feature/cavity_noise_simulation (additive noise implementation) by ehtkarim · Pull Request #210 · project8/katydid

ehtkarim · 2025-05-27T23:15:55Z

The processor "gaussian-noise-generator" (from KTGaussianNoiseGenerator.cc) now correctly adds White Gaussian Noise to the signal. I've added an example config file (named KatydidPSConfig_noise.yaml) demonstrating the processor usage. One can also use a random number seed using the "seed" parameter in the config file

… a working seed for random noise generation

…pectrum issue in PS)

…ator

…nal noise handling

…d corrected cavity noise model power from the PSD

…enerator' processor. Updated 'CMakeLists.txt'

ehtkarim · 2025-10-01T18:02:19Z

KTGaussianNoiseGenerator
- Has a working random seed feature now
- Can configure noise-psd-floor values. Defaults to 2.2e-13 W/Hz (default CCA scenario)
KTCavityNoiseGenerator
- Inherited from KTGaussianNoiseGenerator
- Has a working random seed feature now
- Implements Wouter/Arina's frequency dependent cavity noise model
- Can configure model parameters in the Katydid config file
KTMeasuredNoiseGenerator
- Inherited from KTGaussianNoiseGenerator
- Takes a .yaml file with psd-mean and psd-variance values at different frequencies as input, interpolates them using a 3rd order spline, and implements the psd-mean and psd-variance values per frequency bin to generate noise
I'm attaching some freq. and time slices with each of the three noises applied to a nominal 90 degree Locust_MC signal:

Freq_Spectrum_PSD_Mean_1e-19_Variance_2e-36

Time_Series_PSD_Mean_1e-19_Variance_2e-36

nsoblath · 2025-10-30T14:39:31Z

Examples/ConfigFiles/KatydidPSConfig_noise.yaml

Please rename this to KatydidPSNoiseConfig.yaml to stick with CamelCase and have the "noise" modifier before "Config"

nsoblath · 2025-10-30T14:41:03Z

Examples/ConfigFiles/KatydidPSConfig_noise_cavity.yaml

Please rename this to KatydidPSNoiseCavityConfig.yaml to stick with CamelCase and have the "NoiseCavity" modifier before "Config"

nsoblath · 2025-10-30T14:51:24Z

Source/Simulation/KTCavityNoiseGenerator.hh

+            virtual bool ConfigureDerivedGenerator(const scarab::param_node* node);
+
+        protected:
+            struct ModelPars


Why are the model parameters contained within a struct? I don't see any particular design reason for it. They should just be member variables of the processor.

I suggest making them all member variables in the class using the MEMVAR macros. Try to stick with the CamelCase naming convention.

nsoblath · 2025-10-30T15:02:39Z

Source/Simulation/KTCavityNoiseGenerator.cc

+        const double g   = fPars.Q0 / fPars.Q_L;
+        const double lor = 1.0 / ( 1.0 + std::pow( 2.0 * fPars.Q_L * (f - fPars.f0) / fPars.f0, 2.0 ) );    // Lorentzian
+
+        const double kB  = 1.380649e-23;


These constants should be added to KTMath (in the Utility directory). Please add them there.

Also, while you're editing those files, can you please make the functions that return numeric constants constexpr? Originally the minimum version of C++ I used didn't have that when I first started Katydid, but we require C++17, so we an use it.

nsoblath · 2025-10-30T15:08:18Z

Source/Simulation/KTCavityNoiseGenerator.cc

+
+    double KTCavityNoiseGenerator::NoisePSD(double f) const
+    {
+        const double g   = fPars.Q0 / fPars.Q_L;


I suggest you add g as a private member variable (no setter). And then you calculate it in the setters for Q0 and Q_L. That way you're not calculating it every time NoisePSD is calculated.

nsoblath · 2025-10-30T15:09:18Z

Source/Simulation/KTCavityNoiseGenerator.cc

+
+        const double P_cav  = kB*Tcav  * (4.*g/std::pow(1.+g,2)) * lor;
+        const double P_loss = kB*( fPars.A*fPars.T_line_start + (1.-fPars.A)*fPars.T_line_end );
+        const double P_isol = kB*Tisol * (1. - (4.*g/std::pow(1.+g,2))*lor);


pow() is quite slow. For a simple thing like this one I suggest you just use (1.+g)*(1.+g).

