def params_flat_to_func (self, flat):
"""update function parameters from a flat list"""
# first split according to functions
nshared = [sum(f.varshare) for f in self.functions]
ntotal = [len(f.params) for f in self.functions]
ndatasets = len(self.graph.fit_datasets())
fsizes = [s+(t-s)*ndatasets for s,t in zip(nshared,ntotal)]
fparams = splitlist (flat, fsizes)
# then split by variable
for fn, fparam in zip (self.functions, fparams):
dparams = []
vsizes = map (lambda x: x and 1 or ndatasets, fn.varshare)
vparams = splitlist (fparam, vsizes)
for d in range(ndatasets):
dp = []
for j, vparam in enumerate(vparams):
if fn.varshare [j]:
dp.append (vparam[0])
else:
try:
dp.append (vparam[d])
except:
print >>sys.stderr, d
dparams.append (dp)
for dparam, ds in zip(dparams, self.graph.fit_datasets()):
fn.load(ds.curveid)
fn.set_params (dparam)
fn.save(ds.curveid)
def params_flat_to_func(self, flat):
"""update function parameters from a flat list"""
# first split according to functions
nshared = [sum(f.varshare) for f in self.functions]
ntotal = [len(f.params) for f in self.functions]
ndatasets = len(self.graph.fit_datasets())
fsizes = [s + (t - s) * ndatasets for s, t in zip(nshared, ntotal)]
fparams = splitlist(flat, fsizes)
# then split by variable
for fn, fparam in zip(self.functions, fparams):
dparams = []
vsizes = map(lambda x: x and 1 or ndatasets, fn.varshare)
vparams = splitlist(fparam, vsizes)
for d in range(ndatasets):
dp = []
for j, vparam in enumerate(vparams):
if fn.varshare[j]:
dp.append(vparam[0])
else:
try:
dp.append(vparam[d])
except:
print >> sys.stderr, d
dparams.append(dp)
for dparam, ds in zip(dparams, self.graph.fit_datasets()):
fn.load(ds.curveid)
fn.set_params(dparam)
fn.save(ds.curveid)
def fitfunction (*args):
if len(args) == 3:
niter, actred, wss = args
project.mainwin.statusBar().message ('Fitting: Iteration %d, xsqr=%g, reduced by %g' % (niter, wss, actred))
project.mainwin.progressbar.setProgress(niter*10, 0)
project.main_dict['app'].processEvents()
return
else:
params, x = args
fitdatasets = FitWindow.inst.graph.fit_datasets()
fsdizes = [len(d.x()) for d in fitdatasets]
xreal = map (Numeric.array, splitlist (x, fsdizes))
y = []
FitWindow.inst.params_flat_to_func (params)
for i, xx in enumerate(xreal):
for fn in FitWindow.inst.functions:
fn.load (fitdatasets[i].curveid)
y.append (FitWindow.inst.call(xx))
ret = concatenate (y)
if FitWindow.inst.graph.fitwin.wmethod == 2:
try:
return Numeric.log(ret) # logarithmic fit
except:
print >>sys.stderr, 'Math range error (den Vogel)'
return zeros(ret.shape, 'f')
else:
return ret # linear fit
def fitfunction(*args):
if len(args) == 3:
niter, actred, wss = args
project.mainwin.statusBar().message(
'Fitting: Iteration %d, xsqr=%g, reduced by %g' %
(niter, wss, actred))
project.mainwin.progressbar.setProgress(niter * 10, 0)
project.main_dict['app'].processEvents()
return
else:
params, x = args
fitdatasets = FitWindow.inst.graph.fit_datasets()
fsdizes = [len(d.x()) for d in fitdatasets]
xreal = map(Numeric.array, splitlist(x, fsdizes))
y = []
FitWindow.inst.params_flat_to_func(params)
for i, xx in enumerate(xreal):
for fn in FitWindow.inst.functions:
fn.load(fitdatasets[i].curveid)
y.append(FitWindow.inst.call(xx))
ret = concatenate(y)
if FitWindow.inst.graph.fitwin.wmethod == 2:
try:
return Numeric.log(ret) # logarithmic fit
except:
print >> sys.stderr, 'Math range error (den Vogel)'
return zeros(ret.shape, 'f')
else:
return ret # linear fit
