Signal-to-noise statistic: Difference between revisions

From formulasearchengine
Jump to navigation Jump to search
en>Male1979
relating to classification accuracy
 
en>Yobot
m WP:CHECKWIKI error 61 fix, References after punctuation per WP:REFPUNC and WP:PAIC using AWB (8459)
 
Line 1: Line 1:
Hello and welcome. My title is Ling. One of the things I love most is greeting card gathering but I don't have the time lately. My house is now in Kansas. The occupation he's been occupying for many years is a messenger.<br><br>Here is my web-site; extended auto warranty ([http://Bulls4Bears.com/blog/2014/08/23/auto-repair-tips-youll-wish-youd-read-sooner/ click the next internet page])
In [[numerical analysis]] and [[computational fluid dynamics]], '''Godunov's theorem''' — also known as '''Godunov's order barrier theorem''' — is a mathematical [[theorem]] important in the development of the theory of [[high resolution scheme]]s for the numerical solution of [[partial differential equations]].
 
The theorem states that:
 
:''Linear numerical schemes for solving [[partial differential equations]] (PDE's), having the property of not generating new extrema ([[monotone scheme]]), can be at most first-order accurate.''
 
Professor [[Sergei K. Godunov]] originally proved the theorem as a Ph.D. student at [[Moscow State University]]. It is his most influential work in the area of applied and numerical mathematics and  has had a major impact on science and engineering, particularly in the development of methodologies used in [[computational fluid dynamics]] (CFD) and other computational fields. One of his major contributions was to prove the theorem (Godunov, 1954; Godunov, 1959), that bears his name.
 
==The theorem==
We generally follow Wesseling (2001).
 
'''Aside'''
 
Assume a continuum problem described by a [[partial differential equations|PDE]] is to be computed using a numerical scheme based upon a uniform computational grid and a one-step, constant step-size, ''M'' grid point, integration algorithm, either implicit or explicit. Then if <math> x_{j}  = j\,\Delta x \ </math>  and <math>t^{n}  = n\,\Delta t \ </math>, such a scheme can be described by
 
:<math>
\sum\limits_{m=1}^{M} {\beta _m } \varphi _{j + m}^{n + 1}  = \sum\limits_{m=1}^{M} {\alpha _m \varphi _{j + m}^n }.
\quad  \quad ( 1) </math>
 
In other words, the solution <math>\varphi _j^{n + 1} \ </math> at time <math>n + 1</math> and location <math>j</math> is a linear function of the solution at the previous time step <math>n</math>. We assume that <math>\beta _m \ </math> determines <math>\varphi _j^{n + 1} \ </math> uniquely. Now, since the above equation represents a linear relationship between <math> \varphi _j^{n } \ </math> and <math> \varphi _j^{n + 1} \ </math> we can perform a linear transformation to obtain the following equivalent form,
 
:<math>\varphi _j^{n + 1}  = \sum\limits_m^{M} {\gamma _m \varphi _{j + m}^n }. \quad  \quad ( 2)
</math>
 
'''Theorem 1:''' ''Monotonicity preserving''
 
The above scheme of equation (2) is monotonicity preserving if and only if
 
:<math>\gamma _m  \ge 0,\quad \forall m . \quad  \quad ( 3)</math>
 
''Proof'' - Godunov (1959)
 
'''Case 1: (sufficient condition)'''
 
Assume (3) applies and that <math>\varphi _j^n \ </math> is monotonically increasing with <math>j \ </math>.
 
Then, because <math>\varphi _j^n  \le \varphi _{j + 1}^n  \le \cdots  \le \varphi _{j + m}^n </math> it therefore follows that <math>\varphi _j^{n + 1}  \le \varphi _{j + 1}^{n + 1} \le \cdots  \le \varphi _{j + m}^{n + 1} \ </math> because
 
:<math>
\varphi _j^{n + 1}  - \varphi _{j - 1}^{n + 1}  = \sum\limits_m^{M} {\gamma _m \left( {\varphi _{j + m}^n  - \varphi _{j + m - 1}^n } \right)}  \ge 0 . \quad  \quad ( 4)</math>
 
This means that monotonicity is preserved for this case.
 
