tinyurl.com/IMNN-AICosmo
or even better...
pip install IMNN jupyter matplotlib
git clone https://github.com/tomcharnock/IMNN-LFI_Taskforce.git (branch AICosmo)
cd IMNN-LFI_Taskforce
jupyter notebook
... then you'll have the IMNN on your own machine - you lucky people
Tom Charnock
Institut d'Astrophysique de Paris
Guilhem Lavaux, Benjamin D. Wandelt
Notebook: presentations.charnock.fr/IMNN/AICosmo19
|
|
|
|
|
A neural network (obvs.)
Some simulations ${\bf d}^i$

Some derivatives of the simulations ${\bf d},_{\alpha}^i$
|
|
We work in the space of data ${\bf d}$ and model parameters $\boldsymbol{\theta}$
Posterior is a slice through this space at some given data
Simulate the data and accept simulations close to the true data
We pass all simulations through the network to get summaries.
We pass the real data through the network to get its summary.
Measure how far the summaries of the simulations are from the summary of the real data
Network becomes part of the data aquisition
If simulations are bad, the summaries are less informative

Fast convolutions on the sphere to learn real space kernels in the correct topology


Keep eyes peeled for application to Planck data!
We can massively compress our data to optimal summaries
Using LFI we don't have to worry as much about the simulations
Its exceptionally cheap!
We know physics - how much should we insert into our networks?
Up to what point can we trust regression to parameter values?
Convolutions for translationally invariant data
(images or local signals)
We can combine the IMNN with our known and loved summaries to extract just that little bit extra information



