Skip to content

Latest commit

 

History

History
105 lines (71 loc) · 2.7 KB

TODO.md

File metadata and controls

105 lines (71 loc) · 2.7 KB
Raw

To do

Re-factor to be only stochastic processes and not sub-stochastic ones.

Derive Macros

  • State
    • Needs to read where_clauses to write them again in the implementation of State trait
    • Needs to be given (input) the type of Item (...which is always T).
    • Nice error message!
  • StateIterator
    • ??

Once implemented, clean the code for all the structs.

FiniteMarkovChain

  • Construction
    • Checks of state_space!
      • What is the condition you want
      • state_space must be different elements
    • set_state_space
      • state_space must be different elements
    • From Vec
      • From a sample,
        • Consider the initial state
        • Discover the state space
        • Count each transition
        • Construct the chain
        • Add to lib module documentation
  • Move from panicking to errors
    • Create errors

ContFiniteMarkovChain

  • The same
  • has_absorbing_state(&self) -> bool

Macros

  • Learn to do useful macros!

Organization

  • Modules organization: different algorithms for simulation
    • Exact
    • Fast
      • Sample speed
    • Accurate
    • epsilon-strong
      • Brownian motion

Jump processes

In a few cases, the marginal distributions of the increments have a simple form such as a gamma distribution, a stable distribution, or an inverse Gaussian distribution so that special methods for such distributions allow to generate discrete skeletons.

  • Possion process
  • Levy process

Abstract probabilities

No longer use only f64 to represent probabilities

Requirements:

  • Need to represent cummulative probabilities, to simulate raw_dist!

Options

Best

Unit wrapper from nalgebra!

https://docs.rs/nalgebra/0.26.2/nalgebra/base/struct.Unit.html

Idea: Distribution<[T; N]> wrapper for vectors, with constant generics!

  • Implementation: prob-num crate!

Good

Struct P: and from methods for different T: Copy + Clone + Debug + PartialOrd + Num

  • Pros

    • More general than closed01
    • Checked initialization in debug mode (use assert_debug!)

  • Implement any trait we want, e.g. num_traits and some core::ops.

Rejected

  • closed01: a new type of float with extra methods.

    • Pros:
      • Add
      • WrappingAdd
      • SaturingAdd
    • Cons:
      • Still only floats

  • New trait P: to fix the types you can use.