'''Case 2: (necessary condition)'''  
 
We prove the necessary condition by contradiction. Assume that <math>\gamma _p^{}  < 0 \ </math> for some <math>p \ </math> and choose the following monotonically increasing <math>\varphi_j^n \quad </math>,
 
:<math>\varphi _i^n  = 0, \quad i < k;\quad \varphi _i^n  = 1, \quad i \ge k . \quad  \quad ( 5) </math>
 
Then from equation (2) we get
 
:<math> \varphi _j^{n + 1}  - \varphi _{j-1}^{n+1}  = \sum\limits_m^M {\gamma _m } \left( {\varphi _{j + m}^{n}  - \varphi _{j + m - 1}^{n} } \right) = \left\{ {\begin{array}{*{20}c}
  {0,} & {\left[ {j + m \ne k} \right]}  \\
  {\gamma _m ,} & {\left[ {j + m = k} \right]}  \\
\end{array}} \right . \quad  \quad ( 6)</math>
 
Now choose <math> j=k-p \ </math>, to give
 
:<math>
\varphi _{k-p}^{n + 1}  - \varphi _{k-p-1}^{n + 1}  =  {\gamma _p \left( {\varphi _{k}^n  - \varphi _{k - 1}^n } \right)}  < 0  , \quad  \quad ( 7)</math>
 
which implies that <math>\varphi _j^{n + 1} \ </math> is '''NOT''' increasing, and we have a contradiction. Thus, monotonicity is '''NOT''' preserved for <math>\gamma _p  < 0 \ </math>, which completes the proof.
 
'''Theorem 2:''' ''Godunov’s Order Barrier Theorem''
 
Linear one-step second-order accurate numerical schemes for the convection equation
 
:<math> {{\partial \varphi } \over {\partial t}} + c{ { \partial \varphi } \over {\partial x}} = 0 , \quad t > 0, \quad x \in \mathbb{R} \quad  \quad (10)</math>
 
cannot be monotonicity preserving unless
:<math>\sigma  = \left| c \right|{{\Delta t} \over {\Delta x}} \in \mathbb{ N} , \quad  \quad (11)</math>
 
where <math> \sigma \ </math> is the signed [[Courant–Friedrichs–Lewy condition]] (CFL) number.
 
''Proof'' - Godunov (1959)
 
Assume a numerical scheme of the form described by equation (2) and choose
 
:<math>\varphi \left( {0,x} \right) = \left( {{x \over {\Delta x}} - {1 \over 2}} \right)^2  - {1 \over 4}, \quad \varphi _j^0  = \left( {j - {1 \over 2}} \right)^2  - {1 \over 4} . \quad  \quad (12)</math>
 
The exact solution is
:<math>
\varphi \left( {t,x} \right) = \left( {{{x - ct} \over {\Delta x}} - {1 \over 2}} \right)^2  - {1 \over 4} . \quad  \quad (13)
</math>
 
If we assume the scheme to be at least second-order accurate, it should produce the following solution exactly
 
:<math>
\varphi _j^1  = \left( {j - \sigma  - {1 \over 2}} \right)^2  - {1 \over 4}, \quad \varphi _j^0  = \left( {j - {1 \over 2}} \right)^2  - {1 \over 4}. \quad  \quad (14)
</math>
 
Substituting into equation (2) gives:
 
:<math>
\left( {j - \sigma  - {1 \over 2}} \right)^2  - {1 \over 4} = \sum\limits_m^{M} {\gamma _m \left\{ {\left( {j + m - {1 \over 2}} \right)^2  - {1 \over 4}} \right\}}. \quad  \quad (15)
</math>
 
Suppose that the scheme '''IS''' monotonicity preserving, then according to the theorem 1 above, <math>\gamma _m  \ge 0 \ </math>. 
 
Now, it is clear from equation (15) that
 
:<math> \left( {j - \sigma  - {1 \over 2}} \right)^2  - {1 \over 4} \ge 0, \quad \forall j . \quad  \quad (16)</math>
 
Assume <math>\sigma  > 0, \quad \sigma  \notin \mathbb{ N} \ </math> and choose <math>j \ </math> such that <math> j > \sigma  > \left( j - 1 \right) \ </math> . This implies that <math>\left( {j - \sigma } \right) > 0 \ </math> and <math>\left( {j - \sigma  - 1} \right) < 0 \ </math> .
 
It therefore follows that,
 
:<math>
\left( {j - \sigma  - {1 \over 2}} \right)^2  - {1 \over 4} = \left( j - \sigma \right) \left(j - \sigma - 1 \right) < 0, \quad  \quad (17) </math>
 
which contradicts equation (16) and completes the proof.
 
The exceptional situation whereby <math>\sigma  = \left| c \right|{{\Delta t} \over {\Delta x}} \in \mathbb{N} \ </math> is only of theoretical interest, since this cannot be realised with variable coefficients. Also, integer [[Courant–Friedrichs–Lewy condition|CFL]] numbers greater than unity would not be feasible for practical problems.
 
