minor fix

Author: martinjrobins

paper/paper.md

@@ -45,7 +45,7 @@ ODE solvers require linear algebra containers (e.g. vectors, matrices), operator
 
 ODE solvers have a long history in scientific computing, and many libraries currently exist. Some notable examples include `scipy.integrate.odeint` [@virtanen2020scipy] in Python, `ode45` [@shampine1997matlab] in MATLAB, and the `Sundials` suite of solvers [@gardner2022sundials] in C. Rust is a systems programming language that is gaining popularity in the scientific computing community due to its performance, safety, and ease of use. There is currently no ODE solver library written in Rust that provides the same level of functionality as these other libraries, and this is the gap that `diffsol` aims to fill.
 
-ODE solvers written in lower-level languages like C, Fortran or Rust offer significant performance benefits. However, these solvers are often more difficult to wrap and use in higher-level languages like Python or MATLAB, primarily because users must supply their equations in the language of the solver. `diffsol` solves this issue by providing its own custom `DiffSL` DSL which is JIT compiled to efficient native code at run-time, meaning that users from a higher-level language like Python or R can specify their equations using a simple string-based format while still maintaining similar performance to pure Rust. Two other popular ODE solvers that take advantage of JIT compilation are `DifferentialEquations.jl` [@DifferentialEquations.jl-2017] in Julia, and `diffrax` [@kidger2021on] in Python. However, both these packages compile the entire solver as well as the equations, which is a significant amount of code. `diffSol` only compiles the equations, meaning that it has a significantly smaller "time-to-first-plot" for users. Another popular differential equations solver package utilising a DSL is OpenModelica [@fritzson2020OpenModelica]. Wrappers to this package in higher-level languages like Python rely on messaging to a separate OpenModelica server, which can be slow and more complicated to set up. In contrast, `diffsol` can be integrated directly into higher-level languages using language bindings and linking to a single shared or static library, see for example the `pydiffsol` Python bindings discussed below.
+ODE solvers written in lower-level languages like C, Fortran or Rust offer significant performance benefits. However, these solvers are often more difficult to wrap and use in higher-level languages like Python or MATLAB, primarily because users must supply their equations in the language of the solver. `diffsol` solves this issue by providing its own custom `DiffSL` DSL which is JIT compiled to efficient native code at run-time, meaning that users from a higher-level language like Python or R can specify their equations using a simple string-based format while still maintaining similar performance to pure Rust. Two other popular ODE solvers that take advantage of JIT compilation are `DifferentialEquations.jl` [@DifferentialEquations.jl-2017] in Julia, and `diffrax` [@kidger2021on] in Python. However, both these packages compile the entire solver as well as the equations, which is a significant amount of code. `diffSol` only compiles the equations, meaning that it has a significantly smaller "time-to-first-plot" for users. Another popular differential equations solver package utilising a DSL is OpenModelica [@fritzson2020OpenModelica]. Wrappers to this package in higher-level languages like Python rely on messaging to a separate OpenModelica server, which can be slow and more complicated to set up. In contrast, `diffsol` can be integrated directly into higher-level languages using language bindings and linking to a single shared library, see for example the `pydiffsol` Python bindings discussed below.
 
 # Features