Computing gpr lagtime

[wrshoemaker]

Hello,

@jesszell and I are trying to compute lagtime for the gaussian process regression model using line 55 in halo_serial.py and I get the following error

Traceback (most recent call last):
File "gpr_test.py", line 120, in
get_params(file_name)
File "gpr_test.py", line 83, in get_params
print(lagTime.compute(gpFactory.buildInputFixed(time_min=0,time_max=max(edata.time),size=200,convert=False), gp1))
File "/Python/gp_growth/metric.py", line 148, in compute
ret = self._compute(predictive_data,model,**kwargs)
File "/Python/gp_growth/metric.py", line 271, in _compute
mu,var = gpDerivative(x,gp)
File "/Python/gp_growth/metric.py", line 14, in gpDerivative
if x.ndim == 1:
AttributeError: 'float' object has no attribute 'ndim'

I'm looking at the LagTime() class in metric.py. It looks like the _compute() function treats the object containing the optical density data as a float. I think this is because of how the GrowthMetric() class treats the LagTime() class, but I haven't dug into the GrowthMetric() code.

Do you have a quick fix to get the lag time estimates? I've been excited about getting maximum growth rate and carrying capacity estimates from a gaussian process regression model and would really like to have lag time estimates.

Also, I very much enjoyed the paper.