nsoblath · 2025-10-30T15:14:53Z

Source/Simulation/KTCavityNoiseGenerator.cc

+        const double hbar= 1.054571817e-34;
+        const double omega = 2.0 * M_PI * f;
+
+        auto eta = [](double x){ return x/(std::exp(x)-1.0) + 0.5; };   // Bose-Einstein factor


I suggest implementing this as a member function. That way you can have unit tests for it, and you won't have memory being allocated for the lambda at runtime.

nsoblath · 2026-01-30T18:16:51Z

Examples/ConfigFiles/KatydidPSConfig_noise.yaml

+      - egg
+
+egg:
+    filename: "locust_mc_Seed600_Angle90.000000000_Pos0.0040000_Energy18563.251000000.egg"


Since this is an example, please give a generic filename.

nsoblath · 2026-01-30T18:17:13Z

Examples/ConfigFiles/KatydidPSConfig_noise_cavity.yaml

+      - egg
+
+egg:
+    filename: "locust_mc_Seed600_Angle90.000000000_Pos0.0040000_Energy18563.251000000.egg"


Please give a generic filename

nsoblath · 2026-01-30T18:24:35Z

Source/Simulation/KTMeasuredNoiseGenerator.hh

+
+        protected:
+            // Simple natural cubic spline for doubles
+            class KCubicSpline


Instead of re-implementing a spline here, I suggest using Utility/KTSpline

Actually, you could take your implementation and replace the use of ROOT's TSpline3. I've always been bothered by requiring ROOT for the use of KTSpline, but never enough to fix it.

juniorpena · 2026-02-12T22:22:03Z

Source/Simulation/KTGaussianNoiseGenerator.cc

+        }
+        else if (node->has("noise-temperature"))
+        {
+            static constexpr double kBoltzmann = 1.38064852e-23; // J/K - Boltzmann constant; there's none defined in Source/Utility, so keeping a local constexpr


Could add kB to KTMath in Utility as Noah commented above

ehtkarim added 7 commits April 23, 2025 10:46

Initial commmit for implementing working aWGN in Katydid, need to set…

81000d6

… a working seed for random noise generation

Added a fix for handling noise addition properly (resolves mirrored s…

22c4938

…pectrum issue in PS)

Added a way to incorporate seed for the RNG()

a12bb55

Added an example config file for applying aWGN

de31c1e

Minor fix in KTRandom.hh

8b0b99d

Removed commits and made minor changes before merging to /develop

8265bdd

Minor indentation fix

33a5215

ehtkarim requested a review from nsoblath May 27, 2025 23:15

ehtkarim assigned ehtkarim and unassigned ehtkarim May 27, 2025

ehtkarim requested a review from psurukuc May 27, 2025 23:21

ehtkarim added 3 commits May 29, 2025 12:57

Initial implementation of cavity noise model under KTCavityNoiseGener…

ac7f484

…ator

Fixed cavity parameters reading-in, f_lo implementation & complex sig…

8b0f5a3

…nal noise handling

Added feature to set aWGN by noise-floor-psd or noise-temperature, an…

9be0933

…d corrected cavity noise model power from the PSD

ehtkarim force-pushed the feature/cavity_noise_simulation branch from 94b00d2 to 9be0933 Compare September 17, 2025 16:23

Added documentation for new processors, and created 'KTMeasuredNoiseG…

62dcc25

…enerator' processor. Updated 'CMakeLists.txt'

ehtkarim changed the title ~~feature/cavity_noise_simulation (aWGN implementation)~~ feature/cavity_noise_simulation (additive noise implementation) Oct 10, 2025

Correcting power normalization KTMeasuredNoiseGenerator.cc

4a93ba5

nsoblath requested changes Jan 30, 2026

View reviewed changes

juniorpena reviewed Feb 12, 2026

View reviewed changes

Conversation

ehtkarim commented May 27, 2025

Uh oh!

ehtkarim commented Oct 1, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants