Update PEtab requirement to 5.4.1 in /test

[dependabot]

Updates the requirements on PEtab to permit the latest version.

Release notes

Sourced from PEtab's releases.

v5.4.1

PEtab v5.4.1

Diff since v5.4.0

Merged pull requests:

• Fix logo in docs (#409) (@​sebapersson)

Changelog

Sourced from PEtab's changelog.

Breaking updates and feature summaries across releases

PEtab.jl 5.4.0

• Added support for the new Julia native LBFGSB in Optimn for calibrate and calibrate_multistart.
• Added a show_progress keyword argument to calibrate_multistart for optionally displaying a progress bar over the multistart runs.
• Added support for the sensealg = BacksolveAdjoint for gradient_method = :Adjoint in PEtabODEProblem construction.

PEtab.jl 5.3.0

• Added support for Optimisers.jl update rules in calibrate and calibrate_multistart via a package extension. For example, it is now possible to do: calibrate(prob, x0, Optimisers.Adam(1e-3)).
• Added plotting support for fitted neural network functions in UDE problems via the plot_type = :best_function and plot_type = :function_ensemble options.
• Updated SciMLBase and ODE-solver packages compat.

Thanks to @​TorkelE for contributions to this release.

PEtab.jl 5.2.0

• Added support for ModelingToolkitNeuralNets via a package extension. This enables symbolic Universal Differential Equations (UDEs) (often called grey-box or hybrid Neural ODEs), where mechanistic and machine-learning components are combined in the ODE right-hand side. In particular, neural networks can now be embedded in ODESystem and ReactionSystem models. For examples, see the updated SciML documentation.
• Switched internal parameter handling to SciMLStructures, in particular MTKParameters. Except for dynamics provided as an ODEProblem, PEtab.jl now uses MTKParameters internally, improving long-term compatibility with the SciML ecosystem.
• Updated ODESolver verbosity to use SciMLLogging.jl. Instead of true|false, verbose should now be set to SciMLLogging.None(), SciMLLogging.Minimal(), SciMLLogging.Standard(), SciMLLogging.Detailed(), or SciMLLogging.All().
• Removed support for SciMLSensitivity forward sensitivity sensealgs ForwardSensitivities and ForwardDiffSensitivities for gradient_method = :ForwardEquations in PEtabODEProblem. These were consistently the slowest gradient options and added maintenance overhead. Use sensealg = :ForwardDiff (the current default) instead.

PEtab.jl 5.1.0

• Added support for creating a LikelihoodProfiler.jl ProfileLikelihoodProblem from a parameter estimation result via ProfileLikelihoodProblem(res, prob). This functionality is provided through an extension.

PEtab.jl 5.0.0

PEtab.jl v5.0 is a breaking release that adds support for scientific machine learning

... (truncated)

Commits

• e7823b4 Bump version from 5.4.0 to 5.4.1
• b06ae88 Merge pull request #409 from sebapersson/logo_fix
• 3925d1d Fix logo in docs
• dec7e45 Update logo image width in README
• 75178ac Update logo width in README.md
• b750eb6 Merge pull request #408 from sebapersson/logo
• 8bd0e7a Merge pull request #406 from sebapersson/dependabot/github_actions/actions/ch...
• 9679acc Merge pull request #407 from sebapersson/profile_bug
• 906650e Add logo on homepage
• 1c4444b Merge branch 'profile_bug' into logo
• Additional commits viewable in compare view