| 1 | import math |
|---|
| 2 | # arrays to fill |
|---|
| 3 | resx=numpy.zeros(200) |
|---|
| 4 | resy=numpy.zeros(200) |
|---|
| 5 | rese=numpy.zeros(200) |
|---|
| 6 | resChi=numpy.zeros(200) |
|---|
| 7 | res0=numpy.zeros(200) |
|---|
| 8 | res1=numpy.ones(200) |
|---|
| 9 | resNZ=numpy.zeros(200) |
|---|
| 10 | resIC=numpy.zeros(200) |
|---|
| 11 | resICE=numpy.zeros(200) |
|---|
| 12 | a=0.25 # asymmetry to simulate |
|---|
| 13 | NB=1000 # number of (raw) bins |
|---|
| 14 | x12arr=numpy.zeros(2*NB) |
|---|
| 15 | y12arr=numpy.zeros(2*NB) |
|---|
| 16 | e12arr=numpy.zeros(2*NB) |
|---|
| 17 | ws=CreateWorkspace(x12arr,y12arr,e12arr,2,OutputWorkspace="Hello") |
|---|
| 18 | for x in range(200): |
|---|
| 19 | lam=math.exp((x-75.0)/10.0) # counts per bin, in absence of any signal. Log scale to show detail. |
|---|
| 20 | lam1=lam*(1+a) |
|---|
| 21 | lam2=lam*(1-a) # counts per bin in forward and backward banks |
|---|
| 22 | Xarr=range(NB) # "time" |
|---|
| 23 | Y1arr=numpy.random.poisson(lam1,NB) # measured forward counts |
|---|
| 24 | E1arr=numpy.sqrt(Y1arr) # and their errors by the "standard" formula, Fit() will treat error=0 specially |
|---|
| 25 | Y2arr=numpy.random.poisson(lam2,NB) |
|---|
| 26 | E2arr=numpy.sqrt(Y2arr) |
|---|
| 27 | ws.dataX(0)[:]=Xarr |
|---|
| 28 | ws.dataY(0)[:]=Y1arr |
|---|
| 29 | ws.dataE(0)[:]=E1arr |
|---|
| 30 | ws.dataX(1)[:]=Xarr |
|---|
| 31 | ws.dataY(1)[:]=Y2arr |
|---|
| 32 | ws.dataE(1)[:]=E2arr |
|---|
| 33 | AsymmetryCalc(InputWorkspace="Hello",OutputWorkspace="Asym",ForwardSpectra="0",BackwardSpectra="1",alpha=1.0) # re-generated asymmetry |
|---|
| 34 | (stat,chisq,Covar,params,curves)=Fit(Function="name=FlatBackground,A0=0.1",InputWorkspace="Asym",Output="Asym") |
|---|
| 35 | print lam," -> ",params.column(1)[0]," +- ",params.column(2)[0]," chisq=",chisq," st=",stat |
|---|
| 36 | resx[x]=lam |
|---|
| 37 | resy[x]=params.column(1)[0] |
|---|
| 38 | rese[x]=params.column(2)[0] |
|---|
| 39 | resNZ[x]=(len(Y1arr)-numpy.count_nonzero(Y1arr)+len(Y2arr)-numpy.count_nonzero(Y2arr))/(len(Y1arr)+len(Y2arr)+0.0) |
|---|
| 40 | resChi[x]=chisq |
|---|
| 41 | DeleteWorkspace("Asym") |
|---|
| 42 | |
|---|
| 43 | YS1=numpy.sum(Y1arr,dtype=numpy.float) # integral asymmetry of same data for comparison |
|---|
| 44 | YS2=numpy.sum(Y2arr,dtype=numpy.float) |
|---|
| 45 | if(YS1+YS2>0): |
|---|
| 46 | resIC[x]=(YS1-YS2)/(YS1+YS2) |
|---|
| 47 | resICE[x]=2.0*math.sqrt(YS1*YS2)*(YS1+YS2)**(-1.5) |
|---|
| 48 | else: |
|---|
| 49 | resIC[x]=float('NaN') |
|---|
| 50 | resICE[x]=float('Inf') # no counts, asymmetry completely uncertain! |
|---|
| 51 | |
|---|
| 52 | DeleteWorkspace("Hello") |
|---|
| 53 | CreateWorkspace(resx,resy,rese,1,OutputWorkspace="Summary") # what the fit thought the asymmetry was |
|---|
| 54 | CreateWorkspace(resx,resNZ,res0,1,OutputWorkspace="FractionOfZeros") |
|---|
| 55 | DiffOverErr=(resy-a)/rese |
|---|
| 56 | CreateWorkspace(resx,DiffOverErr,res1,1,OutputWorkspace="NormalisedDifference") # if outside +-1 then systematic errors are significant |
|---|
| 57 | CreateWorkspace(resx,resChi,res0,1,OutputWorkspace="ChiSquared") |
|---|
| 58 | |
|---|
| 59 | CreateWorkspace(resx,resIC,resICE,1,OutputWorkspace="IntegralCounted") |
|---|
| 60 | |
|---|
| 61 | p=plotSpectrum("Summary",0,True) |
|---|
| 62 | l=p.activeLayer() |
|---|
| 63 | l.setAxisScale(Layer.Bottom,0.001,1000000.0,Layer.Log10) |
|---|
| 64 | |
|---|
| 65 | p2=plotSpectrum("NormalisedDifference",0,True) |
|---|
| 66 | l2=p2.activeLayer() |
|---|
| 67 | l2.setAxisScale(Layer.Bottom,0.001,1000000.0,Layer.Log10) |
|---|
