|
|
| Cpe (std::string paramStr, size_t count=10, bool defaultToRange=false) |
|
| Cpe (fvalue q, fvalue alpha) |
| virtual std::complex< fvalue > | execute (fvalue omega) override |
| virtual size_t | paramCount () const override |
| virtual char | getComponantChar () const override |
| virtual std::string | componantName () const override |
| virtual std::vector< eis::Range > | getDefaultParameters (bool range=true) const override |
| virtual std::string | getCode (std::vector< std::string > ¶meters) override |
| virtual std::string | getTorchScript (std::vector< std::string > ¶meters) override |
|
virtual void | setParamRanges (const std::vector< eis::Range > &ranges) |
|
virtual std::vector< eis::Range > & | getParamRanges () |
|
virtual std::vector< eis::Range > | getParamRanges () const |
| virtual std::string | getComponantString (bool currentValue=true, bool withCount=false) const |
| | Gets this componant as a model string.
|
|
virtual bool | compileable () |
|
std::string | getUniqueName () |
| bool | hasDefaultedParameters () const |
| | Checks if this componant had to use default parameters.
|
Constant phase element, model character p.
Z(omega) = 1 / (Q (j omega)^alpha)
= cos(alpha pi/2) / (Q omega^alpha) - j sin(alpha pi/2) / (Q omega^alpha)
|Z(omega)| = 1 / (Q omega^alpha)
phase = -alpha pi/2 constant over frequency
Parameters: p{Q,alpha}
- Q : admittance coefficient (often Y0 or T) in S s^alpha
- alpha : exponent, dimensionless, conventionally 0 <= alpha <= 1
The general element of which the others are special cases. Substituting an exponent turns p into a different element exactly:
alpha = 1 capacitor C = Q
alpha = 0.5
Warburg A = 1 / (Q sqrt(2))
alpha = 0 resistor R = 1 / Q
alpha = -1 inductor L = 1 / Q (outside the usual range)
Warburg element (semi-infinite linear diffusion), model character w.
Definition warburg.h:57
Because of these identities a p whose exponent sits on one of those values is interchangeable with the corresponding element; eis::Model::getEquivalentParameterPermutations() finds them by fitting rather than by comparing model characters.
Numerically the real and imaginary parts are computed as separate cos and sin terms, so at alpha = 1 the real part is not exactly zero but about 6e-17 / (Q omega), the rounding of cos(pi/2). Harmless for spectra, but do not test Re(Z) == 0 against a p.
alpha is the least sensitive parameter in a model string and Q the most, which matters when budgeting sweep steps; see the find-range mode.