3.1. Parameter continuation and bifurcation detection

In this tutorial, you will learn how to perform a 1D numerical parameter continuation in PyRates with automatic fold bifurcation detection. Furthermore, you will learn how to plot a simple bifurcation diagram. Throughout this example, we will use the quadratic integrate-and-fire population model [1], a detailed introduction of which is given in the model introductions example gallery. The dynamic equations of this model read the following:

\[ \begin{align}\begin{aligned}\tau \dot r &= \frac{\Delta}{\pi\tau} + 2 r v,\\\tau \dot v &= v^2 +\bar\eta + I(t) + J r \tau - (\pi r \tau)^2,\end{aligned}\end{align} \]

where \(r\) is the average firing rate and \(v\) is the average membrane potential of the QIF population. It is governed by 4 parameters:

  • \(\tau\) –> the population time constant

  • \(\bar \eta\) –> the mean of a Lorenzian distribution over the neural excitability in the population

  • \(\Delta\) –> the half-width at half maximum of the Lorenzian distribution over the neural excitability

  • \(J\) –> the strength of the recurrent coupling inside the population

In this tutorial, we will demonstrate how to (1) , (2) perform a simple 1D parameter continuation in \(\bar \eta\), and (3) plot the corresponding bifurcation diagram. The latter has also been done in [1], so you can compare the resulting plot with the results reported by Montbrió et al. For parts (2) and (3) of the tutorial, it is required that you have PyCoBi installed in the Python environment you are using.


import matplotlib.pyplot as plt
from pyrates import CircuitTemplate
from pycobi import ODESystem
import sys

path = sys.argv[-1]
auto_dir = path if type(path) is str and ".py" not in path else "~/PycharmProjects/auto-07p"

3.1.1. Part 1: Creating a PyCoBi Instance

In this first part, we will be concerned with how to create a model representation that is compatible with auto-07p, which is the software that is used for parameter continuations and bifurcation analysis in PyRates [2]. Step 1: Load the model into PyRates

As a first step, we have to load the model into PyRates. This is done the usual way. If you are not familiar with this, check out the example galleries for model definitions.

qif = CircuitTemplate.from_yaml("model_templates.neural_mass_models.qif.qif") Step 2: Generate the Fortran routines required by Auto-07p

In the next step, we will translate our model into a Fortran file containing all subroutines required by auto-07p. In short, auto-07p requires a fortran file with the model equations and initial values [2]. This will require using the Fortran backend of PyRates and turning the vectorize option off:

qif.get_run_func(func_name='qif_rhs', file_name='qif', step_size=1e-4, auto=True, backend='fortran', solver='scipy',
                 vectorize=False, float_precision='float64')

Calling the CircuitTemplate.get_run_func method with auto=True creates two files, which we will inspect below. The first file is a .f90 file containing all the Fortran subroutines required by auto-07p:

f = open('qif.f90', 'r')

The second file is a textfile containing all the auto-07p parameters that determine how it performs parameter continuations and automated bifurcation detection:

f = open('c.ivp', 'r')

The default parameters written out by PyRates allow to solve the initial value problem, i.e. perform simple numerical simulations via auto-07p. For a detailed explanation of these parameters, see the Auto-07p documentation. Step 3: Generate a PyCoBi instance

Now that the model equations are compiled, we can generate an instance of pycobi.ODESystem, a Python tool that provides and interface to auto-07p.

qif_auto = ODESystem(working_dir=None, auto_dir=auto_dir, init_cont=False)

Now, we can use all the tools provided by Auto-07p to investigate how the model reacts to changes in its parameterization.

3.1.2. Part 2: Performing Parameter Continuations

In this part, we will demonstrate how to perform simple 1D parameter continuations via the ODESystem.run() method. Step 1: Time continuation

In parameter continuations, it is required that you start continuing the parameters from an equilibrium or periodic orbit, i.e. that the solution to the initial value problem would be constant or periodic in time. To achieve this, you can either set the initial values and parameters of your system to a known solution in the model definition (e.g. the YAML template), or choose a model parameterization for which a finite solution exists and then calculate the solution of the model in time until it converges to an equilibrium (or periodic orbit). If you go with the latter, you can then start to perform parameter continuations using the values of the state variables at the end point of your solution in time. This can be simply achieved by a call to the CircuitTemplate.run() method, before calling the get_run_func() method. Then, PyRates will automatically use the values of the state variables from the last simulation step. Alternatively, you can perform simulations in time via PyCoBi as follows:

t_sols, t_cont = qif_auto.run(
    e='qif', c='ivp', name='time', DS=1e-4, DSMIN=1e-10, EPSL=1e-08, EPSU=1e-08, EPSS=1e-06,
    DSMAX=1e-2, NMX=1000, UZR={14: 4.0}, STOP={'UZ1'})

qif_auto.plot_continuation('PAR(14)', 'U(1)', cont='time')

In this function call, you see how the general interface of the ODESystem.run() method works. In every first call of this method, the name of a fortran equations files needs to be specified by the keyword argument e. The standard name of this function is rhs_func when the file was automatically generated via PyRates. Also, we declare a constants file via c='ivp'. This file has been automatically generated by PyRates as well and contains the constants that are required by auto-07p for time continuations. Both files can be expected in the build directory of PyRates (pyrates_build/qif in this case). All other arguments just overwrite important auto-07p constants that are also declared in the constants file c.ivp. For a detailed explanation of those constants, please have a look at the auto documentation. Here, we will provide a short explanation of what each specific parameter does in our context:

