Serialisation Functions

In this section we document methods for serialising and de-serialising DiffFusion.jl objects.

We serialise objects into ordered dictionaries of the general form

typename : TypeName
constructor : function_name
field_name_1 : field_value_1
...
field_name_n : field_value_n

Dictionary keys are strings, dictionary values are strings, numbers o ordered dictionaries of component structures.

We utilise multiple dispatch to specify serialisation recursively.

Object Serialisation

DiffFusion.serialise — Function

serialise(o::Any)

Serialise an arbitrary object.

serialise(o::Future)

Serialise a Future object from a remotecall(...).

This operation is blocking. We require that the result is calculated such that it actually can be serialised.

serialise(o::Nothing)

Serialise Nothing.

serialise(o::AbstractString)

Serialise String.

serialise(o::Integer)

Serialise Integer.

serialise(o::ModelValue)

Serialise Float.

serialise(o::AbstractDict)

Serialise dictionaries.

serialise(o::AbstractVector)

Serialise vectors.

serialise(o::AbstractMatrix)

Serialise matrices.

serialise(o::AbstractArray)

Serialise d-dimensional arrays.

serialise(o::FlatForward)

Serialise FlatForward.

serialise(o::ZeroCurve)

Serialise ZeroCurve.

serialise(o::LinearZeroCurve)

Serialise LinearZeroCurve.

serialise(o::BackwardFlatVolatility)

Serialise BackwardFlatVolatility.

serialise(o::PiecewiseFlatParameter)

Serialise PiecewiseFlatParameter.

serialise(o::CorrelationHolder)

Serialise CorrelationHolder.

serialise(o::Context)

Serialise model Context.

serialise(o::BenchmarkTimesScaling)

Serialise a BenchmarkTimesScaling enumeration object.

serialise(o::GaussianHjmModel)

Serialise GaussianHjmModel.

serialise(o::LognormalAssetModel)

Serialise LognormalAssetModel.

serialise(o::CevAssetModel)

Serialise CevAssetModel.

serialise(o::SimpleModel)

Serialise SimpleModel.

DiffFusion.serialise_struct — Function

serialise_struct(o::Any)

Create a dictionary from an arbitrary struct object

DiffFusion._serialise_key_references — Constant

Identifyiers used to specify a reference for serialisation.

DiffFusion.serialise_key — Function

serialise_key(alias::String)

Serialise an alias as a key.

This is required to capture object dependencies for de-serialisation.

DiffFusion.serialise_as_list — Function

serialise_as_list(o::SimpleModel)

Serialise SimpleModel as a list and capture references.

Object De-Serialisation

DiffFusion.deserialise — Function

deserialise(o::String, d::Union{AbstractDict, Nothing} = nothing)

De-serialise strings.

We incorporate some logic to handle external references.

We allow that the repository disctionary d contains remote call Futures. However, we want to ensure that the method returns actual objects. Thus, we fetch any Future within this method.

deserialise(o::Number, d::Union{AbstractDict, Nothing} = nothing)

De-serialise numbers.

deserialise(o::AbstractDict, d::Union{AbstractDict, Nothing} = nothing)

De-serialise dictionary.

deserialise(o::AbstractVector, d::Union{AbstractDict, Nothing} = nothing)

De-serialise vector or matrix.

DiffFusion.deserialise_object — Function

deserialise_object(o::OrderedDict, d::Union{AbstractDict, Nothing} = nothing)

De-serialise objects.

Caution, this method bares the potential risk of code injection.

DiffFusion.deserialise_from_list — Function

deserialise_from_list(dict_list::AbstractVector)

De-serialise a list of objects and capture references.

DiffFusion.array — Function

array(data::AbstractVector, dims)

Convert vector data into d-dimensional array. Here, d is length(dims).

Parameter dims is a vector or tuple of number of elements per dimension.

Data vector data must be of appropriate length, i.e. length(data) == prod(dims).

Rebuild Models

This subsection contains methods to extract term structures and re-build models.

Methods are intended to be used for sensitivity calculations. For that purpose we need to identify model parameters as inputs to the valuation function.

Model parameters for a model m::Model are stored in a Dict{String, Any} of the form

"type" => typeof(m),
"alias" => m.alias,
[parameter identifier] => [parameter value(s)]
...

DiffFusion.model_parameters — Function

model_parameters(m::GaussianHjmModel)

Extract model parameters from GaussianHjmModel.

model_parameters(m::LognormalAssetModel)

Extract model parameters from LognormalAssetModel.

model_parameters(m::CevAssetModel)

Extract model parameters from CevAssetModel.

model_parameters(m::SimpleModel)

Extract model parameters from SimpleModel.

DiffFusion.build_model — Function

build_model(
     alias::String,
     param_dict::Dict,
     model_dict::Dict,
     )

Re-build a model from model parameters dictionary.

Alias identifies the model which is to be build. Input parameter term structures are stored in paramdict. The modeldict is used to reference quanto models.

DiffFusion.termstructure_values — Function

termstructure_values(ts_dict::AbstractDict)

Extract term structure labels and values from term structure dictionary

DiffFusion.termstructure_dictionary! — Function

termstructure_dictionary!(
    ts_dict::AbstractDict,
    ts_labels::AbstractVector,
    ts_values::AbstractVector,
    )

Re-build term structure dictionary from labels and values.

DiffFusion.model_volatility_values — Function

model_volatility_values(
    alias::String,
    param_dict::Dict,
    )

Extract volatility labels and values from model parameters.

DiffFusion.model_parameters! — Function

model_parameters!(
    param_dict::Dict,
    param_labels::AbstractVector,
    param_values::AbstractVector,
    )

Re-build model parameter dictionary from volatility labels and values.