cmf::math Namespace Reference

Contains classes for numerical solving of ODE's. More...

Detailed Description

Contains classes for numerical solving of ODE's.

Classes
class	BDF2
	An order 2 BDF-Method with fixed-point iteration and variable step size. More...
class	cubicspline
	Interpolates points with a cubic spline interpolation. More...
class	CVodeAdams
	Explizit multistep solver using CVode. More...
class	CVodeBanded
	implicit BDF CVode solver with a banded Jacobian approximation More...
class	CVodeBase
	Abstract base class for different modes of the CVode solver. More...
class	CVodeDense
	implicit BDF CVode solver with full Jacobian approximation More...
class	CVodeDiag
	implicit BDF CVode solver with a one line diagonal Jacobian approximation More...
struct	CVodeInfo
	Reports the current state of a CVode solver. More...
class	CVodeKrylov
	implicit BDF CVode solver with a Krylov preconditioner More...
struct	CVodeOptions
	A set of options for all CVode3 solver. More...
struct	Date
	An absolute time, not for calculation. Date and Time are interchangable. More...
class	ExplicitEuler_fixed
	An explicit Euler integrator, with a fixed time step. More...
class	HeunIntegrator
	A simple predictor - corrector solver. More...
class	ImplicitEuler
	An implicit (backward) Euler integrator using fixpoint iteration. More...
class	Integrator
	Base class for any kind of integrator. More...
class	MultiIntegrator
	The MultiIntegrator is a wrapper for a bunch integrators. The states of the integrators should not have direct connections over integrator boundaries. More...
class	num_array
	A valarray kind of vector implementation with OpenMP capabilities. More...
class	RKFIntegrator
	Integrates a vector of cmf::math::StateVariable with the Runge-Kutta-Fehlberg (RKF54) method. More...
class	SoluteWaterIntegrator
	A SoluteWaterIntegrator implements the cmf::math::Integrator interface, but consists of two independent ODE-solvers. More...
class	StateVariable
	Abstract class state variable. More...
class	Time
	A time class, used to pass around current modelling times. More...
class	timeseries
	A timeseries is a list of values, equally distributed over time. More...

Functions
int	get_parallel_threads ()
	Returns the max number of threads used by OpenMP in parallel sections of the code.
double	nash_sutcliffe (const cmf::math::timeseries &model, const cmf::math::timeseries &observation)
	Calculates the Nash-Sutcliffe efficiency of a modeled timeseries in comparison with an observed timeseries.
int	set_parallel_threads (int numthreads)
	Set the number of threads used by OpenMP in parallel sections of the code.

Variables
const Time	day = 24 * h
	1 day
const Time	h = 60 * min
	1 hour
const Time	min = 60 * sec
	1 minute
const Time	month = year/12
	1 month = 30*day+10*h
const Time	ms = timespan(1)
	1 milli second
const Time	never =Time(0,0,-9999)
	A date long before our time to indicate 'never'.
const Time	sec = 1000 * ms
	1 second
const Time	week = 7 * day
	1 week
const Time	year = 365 * day
	1 year = 365*day

Function Documentation

nash_sutcliffe()

double nash_sutcliffe	(	const cmf::math::timeseries &	model,
		const cmf::math::timeseries &	observation )

Calculates the Nash-Sutcliffe efficiency of a modeled timeseries in comparison with an observed timeseries.

The Nash-Sutcliffe efficiancy is defined as:

\[ E = 1 - \frac{\sum_{t=1}^{T}(M_t - O_t)^2}{\sum_{t=1}^{T}(O_t - \overline{O})^2}\]

where

• \(T\) is the number of observation time steps
• \(M\) is the timeseries of model results matchinig O
• \(O\) is the timeseries containing observations
• \(\overline{O}\) is the arithmetic mean of observations