
A. Using mean function: 0.5*t^2 + 1.5*sin(10*t) + cos(20*t) on [0,1] (features a valley)

	Using standard Trunc method, the bias for the first 5 eigenvalues are
	
	   -0.0790   -0.0250   -0.0125   -0.0201   -0.0229

	and the corresponding std dev/sqrt(1000) are
	
	    0.0074    0.0024    0.0014    0.0009    0.0005


	Similarly, when using true mean function, the bias are
	
	    0.0754    0.0235    0.0201    0.0011   -0.0086

	and the std dev/sqrt(1000) are
   
	    0.0081    0.0024    0.0014    0.0009    0.0006


A. Using mean function: 0.5*t^2 + -0.4*log(1+t) on [0,1] (mild)

	Using standard Trunc method, the bias for the first 5 eigenvalues are
	
	   -0.0340   -0.0159   -0.0117   -0.0200   -0.0228

	and the corresponding std dev/sqrt(1000) are
	
	    0.0076    0.0025    0.0014    0.0009    0.0005


	Similarly, when using true mean function, the bias are
	
	    0.0879    0.0266    0.0180   -0.0007   -0.0093

	and the std dev/sqrt(1000) are
   
	    0.0080    0.0025    0.0013    0.0009    0.0006
