pints
pints copied to clipboard
pints-team
Probabilistic Inference on Noisy Time Series
I have now worked out a numerically stable sensitivity for this case. See [here](https://mathematica.stackexchange.com/questions/191822/derivative-of-integrated-noise-gaussian-likelihood/193066#193066) and so makes sense to add this.
Add a sentence in doc saying what PINTS expect the model to return if it cannot solve the given set of parameters. E.g. it is expecting None or np.NaN or...
The pints.vector(array) command copies the input array with dtype=float, which casts the imaginary components from a complex input to real.
This is necessary for #282. There will be some priors for which this is not possible, in which case NotImplementedError should be returned.
It would be much easier for specifying Gaussian priors on lots of parameters (as is common).
[This](https://arxiv.org/abs/1903.08008) paper has all that is needed and [this](https://avehtari.github.io/rhat_ess/rhat_ess.html) repo contains R code that could be followed.
[This](https://arxiv.org/abs/1903.08008) paper has all that is needed and [this](https://avehtari.github.io/rhat_ess/rhat_ess.html) repo contains R code that could be followed.
E.g. - ProblemErrorMeasure.problem() - SingleOutputProblem.model() - MultiOutputProblem.model() Not needed from point of view of evaluating / simulating, but very convenient when e.g. passing a problem to a plotting method if...
``` sys:1: ResourceWarning: unclosed file ``` See very bottom of logs at e.g. https://ci.appveyor.com/project/martinjrobins/pints/builds/26014056/job/hdhmjs2jkj39tax2
Should either document this, or maybe add methods `set_initial_conditions` and `initial_conditions` to `ToyODEModel` class and use `initial_conditions()` inside `simulate` instead. Might also want to think about the case where we...