diff --git a/README.md b/README.md
index c974b3f..6c0cff2 100644
--- a/README.md
+++ b/README.md
@@ -7,3 +7,17 @@ This is a [`yt`](https://yt-project.org/) frontend for [`libyt`](https://github.
 * **yt**: https://yt-project.org/
 * **libyt Repo**: https://github.com/yt-project/libyt
 * **libyt Doc**: https://libyt.readthedocs.io/en/latest/
+
+
+### Install
+
+- From PyPI:
+  ```bash
+  pip install yt-libyt
+  ```
+- From source:
+  ```bash
+  git clone https://github.com/data-exp-lab/yt_libyt.git
+  cd yt_libyt
+  pip install .
+  ```
diff --git a/yt_libyt/_version.py b/yt_libyt/_version.py
index 6526deb..a73339b 100644
--- a/yt_libyt/_version.py
+++ b/yt_libyt/_version.py
@@ -1 +1 @@
-__version__ = "0.0.7"
+__version__ = "0.0.8"
diff --git a/yt_libyt/data_structures.py b/yt_libyt/data_structures.py
index 9afea2b..c82ebdb 100644
--- a/yt_libyt/data_structures.py
+++ b/yt_libyt/data_structures.py
@@ -53,6 +53,7 @@ def __repr__(self):
 class libytHierarchy(GridIndex):
     grid = libytGrid
     libyt = None
+
     # _preload_implemented = True # Not sure about this option
 
     def __init__(self, ds, dataset_type="libyt"):
@@ -274,6 +275,9 @@ def _set_code_unit_attributes(self):
         setdefaultattr(
             self, "magnetic_unit", self.quan(self.libyt.param_yt["magnetic_unit"], "gauss")
         )
+        setdefaultattr(
+            self, "velocity_unit", self.quan(self.libyt.param_yt["velocity_unit"], "cm/s")
+        )
 
     def _parse_parameter_file(self):
         # dataset identifier
diff --git a/yt_libyt/io.py b/yt_libyt/io.py
index d1e9f9a..cc8db8e 100644
--- a/yt_libyt/io.py
+++ b/yt_libyt/io.py
@@ -289,12 +289,25 @@ def _read_fluid_selection(self, chunks, selector, fields, size):
 
     @staticmethod
     def _get_my_rank():
-        from mpi4py import MPI
+        import libyt
 
-        comm = MPI.COMM_WORLD
-        return comm.Get_rank()
+        if libyt.libyt_info["SERIAL_MODE"] is False:
+            try:
+                from mpi4py import MPI
+
+                comm = MPI.COMM_WORLD
+                return comm.Get_rank()
+            except ImportError:
+                raise ImportError("Need mpi4py in parallel mode (SERIAL_MODE = false)")
+        else:
+            return 0
 
     def _distinguish_nonlocal_grids(self, chunks):
+
+        if self.libyt.libyt_info["SERIAL_MODE"] is True:
+            # we don't need rma
+            return False, [], [], []
+
         # Split local and non-local grids.
         from mpi4py import MPI