
        check(complex(nan, nan), (nan, nan))
        check(complex(inf, nan), (inf, nan))
        check(complex(-inf, nan), (inf, nan))
        check(complex(nan, inf), (inf, nan))
        check(complex(nan, -inf), (inf, nan)) 

# 需要导入模块: import cmath [as 别名]
# 或者: from cmath import e [as 别名]
def computeBaseClassifierCoefficient(self, classifierIdx):
        '''
        输入当前正在训练的基分类器下标(从0开始)，计算当前的基分类器系数 alpha 。
        :param classifierIdx: 当前分类器下标(从0开始)
        :return:
        '''
        self.alphaList[classifierIdx] = (1.0 / 2 * \
                                         cmath.log((1.0-self.eList[classifierIdx])/self.eList[classifierIdx], cmath.e)\
                                         ).real 

# 需要导入模块: import cmath [as 别名]
# 或者: from cmath import e [as 别名]
def setPval(self,p): self.p = p ### Typically re-set when a moderated statistic is calculated (e.g., emperical Bayesian - eBayes) 

# 需要导入模块: import cmath [as 别名]
# 或者: from cmath import e [as 别名]
def moderateTestStats(pval_db,probability_statistic):
    """ Calculate a moderated variance for each biological comparison based, based on the average variance of all genes or molecules.
    This calculation should be identical for moderated student t-test p-values from the R package limma. Small variances might arrise
    from differences in the precision float values stored by the different languages and threshold from the Newton Iteration step. This
    implementation currently relies on first, second and third derivitive calculations (e.g., polygamma aka psi functions) from mpmath."""
    
    #tst = salstat_stats.TwoSampleTests([],[]) ### Create object with two empty lists - will analyze in object database afterwards
    #d0, s0_squared = tst.getModeratedStandardDeviation(pval_db)
    d0, s0_squared = getModeratedStandardDeviation(pval_db,probability_statistic)
    #print 'Prior degrees of freedom:',d0, 'and Prior s0 squared:',s0_squared
    #d0 = 2.054191
    #s0_squared = 0.01090202
    for uid in pval_db:
        gs = pval_db[uid]
        if 'Welch' in probability_statistic:
            ModeratedWelchTest(gs,d0, s0_squared)
        else:
            #tst.ModeratedTTestUnpaired(gs,d0, s0_squared)
            ModeratedTTestUnpaired(gs,d0,s0_squared)
        """
        if uid == '10367120':
            print gs.Avg1(), gs.Avg2(), gs.FeatureVariance(), math.sqrt(gs.FeatureVariance()), gs.AdjP()
            #gs.setFeatureVariance(math.sqrt(gs.FeatureVariance()))
            #tst.ModeratedTTestUnpaired(gs,d0, s0_squared)
            #print gs.Avg1(), gs.Avg2(), gs.FeatureVariance(), math.sqrt(gs.FeatureVariance()), gs.AdjP()
        """ 

# 需要导入模块: import cmath [as 别名]
# 或者: from cmath import e [as 别名]
def getModeratedStandardDeviation(comparison_db,probability_statistic):
    variance_ls=[]; e_sum=0; d0_2nd_moment_gene_sum = 0
    for uid in comparison_db:
        gs = comparison_db[uid] ### Object containing summary statistics needed for each uid (aka feature)
        if 'Welch' in probability_statistic:
            df = gs.DF()
        else:
            try: df = (gs.N1() + gs.N2()) - 2
            except Exception,e: print e, gs, [gs.N1(), gs.N2()];kill
        sg_squared = gs.FeatureVariance()
        #print uid, df, sg_squared;kill
        ###calculate s0 and d0
        if sg_squared > 1e-11:
            zg = math.log(sg_squared)
            eg = zg - psi(0,df/2.0) + math.log(df/2.0)
            variance_ls.append((eg,df))

    n = len(variance_ls) ### number of uids analyzed
    
    ### Get the mean eg for all IDs
    for (eg,df) in variance_ls:
        e_sum+=eg
        e_avg = e_sum/len(variance_ls)

    ### Calculate the d0 2nd derivitive that will later need to be solved for d0
    for (eg,df) in variance_ls:
        d0_2nd_moment_gene_sum += ((math.pow(eg-e_avg,2)*n)/(n-1)) - psi(1,df/2)
        
    d0_2nd_moment_solve = d0_2nd_moment_gene_sum/len(variance_ls)
    #print [d0_2nd_moment_solve]
    d0 = NewtonInteration(d0_2nd_moment_solve)*2
    #print [d0]
    d0 = float(d0)
    e = cm.e
    s0_squared = math.pow(e,e_avg+psi(0,d0/2) - math.log(d0/2))
    return d0, s0_squared 

# 需要导入模块: import cmath [as 别名]
# 或者: from cmath import e [as 别名]
def chisqprob(chisq,df):
    """
    Returns the (1-tailed) probability value associated with the provided
    chi-square value and df.  Adapted from chisq.c in Gary Perlman's |Stat.
    
    Usage:   chisqprob(chisq,df)
    """
    BIG = 20.0
    def ex(x):
	BIG = 20.0
	if x < -BIG:
	    return 0.0
	else:
	    return math.exp(x)

    if chisq <=0 or df < 1:
	return 1.0
    a = 0.5 * chisq
    if df%2 == 0:
	even = 1
    else:
	even = 0
    if df > 1:
	y = ex(-a)
    if even:
	s = y
    else:
	s = 2.0 * zprob(-math.sqrt(chisq))
    if (df > 2):
	chisq = 0.5 * (df - 1.0)
	if even:
	    z = 1.0
	else:
	    z = 0.5
	if a > BIG:
	    if even:
		e = 0.0
	    else:
		e = math.log(math.sqrt(math.pi))
	    c = math.log(a)
	    while (z <= chisq):
		e = math.log(z) + e
		s = s + ex(c*z-a-e)
		z = z + 1.0
	    return s
	else:
	    if even:
		e = 1.0
	    else:
		e = 1.0 / math.sqrt(math.pi) / math.sqrt(a)
		c = 0.0
		while (z <= chisq):
		    e = e * (a/float(z))
		    c = c + e
		    z = z + 1.0
		return (c*y+s)
    else:
	return s