  • DS=1e-3 defines the initial step-size of the time continuation (in ms)

  • DSMIN=1e-4 defines the minimal step-size of the time continuation (in ms)

  • DSMAX=1.0 defines the maximal step-size of the time continuation (in ms)

  • NMX=10000 defines the maximum number of continuation steps to perform

  • UZR={14: 1000.0} tells auto-07p to create a user-specified marker when the parameter 14, which is the default parameter field in auto-07p in which time is stored, reaches a value of 1000.0 (ms)

  • STOP={'UZ1'} tells auto-07p to stop the continuation ones it hits the first user-specified marker

The output of this call to .run() is a tuple, with the following two entries:

  • dict –> A dictionary that contains a summary of the parameter, variables and other characteristics of each solution along the branch that has been generated by the continuation.

  • branch –> An auto-07p object that has been generated during the .run() call and is required for subsequent parameter continuations. Step 2: Continuation of \(\bar \eta\)

If you look at the auto-07p output in the terminal, you will see that the values for U(1) and U(2) converged to certain values. These two values represent the current values of our state variables \(r\) and \(v\). Thus, our model converged to an equilibrium and we are now save to perform the continuation in our parameter of interest: \(\bar \eta\). This follows a very similar syntax:

eta_sols, eta_cont = qif_auto.run(
    origin=t_cont, starting_point='UZ1', name='eta', bidirectional=True,
    ICP=4, RL0=-20.0, RL1=20.0, IPS=1, ILP=1, ISP=2, ISW=1, NTST=400,
    NCOL=4, IAD=3, IPLT=0, NBC=0, NINT=0, NMX=2000, NPR=10, MXBF=5, IID=2,
    ITMX=40, ITNW=40, NWTN=12, JAC=0, EPSL=1e-06, EPSU=1e-06, EPSS=1e-04,
    DS=1e-4, DSMIN=1e-8, DSMAX=5e-2, IADS=1, THL={}, THU={}, UZR={}, STOP={}

In this call, we specified the full set of auto-07p constants. Don’t worry, usually, you do not have to bother with most of them. It is common practice to specify most of them in constants files that you would refer to the same way as in the previous call to the .run() method, via c=name. In such a case, you would specify all auto-07p constants that do not change between calls to the .run() method in a file with the name c.name and only provide the constants that need to be altered between .run() calls directly to the .run() method.

While it is out of the scope of this tutorial to explain all auto-07p constants here. Instead, we will provide an intuitive explanation of the most important ones and refer to the auto documentation for the rest:

  • origin='t_cont' is a keyword argument specific to PyCoBi. It tells auto-07p from which branch of solutions to start the parameter continuation from. This needs to be specified for every call to the .run() method, except for the first (since their is no solutions branch at this point). Here, we specified the solution branch from our initial continuation in time.

  • starting_point='UZ1' is a keyword argument specific to PyCoBi. It tells auto-07p to start the continuation from the first user-specified marker of the provided origin

  • name='eta' is a keyword argument specific to PyCoBi. It tells PyCoBi to store the results of the continuation using this particular name. We can use this name for later continuations or for plotting, to indicate which continuation to start from or to plot the results of.

  • bidirectional=True is a keyword argument specific to PyCoBi. It tells PyCoBi to change \(\bar\eta\) both in the positive and the negative direction.

  • ICP=4 tells auto-07p to perform a 1D continuation over parameter number 4 (you can check in the fortran file rhs_func.f that \(\bar\eta\) is indeed the 4th parameter)

  • RL0=-20.0 and RL1=20.0 specify the boundaries of the parameter continuation. If \(\bar\eta\) is continued beyond any of these borders, auto-07p stops the parameter continuation.

  • IPS=1 indicates auto-07p that it is supposed to continue an equilibrium of an ODE system

  • ILP=1 turns on the detection of fold bifurcations in auto-07p

  • ISP=2 turns on full automatic bifurcation detection in auto-07p

Checking the terminal output of auto-07p, you will realize that the output in column TY shows LP for two of the solutions we computed along our branch in \(\bar\eta\). These indicate the detection of limit point or fold bifurcations. We can visualize the full bifurcation diagram via the following call:

qif_auto.plot_continuation('PAR(4)', 'U(1)', cont='eta')

The curve in this plot represents the value of \(r\) (y-axis) at the equilibrium solutions that exist for each value of \(\bar\eta\) (x-axis). A solid line indicates that the equilibrium is stable, whereas a dotted line indicated that the equilibrium is unstable. The triangles mark the points at which auto-07p detected fold bifurcations. At a fold bifurcation, the critical eigenvalue of the vector field defined by the right-hand sides of our model’s ODEs crosses the imaginary axis (i.e. its real part changes the sign). This indicates a change of stability of an equilibrium solution, which happens because an unstable and a stable equilibiurm approach and annihilate each other. This behavior can be read from the plot as well. The solid and the dotted line approach each other towards the fold bifurcation marks. After they meet at the fold bifurcation, both cease to exist. Final step: Clean up all temporary files

As a last step, it is good practice to clean up all temporary files created by PyCoBi and PyRates. This can be achieved with the following simple call:


Gallery generated by Sphinx-Gallery