==See also==
*[[Finite volume method]]
*[[Flux limiter]]
*[[Total variation diminishing]]
 
==References==
*'''Godunov, Sergei K.''' (1954), ''Ph.D. Dissertation: Different Methods for Shock Waves'',  Moscow State University.
*'''Godunov, Sergei K.''' (1959), A Difference Scheme for Numerical Solution of Discontinuous Solution of Hydrodynamic Equations, ''Math. Sbornik, 47, 271-306'', translated US Joint Publ. Res. Service, JPRS 7226, 1969.
*'''Wesseling, Pieter ''' (2001), ''Principles of Computational Fluid Dynamics'', Springer-Verlag.
 
==Further reading==
*'''Hirsch, C.''' (1990), ''Numerical Computation of Internal and External Flows'', vol 2, Wiley.
*'''Laney, Culbert B.''' (1998), ''Computational Gas Dynamics'', Cambridge University Press.
*'''Toro, E. F.''' (1999), ''Riemann Solvers and Numerical Methods for Fluid Dynamics'', Springer-Verlag.
*'''Tannehill, John C., et al.,''' (1997), ''Computational Fluid mechanics and Heat Transfer'', 2nd Ed., Taylor and Francis.
 
[[Category:Numerical differential equations]]
[[Category:Theorems in analysis]]
[[Category:Computational fluid dynamics]]

Latest revision as of 12:11, 12 October 2012

In numerical analysis and computational fluid dynamics, Godunov's theorem — also known as Godunov's order barrier theorem — is a mathematical theorem important in the development of the theory of high resolution schemes for the numerical solution of partial differential equations.

The theorem states that:

Linear numerical schemes for solving partial differential equations (PDE's), having the property of not generating new extrema (monotone scheme), can be at most first-order accurate.

Professor Sergei K. Godunov originally proved the theorem as a Ph.D. student at Moscow State University. It is his most influential work in the area of applied and numerical mathematics and has had a major impact on science and engineering, particularly in the development of methodologies used in computational fluid dynamics (CFD) and other computational fields. One of his major contributions was to prove the theorem (Godunov, 1954; Godunov, 1959), that bears his name.

The theorem

We generally follow Wesseling (2001).

Aside

Assume a continuum problem described by a PDE is to be computed using a numerical scheme based upon a uniform computational grid and a one-step, constant step-size, M grid point, integration algorithm, either implicit or explicit. Then if and , such a scheme can be described by

In other words, the solution at time and location is a linear function of the solution at the previous time step . We assume that determines uniquely. Now, since the above equation represents a linear relationship between and we can perform a linear transformation to obtain the following equivalent form,

Theorem 1: Monotonicity preserving

The above scheme of equation (2) is monotonicity preserving if and only if

Proof - Godunov (1959)

Case 1: (sufficient condition)

Assume (3) applies and that is monotonically increasing with .

Then, because it therefore follows that because

This means that monotonicity is preserved for this case.

Case 2: (necessary condition)

We prove the necessary condition by contradiction. Assume that for some and choose the following monotonically increasing ,

Then from equation (2) we get

Now choose , to give


which implies that is NOT increasing, and we have a contradiction. Thus, monotonicity is NOT preserved for , which completes the proof.

Theorem 2: Godunov’s Order Barrier Theorem

Linear one-step second-order accurate numerical schemes for the convection equation

cannot be monotonicity preserving unless

where is the signed Courant–Friedrichs–Lewy condition (CFL) number.

Proof - Godunov (1959)

Assume a numerical scheme of the form described by equation (2) and choose

The exact solution is

If we assume the scheme to be at least second-order accurate, it should produce the following solution exactly

Substituting into equation (2) gives:

Suppose that the scheme IS monotonicity preserving, then according to the theorem 1 above, .

Now, it is clear from equation (15) that

Assume and choose such that . This implies that and .

It therefore follows that,

which contradicts equation (16) and completes the proof.

The exceptional situation whereby is only of theoretical interest, since this cannot be realised with variable coefficients. Also, integer CFL numbers greater than unity would not be feasible for practical problems.

See also

References

  • Godunov, Sergei K. (1954), Ph.D. Dissertation: Different Methods for Shock Waves, Moscow State University.
  • Godunov, Sergei K. (1959), A Difference Scheme for Numerical Solution of Discontinuous Solution of Hydrodynamic Equations, Math. Sbornik, 47, 271-306, translated US Joint Publ. Res. Service, JPRS 7226, 1969.
  • Wesseling, Pieter (2001), Principles of Computational Fluid Dynamics, Springer-Verlag.

Further reading

  • Hirsch, C. (1990), Numerical Computation of Internal and External Flows, vol 2, Wiley.
  • Laney, Culbert B. (1998), Computational Gas Dynamics, Cambridge University Press.
  • Toro, E. F. (1999), Riemann Solvers and Numerical Methods for Fluid Dynamics, Springer-Verlag.
  • Tannehill, John C., et al., (1997), Computational Fluid mechanics and Heat Transfer, 2nd Ed., Taylor and Francis.